You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@tvm.apache.org by tq...@apache.org on 2023/03/17 21:19:01 UTC
[tvm-site] branch asf-site updated: deploying docs (apache/tvm@4ae08d87105afa663ac1c0392a30058912093eb4)
This is an automated email from the ASF dual-hosted git repository.
tqchen pushed a commit to branch asf-site
in repository https://gitbox.apache.org/repos/asf/tvm-site.git
The following commit(s) were added to refs/heads/asf-site by this push:
new 7d4277066f deploying docs (apache/tvm@4ae08d87105afa663ac1c0392a30058912093eb4)
7d4277066f is described below
commit 7d4277066f44e5840bcacbc01a2e653b9bb157fb
Author: tvm-bot <95...@users.noreply.github.com>
AuthorDate: Fri Mar 17 21:18:52 2023 +0000
deploying docs (apache/tvm@4ae08d87105afa663ac1c0392a30058912093eb4)
---
docs/_images/sphx_glr_micro_train_001.png | Bin 349354 -> 327396 bytes
docs/_images/sphx_glr_micro_train_thumb.png | Bin 24292 -> 23429 bytes
.../how_to/compile_models/from_darknet.rst.txt | 2 +-
.../how_to/compile_models/from_keras.rst.txt | 2 +-
.../how_to/compile_models/from_mxnet.rst.txt | 2 +-
.../how_to/compile_models/from_oneflow.rst.txt | 2 +-
.../how_to/compile_models/from_pytorch.rst.txt | 2 +-
.../how_to/compile_models/from_tensorflow.rst.txt | 2 +-
.../compile_models/sg_execution_times.rst.txt | 22 +-
.../deploy_models/deploy_model_on_adreno.rst.txt | 2 +-
.../deploy_model_on_adreno_tvmc.rst.txt | 2 +-
.../deploy_models/deploy_model_on_android.rst.txt | 2 +-
.../deploy_object_detection_pytorch.rst.txt | 4 +-
.../deploy_models/deploy_prequantized.rst.txt | 6 +-
.../deploy_prequantized_tflite.rst.txt | 4 +-
.../how_to/deploy_models/deploy_quantized.rst.txt | 2 +-
.../deploy_models/deploy_ssd_gluoncv.rst.txt | 4 +-
.../deploy_models/sg_execution_times.rst.txt | 22 +-
.../extend_tvm/bring_your_own_datatypes.rst.txt | 2 +-
.../how_to/extend_tvm/sg_execution_times.rst.txt | 10 +-
.../how_to/extend_tvm/use_pass_instrument.rst.txt | 16 +-
.../optimize_operators/opt_conv_cuda.rst.txt | 2 +-
.../optimize_operators/opt_conv_tensorcore.rst.txt | 2 +-
.../how_to/optimize_operators/opt_gemm.rst.txt | 16 +-
.../optimize_operators/sg_execution_times.rst.txt | 8 +-
.../sg_execution_times.rst.txt | 14 +-
.../tune_conv2d_layer_cuda.rst.txt | 2560 +++++++++++++++++++-
.../tune_network_cuda.rst.txt | 4 +-
.../tune_network_x86.rst.txt | 4 +-
.../tune_sparse_x86.rst.txt | 104 +-
.../tune_with_autotvm/sg_execution_times.rst.txt | 10 +-
.../tune_with_autotvm/tune_conv2d_cuda.rst.txt | 672 ++---
.../work_with_microtvm/micro_autotune.rst.txt | 18 +-
.../work_with_microtvm/micro_pytorch.rst.txt | 4 +-
.../how_to/work_with_microtvm/micro_train.rst.txt | 18 +-
.../work_with_microtvm/sg_execution_times.rst.txt | 18 +-
.../work_with_relay/sg_execution_times.rst.txt | 10 +-
.../how_to/work_with_schedules/intrin_math.rst.txt | 2 +-
.../work_with_schedules/sg_execution_times.rst.txt | 18 +-
.../tutorials/autotvm/sg_execution_times.rst.txt | 6 +-
.../frontend/deploy_classification.rst.txt | 2 +-
.../tutorials/frontend/deploy_detection.rst.txt | 2 +-
.../tutorials/frontend/sg_execution_times.rst.txt | 6 +-
.../tutorials/optimize/sg_execution_times.rst.txt | 6 +-
.../topic/vta/tutorials/sg_execution_times.rst.txt | 4 +-
.../tutorial/auto_scheduler_matmul_x86.rst.txt | 4 +-
docs/_sources/tutorial/autotvm_matmul_x86.rst.txt | 20 +-
docs/_sources/tutorial/autotvm_relay_x86.rst.txt | 58 +-
.../tutorial/cross_compilation_and_rpc.rst.txt | 2 +-
docs/_sources/tutorial/intro_topi.rst.txt | 2 +-
docs/_sources/tutorial/sg_execution_times.rst.txt | 22 +-
.../tutorial/tensor_expr_get_started.rst.txt | 49 +-
docs/commit_hash | 2 +-
docs/how_to/compile_models/from_darknet.html | 2 +-
docs/how_to/compile_models/from_keras.html | 2 +-
docs/how_to/compile_models/from_mxnet.html | 2 +-
docs/how_to/compile_models/from_oneflow.html | 11 +-
docs/how_to/compile_models/from_pytorch.html | 9 +-
docs/how_to/compile_models/from_tensorflow.html | 2 +-
docs/how_to/compile_models/sg_execution_times.html | 22 +-
.../deploy_models/deploy_model_on_adreno.html | 2 +-
.../deploy_models/deploy_model_on_adreno_tvmc.html | 37 +-
.../deploy_models/deploy_model_on_android.html | 2 +-
.../deploy_object_detection_pytorch.html | 41 +-
docs/how_to/deploy_models/deploy_prequantized.html | 8 +-
.../deploy_models/deploy_prequantized_tflite.html | 4 +-
docs/how_to/deploy_models/deploy_quantized.html | 2 +-
docs/how_to/deploy_models/deploy_ssd_gluoncv.html | 37 +-
docs/how_to/deploy_models/sg_execution_times.html | 26 +-
.../extend_tvm/bring_your_own_datatypes.html | 2 +-
docs/how_to/extend_tvm/sg_execution_times.html | 10 +-
docs/how_to/extend_tvm/use_pass_instrument.html | 16 +-
docs/how_to/optimize_operators/opt_conv_cuda.html | 2 +-
.../optimize_operators/opt_conv_tensorcore.html | 2 +-
docs/how_to/optimize_operators/opt_gemm.html | 16 +-
.../optimize_operators/sg_execution_times.html | 8 +-
.../sg_execution_times.html | 14 +-
.../tune_conv2d_layer_cuda.html | 2560 +++++++++++++++++++-
.../tune_with_autoscheduler/tune_network_cuda.html | 4 +-
.../tune_with_autoscheduler/tune_network_x86.html | 4 +-
.../tune_with_autoscheduler/tune_sparse_x86.html | 104 +-
.../tune_with_autotvm/sg_execution_times.html | 10 +-
.../how_to/tune_with_autotvm/tune_conv2d_cuda.html | 672 ++---
docs/how_to/work_with_microtvm/micro_autotune.html | 18 +-
docs/how_to/work_with_microtvm/micro_pytorch.html | 5 +-
docs/how_to/work_with_microtvm/micro_train.html | 16 +-
.../work_with_microtvm/sg_execution_times.html | 18 +-
.../how_to/work_with_relay/sg_execution_times.html | 10 +-
docs/how_to/work_with_schedules/intrin_math.html | 2 +-
.../work_with_schedules/sg_execution_times.html | 18 +-
docs/install/nnpack.html | 12 +-
docs/reference/api/python/auto_scheduler.html | 4 +-
.../api/typedoc/classes/bytestreamreader.html | 12 +-
.../api/typedoc/classes/cachedcallstack.html | 34 +-
docs/reference/api/typedoc/classes/dldatatype.html | 12 +-
docs/reference/api/typedoc/classes/dldevice.html | 10 +-
.../reference/api/typedoc/classes/environment.html | 12 +-
docs/reference/api/typedoc/classes/ffilibrary.html | 20 +-
docs/reference/api/typedoc/classes/instance.html | 58 +-
docs/reference/api/typedoc/classes/memory.html | 34 +-
docs/reference/api/typedoc/classes/module.html | 10 +-
docs/reference/api/typedoc/classes/ndarray.html | 22 +-
.../api/typedoc/classes/packedfunccell.html | 6 +-
docs/reference/api/typedoc/classes/rpcserver.html | 14 +-
.../api/typedoc/classes/runtimecontext.html | 22 +-
docs/reference/api/typedoc/classes/scalar.html | 6 +-
docs/reference/api/typedoc/classes/tvmarray.html | 16 +-
docs/reference/api/typedoc/classes/tvmobject.html | 12 +-
.../api/typedoc/classes/webgpucontext.html | 12 +-
docs/reference/api/typedoc/enums/argtypecode.html | 30 +-
.../api/typedoc/enums/aynccallbackcode.html | 4 +-
.../api/typedoc/enums/dldatatypecode.html | 8 +-
.../api/typedoc/enums/rpcserverstate.html | 12 +-
docs/reference/api/typedoc/enums/sizeof.html | 18 +-
docs/reference/api/typedoc/index.html | 124 +-
.../api/typedoc/interfaces/disposable.html | 2 +-
.../api/typedoc/interfaces/functioninfo.html | 6 +-
.../api/typedoc/interfaces/libraryprovider.html | 4 +-
docs/searchindex.js | 2 +-
.../vta/tutorials/autotvm/sg_execution_times.html | 6 +-
.../tutorials/frontend/deploy_classification.html | 2 +-
.../vta/tutorials/frontend/deploy_detection.html | 2 +-
.../vta/tutorials/frontend/sg_execution_times.html | 6 +-
.../vta/tutorials/optimize/sg_execution_times.html | 6 +-
docs/topic/vta/tutorials/sg_execution_times.html | 4 +-
docs/tutorial/auto_scheduler_matmul_x86.html | 4 +-
docs/tutorial/autotvm_matmul_x86.html | 20 +-
docs/tutorial/autotvm_relay_x86.html | 272 +--
docs/tutorial/cross_compilation_and_rpc.html | 2 +-
docs/tutorial/intro_topi.html | 2 +-
docs/tutorial/sg_execution_times.html | 22 +-
docs/tutorial/tensor_expr_get_started.html | 45 +-
132 files changed, 6292 insertions(+), 2138 deletions(-)
diff --git a/docs/_images/sphx_glr_micro_train_001.png b/docs/_images/sphx_glr_micro_train_001.png
index b147d290fc..6a9ff83b6f 100644
Binary files a/docs/_images/sphx_glr_micro_train_001.png and b/docs/_images/sphx_glr_micro_train_001.png differ
diff --git a/docs/_images/sphx_glr_micro_train_thumb.png b/docs/_images/sphx_glr_micro_train_thumb.png
index a474f132c9..faf94be3ac 100644
Binary files a/docs/_images/sphx_glr_micro_train_thumb.png and b/docs/_images/sphx_glr_micro_train_thumb.png differ
diff --git a/docs/_sources/how_to/compile_models/from_darknet.rst.txt b/docs/_sources/how_to/compile_models/from_darknet.rst.txt
index 5b0568b654..0e3617a9e7 100644
--- a/docs/_sources/how_to/compile_models/from_darknet.rst.txt
+++ b/docs/_sources/how_to/compile_models/from_darknet.rst.txt
@@ -318,7 +318,7 @@ The process is no different from other examples.
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 25.999 seconds)
+ **Total running time of the script:** ( 1 minutes 17.208 seconds)
.. _sphx_glr_download_how_to_compile_models_from_darknet.py:
diff --git a/docs/_sources/how_to/compile_models/from_keras.rst.txt b/docs/_sources/how_to/compile_models/from_keras.rst.txt
index 91bdc62360..3abdf5da34 100644
--- a/docs/_sources/how_to/compile_models/from_keras.rst.txt
+++ b/docs/_sources/how_to/compile_models/from_keras.rst.txt
@@ -232,7 +232,7 @@ Look up prediction top 1 index in 1000 class synset.
.. code-block:: none
Relay top-1 id: 285, class name: Egyptian cat
-
1/1 [==============================] - ETA: 0s
1/1 [==============================] - 1s 1s/step
+
1/1 [==============================] - ETA: 0s
1/1 [==============================] - 1s 945ms/step
Keras top-1 id: 285, class name: Egyptian cat
diff --git a/docs/_sources/how_to/compile_models/from_mxnet.rst.txt b/docs/_sources/how_to/compile_models/from_mxnet.rst.txt
index 093a70b9e1..eba9703613 100644
--- a/docs/_sources/how_to/compile_models/from_mxnet.rst.txt
+++ b/docs/_sources/how_to/compile_models/from_mxnet.rst.txt
@@ -116,7 +116,7 @@ In this section, we download a pretrained imagenet model and classify an image.
.. code-block:: none
- Downloading /workspace/.mxnet/models/resnet18_v1-a0666292.zipb9f2e74b-857a-4d43-8ef5-b9cf3ae31ea9 from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/resnet18_v1-a0666292.zip...
+ Downloading /workspace/.mxnet/models/resnet18_v1-a0666292.zipa4a1b008-0ef6-41a2-8421-d5c49a2c614a from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/resnet18_v1-a0666292.zip...
x (1, 3, 224, 224)
diff --git a/docs/_sources/how_to/compile_models/from_oneflow.rst.txt b/docs/_sources/how_to/compile_models/from_oneflow.rst.txt
index 14d9b1d0e6..8bed8d2a71 100644
--- a/docs/_sources/how_to/compile_models/from_oneflow.rst.txt
+++ b/docs/_sources/how_to/compile_models/from_oneflow.rst.txt
@@ -121,7 +121,7 @@ Load a pretrained OneFlow model and save model
.. code-block:: none
Downloading: "https://oneflow-public.oss-cn-beijing.aliyuncs.com/model_zoo/flowvision/classification/ResNet/resnet18.zip" to /workspace/.oneflow/flowvision_cache/resnet18.zip
-
0%| | 0.00/41.5M [00:00<?, ?B/s]
19%|#9 | 7.99M/41.5M [00:00<00:00, 57.2MB/s]
35%|###4 | 14.3M/41.5M [00:00<00:00, 60.5MB/s]
54%|#####3 | 22.3M/41.5M [00:00<00:00, 54.8MB/s]
67%|######6 | 27.7M/41.5M [00:00<00:00, 53.9MB/s]
92%|#########2| 38.3M/41.5M [00:00<00:00, 63.9MB/s]
100%|##########| 41.5M/41.5M [00:00<00:00, 58.2MB/s]
+
0%| | 0.00/41.5M [00:00<?, ?B/s]
19%|#9 | 7.99M/41.5M [00:00<00:00, 69.3MB/s]
39%|###8 | 16.0M/41.5M [00:00<00:00, 58.5MB/s]
58%|#####7 | 24.0M/41.5M [00:00<00:00, 61.8MB/s]
87%|########7 | 36.2M/41.5M [00:00<00:00, 81.0MB/s]
100%|##########| 41.5M/41.5M [00:00<00:00, 74.6MB/s]
diff --git a/docs/_sources/how_to/compile_models/from_pytorch.rst.txt b/docs/_sources/how_to/compile_models/from_pytorch.rst.txt
index 3a5ddc7762..117d791fc2 100644
--- a/docs/_sources/how_to/compile_models/from_pytorch.rst.txt
+++ b/docs/_sources/how_to/compile_models/from_pytorch.rst.txt
@@ -101,7 +101,7 @@ Load a pretrained PyTorch model
/venv/apache-tvm-py3.7/lib/python3.7/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and will be removed in 0.15. The current behavior is equivalent to passing `weights=ResNet18_Weights.IMAGENET1K_V1`. You can also use `weights=ResNet18_Weights.DEFAULT` to get the most up-to-date weights.
warnings.warn(msg)
Downloading: "https://download.pytorch.org/models/resnet18-f37072fd.pth" to /workspace/.cache/torch/hub/checkpoints/resnet18-f37072fd.pth
-
0%| | 0.00/44.7M [00:00<?, ?B/s]
18%|#8 | 8.12M/44.7M [00:00<00:00, 81.0MB/s]
40%|#### | 18.1M/44.7M [00:00<00:00, 93.8MB/s]
61%|###### | 27.0M/44.7M [00:00<00:00, 67.9MB/s]
90%|########9 | 40.0M/44.7M [00:00<00:00, 84.1MB/s]
100%|##########| 44.7M/44.7M [00:00<00:00, 76.1MB/s]
+
0%| | 0.00/44.7M [00:00<?, ?B/s]
30%|##9 | 13.4M/44.7M [00:00<00:00, 140MB/s]
60%|#####9 | 26.7M/44.7M [00:00<00:00, 111MB/s]
84%|########4 | 37.6M/44.7M [00:00<00:00, 108MB/s]
100%|##########| 44.7M/44.7M [00:00<00:00, 105MB/s]
diff --git a/docs/_sources/how_to/compile_models/from_tensorflow.rst.txt b/docs/_sources/how_to/compile_models/from_tensorflow.rst.txt
index b03eb44c78..e1b4e7509a 100644
--- a/docs/_sources/how_to/compile_models/from_tensorflow.rst.txt
+++ b/docs/_sources/how_to/compile_models/from_tensorflow.rst.txt
@@ -424,7 +424,7 @@ Run the corresponding model on tensorflow
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 30.735 seconds)
+ **Total running time of the script:** ( 1 minutes 22.350 seconds)
.. _sphx_glr_download_how_to_compile_models_from_tensorflow.py:
diff --git a/docs/_sources/how_to/compile_models/sg_execution_times.rst.txt b/docs/_sources/how_to/compile_models/sg_execution_times.rst.txt
index 47fa8aa920..02cece0cfa 100644
--- a/docs/_sources/how_to/compile_models/sg_execution_times.rst.txt
+++ b/docs/_sources/how_to/compile_models/sg_execution_times.rst.txt
@@ -5,26 +5,26 @@
Computation times
=================
-**06:58.560** total execution time for **how_to_compile_models** files:
+**06:32.674** total execution time for **how_to_compile_models** files:
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_tensorflow.py` (``from_tensorflow.py``) | 01:30.735 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_tensorflow.py` (``from_tensorflow.py``) | 01:22.350 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_darknet.py` (``from_darknet.py``) | 01:25.999 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_darknet.py` (``from_darknet.py``) | 01:17.208 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_paddle.py` (``from_paddle.py``) | 00:55.487 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_paddle.py` (``from_paddle.py``) | 00:55.122 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_oneflow.py` (``from_oneflow.py``) | 00:37.646 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_oneflow.py` (``from_oneflow.py``) | 00:37.119 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_coreml.py` (``from_coreml.py``) | 00:35.740 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_coreml.py` (``from_coreml.py``) | 00:31.687 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_mxnet.py` (``from_mxnet.py``) | 00:33.060 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_mxnet.py` (``from_mxnet.py``) | 00:31.006 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_tflite.py` (``from_tflite.py``) | 00:29.017 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_tflite.py` (``from_tflite.py``) | 00:27.332 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_pytorch.py` (``from_pytorch.py``) | 00:26.231 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_pytorch.py` (``from_pytorch.py``) | 00:25.712 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_keras.py` (``from_keras.py``) | 00:21.903 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_keras.py` (``from_keras.py``) | 00:22.387 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_onnx.py` (``from_onnx.py``) | 00:02.741 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_onnx.py` (``from_onnx.py``) | 00:02.750 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/how_to/deploy_models/deploy_model_on_adreno.rst.txt b/docs/_sources/how_to/deploy_models/deploy_model_on_adreno.rst.txt
index b88eea6b85..0078ab68d9 100644
--- a/docs/_sources/how_to/deploy_models/deploy_model_on_adreno.rst.txt
+++ b/docs/_sources/how_to/deploy_models/deploy_model_on_adreno.rst.txt
@@ -637,7 +637,7 @@ well as provides information about the model's performance
Evaluate inference time cost...
Execution time summary:
mean (ms) median (ms) max (ms) min (ms) std (ms)
- 2539.6140 2539.3624 2542.1952 2537.9811 1.3877
+ 2540.1147 2539.6340 2544.0332 2538.5511 1.6081
diff --git a/docs/_sources/how_to/deploy_models/deploy_model_on_adreno_tvmc.rst.txt b/docs/_sources/how_to/deploy_models/deploy_model_on_adreno_tvmc.rst.txt
index 32e17fa71a..763240eed5 100644
--- a/docs/_sources/how_to/deploy_models/deploy_model_on_adreno_tvmc.rst.txt
+++ b/docs/_sources/how_to/deploy_models/deploy_model_on_adreno_tvmc.rst.txt
@@ -127,7 +127,7 @@ Make a Keras Resnet50 Model
.. code-block:: none
Downloading data from https://storage.googleapis.com/tensorflow/keras-applications/resnet/resnet50_weights_tf_dim_ordering_tf_kernels.h5
-
8192/102967424 [..............................] - ETA: 0s
7725056/102967424 [=>............................] - ETA: 0s
12697600/102967424 [==>...........................] - ETA: 0s
16908288/102967424 [===>..........................] - ETA: 0s
22945792/102967424 [=====>........................] - ETA: 0s
27918336/102967424 [=======>......................] - ETA: 0s
32964608/102967424 [========>.....................] - ETA: 0s
37969920/102967424 [==========>...................] - ETA: 0s
42328064/102967424 [===========>..................] - ETA: 0s
48013312/102967424 [============>.................] - ETA: 0s
53018624/102967424 [==============>...............] - ETA: 0s
58023936/102967424 [===============>..............] - ETA: 0s
63004672/102967424 [=================>............] - ETA: 0s
68001792/102967424 [==================>...........] - ETA: 0s
70123520/102967424 [===================>..........] - ETA: 0s
78610432/102967424 [=====================>........] -
ETA: 0s
80936960/102967424 [======================>.......] - ETA: 0s
91070464/102967424 [=========================>....] - ETA: 0s
99901440/102967424 [============================>.] - ETA: 0s
102967424/102967424 [==============================] - 1s 0us/step
+
8192/102967424 [..............................] - ETA: 0s
131072/102967424 [..............................] - ETA: 39s
7012352/102967424 [=>............................] - ETA: 1s
14532608/102967424 [===>..........................] - ETA: 0s
16769024/102967424 [===>..........................] - ETA: 1s
25157632/102967424 [======>.......................] - ETA: 1s
33546240/102967424 [========>.....................] - ETA: 0s
39198720/102967424 [==========>...................] - ETA: 0s
44130304/102967424 [===========>..................] - ETA: 0s
49438720/102967424 [=============>................] - ETA: 0s
51462144/102967424 [=============>................] - ETA: 0s
58712064/102967424 [================>.............] - ETA: 0s
67305472/102967424 [==================>...........] - ETA: 0s
69296128/102967424 [===================>..........] - ETA: 0s
72540160/102967424 [====================>.........] - ETA: 0s
75489280/102967424 [====================>.........
] - ETA: 0s
83877888/102967424 [=======================>......] - ETA: 0s
88711168/102967424 [========================>.....] - ETA: 0s
92266496/102967424 [=========================>....] - ETA: 0s
100794368/102967424 [============================>.] - ETA: 0s
102967424/102967424 [==============================] - 1s 0us/step
diff --git a/docs/_sources/how_to/deploy_models/deploy_model_on_android.rst.txt b/docs/_sources/how_to/deploy_models/deploy_model_on_android.rst.txt
index 0a23d2a495..f5bd2f8aa7 100644
--- a/docs/_sources/how_to/deploy_models/deploy_model_on_android.rst.txt
+++ b/docs/_sources/how_to/deploy_models/deploy_model_on_android.rst.txt
@@ -437,7 +437,7 @@ Execute on TVM
Evaluate inference time cost...
Execution time summary:
mean (ms) median (ms) max (ms) min (ms) std (ms)
- 16.3628 16.2917 17.5737 15.7696 0.5427
+ 15.4342 15.4281 15.5401 15.3692 0.0423
diff --git a/docs/_sources/how_to/deploy_models/deploy_object_detection_pytorch.rst.txt b/docs/_sources/how_to/deploy_models/deploy_object_detection_pytorch.rst.txt
index 598194c964..b8887a04d4 100644
--- a/docs/_sources/how_to/deploy_models/deploy_object_detection_pytorch.rst.txt
+++ b/docs/_sources/how_to/deploy_models/deploy_object_detection_pytorch.rst.txt
@@ -130,7 +130,7 @@ Load pre-trained maskrcnn from torchvision and do tracing
/venv/apache-tvm-py3.7/lib/python3.7/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and will be removed in 0.15. The current behavior is equivalent to passing `weights=MaskRCNN_ResNet50_FPN_Weights.COCO_V1`. You can also use `weights=MaskRCNN_ResNet50_FPN_Weights.DEFAULT` to get the most up-to-date weights.
warnings.warn(msg)
Downloading: "https://download.pytorch.org/models/maskrcnn_resnet50_fpn_coco-bf2d0c1e.pth" to /workspace/.cache/torch/hub/checkpoints/maskrcnn_resnet50_fpn_coco-bf2d0c1e.pth
-
0%| | 0.00/170M [00:00<?, ?B/s]
4%|3 | 6.30M/170M [00:00<00:06, 26.9MB/s]
5%|5 | 8.88M/170M [00:00<00:06, 26.8MB/s]
9%|9 | 16.0M/170M [00:00<00:04, 38.7MB/s]
14%|#4 | 24.0M/170M [00:00<00:02, 51.2MB/s]
19%|#9 | 32.6M/170M [00:00<00:02, 63.0MB/s]
26%|##5 | 43.4M/170M [00:00<00:01, 77.8MB/s]
30%|### | 51.3M/170M [00:00<00:01, 74.0MB/s]
35%|###4 | 58.7M/170M [00:01<00:02, 52.3MB/s]
38%|###8 | 64.7M/170M [00:01<00:02, 46.7MB/s]
42%|####2 | 72.0M/170M [00:01<00:02, 47.4MB/s]
47%|####7 | 80.0M/170M [00:01<00:01, 53.4MB/s]
52%|#####1 | 88.0M/170M [00:01<00:01, 59.5MB/s]
57%|#####6 | 96.0M/170M [00:01<00:01, 56.8MB/s]
61%|######1 | 104M/170M [00:02<00:01, 53.3MB/s]
66%|######5 | 112M/170M [00:02<00:01, 56.2MB/s]
71%|####### | 120M/170M [00:02<00:00, 55.4MB/s]
75%|#######5 | 128M/170M [00:02<00:00, 57.5MB/s]
80%|######## | 136M/170M [00:02<00:00, 52.7MB/s]
85%|########4 | 144M/170M [00:02<00:00, 54.6MB/s]
89%|########9 | 152M/170M [00:02<00:00, 56.6MB/s]
93%|#########3| 158M/170M [00:03<00:00, 54.4MB/s]
96%|#########6| 164M/170M [00:03<00:00, 48.9MB/s]
99%|#########9| 168M/170M [00:03<00:00, 44.3MB/s]
100%|##########| 170M/170M [00:03<00:00, 52.5MB/s]
+
0%| | 0.00/170M [00:00<?, ?B/s]
8%|7 | 12.8M/170M [00:00<00:01, 134MB/s]
15%|#5 | 25.5M/170M [00:00<00:01, 111MB/s]
22%|##1 | 36.6M/170M [00:00<00:01, 113MB/s]
28%|##8 | 48.0M/170M [00:00<00:01, 99.2MB/s]
34%|###3 | 57.7M/170M [00:00<00:01, 99.7MB/s]
42%|####2 | 71.7M/170M [00:00<00:00, 107MB/s]
49%|####8 | 82.6M/170M [00:00<00:01, 91.4MB/s]
54%|#####3 | 91.7M/170M [00:00<00:00, 89.9MB/s]
59%|#####9 | 100M/170M [00:01<00:00, 85.4MB/s]
66%|######5 | 112M/170M [00:01<00:00, 90.9MB/s]
75%|#######4 | 127M/170M [00:01<00:00, 109MB/s]
81%|########1 | 138M/170M [00:01<00:00, 82.0MB/s]
89%|########9 | 152M/170M [00:01<00:00, 94.7MB/s]
95%|#########5| 162M/170M [00:01<00:00, 85.6MB/s]
100%|##########| 170M/170M [00:01<00:00, 94.8MB/s]
/venv/apache-tvm-py3.7/lib/python3.7/site-packages/torch/nn/functional.py:3897: UserWarning: To copy construct from a tensor, it is recommended to use sourceTensor.clone().detach() or sourceTensor.clone().detach().requires_grad_(True), rather than torch.tensor(sourceTensor).
for i in range(dim)
/venv/apache-tvm-py3.7/lib/python3.7/site-packages/torchvision/models/detection/anchor_utils.py:124: UserWarning: __floordiv__ is deprecated, and its behavior will change in a future version of pytorch. It currently rounds toward 0 (like the 'trunc' function NOT 'floor'). This results in incorrect rounding for negative values. To keep the current behavior, use torch.div(a, b, rounding_mode='trunc'), or for actual floor division, use torch.div(a, b, rounding_mode='floor').
@@ -299,7 +299,7 @@ Get boxes with score larger than 0.9
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 3 minutes 48.306 seconds)
+ **Total running time of the script:** ( 3 minutes 28.701 seconds)
.. _sphx_glr_download_how_to_deploy_models_deploy_object_detection_pytorch.py:
diff --git a/docs/_sources/how_to/deploy_models/deploy_prequantized.rst.txt b/docs/_sources/how_to/deploy_models/deploy_prequantized.rst.txt
index 8d3ab272cd..e0addded48 100644
--- a/docs/_sources/how_to/deploy_models/deploy_prequantized.rst.txt
+++ b/docs/_sources/how_to/deploy_models/deploy_prequantized.rst.txt
@@ -227,7 +227,7 @@ training. Other models require a full post training calibration.
/venv/apache-tvm-py3.7/lib/python3.7/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and will be removed in 0.15. The current behavior is equivalent to passing `weights=MobileNet_V2_Weights.IMAGENET1K_V1`. You can also use `weights=MobileNet_V2_Weights.DEFAULT` to get the most up-to-date weights.
warnings.warn(msg)
Downloading: "https://download.pytorch.org/models/mobilenet_v2-b0353104.pth" to /workspace/.cache/torch/hub/checkpoints/mobilenet_v2-b0353104.pth
-
0%| | 0.00/13.6M [00:00<?, ?B/s]
59%|#####8 | 7.99M/13.6M [00:00<00:00, 37.8MB/s]
100%|##########| 13.6M/13.6M [00:00<00:00, 49.3MB/s]
+
0%| | 0.00/13.6M [00:00<?, ?B/s]
59%|#####8 | 7.99M/13.6M [00:00<00:00, 78.7MB/s]
100%|##########| 13.6M/13.6M [00:00<00:00, 85.4MB/s]
@@ -409,7 +409,7 @@ Here we give an example of how to measure performance of TVM compiled models.
Execution time summary:
mean (ms) median (ms) max (ms) min (ms) std (ms)
- 90.0792 90.0454 90.8264 89.8506 0.1604
+ 90.3108 90.1330 102.9750 89.7820 1.3601
@@ -458,7 +458,7 @@ TODO
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 23.441 seconds)
+ **Total running time of the script:** ( 1 minutes 15.641 seconds)
.. _sphx_glr_download_how_to_deploy_models_deploy_prequantized.py:
diff --git a/docs/_sources/how_to/deploy_models/deploy_prequantized_tflite.rst.txt b/docs/_sources/how_to/deploy_models/deploy_prequantized_tflite.rst.txt
index 92598d7a88..64ed3c4dc8 100644
--- a/docs/_sources/how_to/deploy_models/deploy_prequantized_tflite.rst.txt
+++ b/docs/_sources/how_to/deploy_models/deploy_prequantized_tflite.rst.txt
@@ -423,7 +423,7 @@ Here we give an example of how to measure performance of TVM compiled models.
Execution time summary:
mean (ms) median (ms) max (ms) min (ms) std (ms)
- 117.3747 117.1252 123.4665 115.3360 1.2965
+ 118.6768 118.4045 121.8308 116.8414 1.0319
@@ -460,7 +460,7 @@ Here we give an example of how to measure performance of TVM compiled models.
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 2 minutes 35.759 seconds)
+ **Total running time of the script:** ( 2 minutes 29.659 seconds)
.. _sphx_glr_download_how_to_deploy_models_deploy_prequantized_tflite.py:
diff --git a/docs/_sources/how_to/deploy_models/deploy_quantized.rst.txt b/docs/_sources/how_to/deploy_models/deploy_quantized.rst.txt
index 1971f9efe7..7706bebf21 100644
--- a/docs/_sources/how_to/deploy_models/deploy_quantized.rst.txt
+++ b/docs/_sources/how_to/deploy_models/deploy_quantized.rst.txt
@@ -257,7 +257,7 @@ We create a Relay VM to build and execute the model.
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 43.626 seconds)
+ **Total running time of the script:** ( 1 minutes 34.221 seconds)
.. _sphx_glr_download_how_to_deploy_models_deploy_quantized.py:
diff --git a/docs/_sources/how_to/deploy_models/deploy_ssd_gluoncv.rst.txt b/docs/_sources/how_to/deploy_models/deploy_ssd_gluoncv.rst.txt
index 6f15846fc7..a91f3ba6f4 100644
--- a/docs/_sources/how_to/deploy_models/deploy_ssd_gluoncv.rst.txt
+++ b/docs/_sources/how_to/deploy_models/deploy_ssd_gluoncv.rst.txt
@@ -170,7 +170,7 @@ Convert and compile model for CPU.
data: None
input_sym_arg_type = in_param.infer_type()[0]
Downloading /workspace/.mxnet/models/ssd_512_resnet50_v1_voc-9c8b225a.zip from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/ssd_512_resnet50_v1_voc-9c8b225a.zip...
-
0%| | 0/132723 [00:00<?, ?KB/s]
4%|3 | 5069/132723 [00:00<00:02, 50681.95KB/s]
10%|9 | 13250/132723 [00:00<00:01, 68989.48KB/s]
15%|#5 | 20150/132723 [00:00<00:01, 63403.97KB/s]
21%|##1 | 28398/132723 [00:00<00:01, 70549.80KB/s]
27%|##6 | 35521/132723 [00:00<00:02, 45403.86KB/s]
33%|###2 | 43717/132723 [00:00<00:01, 54254.78KB/s]
39%|###9 | 51912/132723 [00:00<00:01, 61372.09KB/s]
45%|####5 | 60122/132723 [00:00<00:01, 66936.75KB/s]
52%|#####1 | 68397/132723 [00:01<00:00, 71320.27KB/s]
58%|#####7 | 76670/132723 [00:01<00:00, 74555.92KB/s]
64%|######4 | 84979/132723 [00:01<00:00, 77015.80KB/s]
70%|####### | 93264/132723 [00:01<00:00, 78715.69KB/s]
76%|#######6 | 101347/132723 [00:01<00:00, 75423.52KB/s]
83%|########2 | 109619/132723 [00:01<00:00, 77504.71KB/s]
89%|########8 | 117503/132723 [00:01<00:00, 65667.40KB/s]
95%|########
#4| 125700/132723 [00:01<00:00, 69882.50KB/s]
100%|##########| 132723/132723 [00:01<00:00, 68132.15KB/s]
+
0%| | 0/132723 [00:00<?, ?KB/s]
5%|4 | 6208/132723 [00:00<00:02, 62072.22KB/s]
10%|# | 13908/132723 [00:00<00:01, 70849.27KB/s]
16%|#6 | 21588/132723 [00:00<00:01, 73564.38KB/s]
22%|##2 | 29313/132723 [00:00<00:01, 75014.22KB/s]
28%|##7 | 36979/132723 [00:00<00:01, 75605.60KB/s]
34%|###3 | 44633/132723 [00:00<00:01, 75922.07KB/s]
39%|###9 | 52295/132723 [00:00<00:01, 76148.00KB/s]
45%|####5 | 60042/132723 [00:00<00:00, 76566.82KB/s]
51%|#####1 | 67705/132723 [00:00<00:00, 76560.29KB/s]
57%|#####6 | 75362/132723 [00:01<00:00, 76424.57KB/s]
63%|######2 | 83005/132723 [00:01<00:00, 68789.41KB/s]
68%|######8 | 90464/132723 [00:01<00:00, 70424.65KB/s]
74%|#######3 | 98172/132723 [00:01<00:00, 72333.93KB/s]
80%|#######9 | 105588/132723 [00:01<00:00, 72864.33KB/s]
85%|########5 | 113207/132723 [00:01<00:00, 73838.60KB/s]
91%|#########
| 120638/132723 [00:01<00:00, 65562.52KB/s]
97%|#########6| 128183/132723 [00:01<00:00, 68252.09KB/s]
100%|##########| 132723/132723 [00:01<00:00, 72121.45KB/s]
@@ -246,7 +246,7 @@ Display result
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 4 minutes 8.321 seconds)
+ **Total running time of the script:** ( 3 minutes 43.650 seconds)
.. _sphx_glr_download_how_to_deploy_models_deploy_ssd_gluoncv.py:
diff --git a/docs/_sources/how_to/deploy_models/sg_execution_times.rst.txt b/docs/_sources/how_to/deploy_models/sg_execution_times.rst.txt
index 19ac7cf49d..92ea5be1ea 100644
--- a/docs/_sources/how_to/deploy_models/sg_execution_times.rst.txt
+++ b/docs/_sources/how_to/deploy_models/sg_execution_times.rst.txt
@@ -5,28 +5,28 @@
Computation times
=================
-**17:09.664** total execution time for **how_to_deploy_models** files:
+**15:54.268** total execution time for **how_to_deploy_models** files:
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_ssd_gluoncv.py` (``deploy_ssd_gluoncv.py``) | 04:08.321 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_ssd_gluoncv.py` (``deploy_ssd_gluoncv.py``) | 03:43.650 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_object_detection_pytorch.py` (``deploy_object_detection_pytorch.py``) | 03:48.306 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_object_detection_pytorch.py` (``deploy_object_detection_pytorch.py``) | 03:28.701 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_prequantized_tflite.py` (``deploy_prequantized_tflite.py``) | 02:35.759 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_prequantized_tflite.py` (``deploy_prequantized_tflite.py``) | 02:29.659 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_quantized.py` (``deploy_quantized.py``) | 01:43.626 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_quantized.py` (``deploy_quantized.py``) | 01:34.221 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_prequantized.py` (``deploy_prequantized.py``) | 01:23.441 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_prequantized.py` (``deploy_prequantized.py``) | 01:15.641 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_adreno.py` (``deploy_model_on_adreno.py``) | 00:54.517 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_adreno.py` (``deploy_model_on_adreno.py``) | 00:54.542 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_adreno_tvmc.py` (``deploy_model_on_adreno_tvmc.py``) | 00:50.879 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_adreno_tvmc.py` (``deploy_model_on_adreno_tvmc.py``) | 00:50.612 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_android.py` (``deploy_model_on_android.py``) | 00:47.062 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_android.py` (``deploy_model_on_android.py``) | 00:41.939 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_rasp.py` (``deploy_model_on_rasp.py``) | 00:29.830 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_nano.py` (``deploy_model_on_nano.py``) | 00:27.808 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_nano.py` (``deploy_model_on_nano.py``) | 00:27.917 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_rasp.py` (``deploy_model_on_rasp.py``) | 00:27.491 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
| :ref:`sphx_glr_how_to_deploy_models_deploy_sparse.py` (``deploy_sparse.py``) | 00:00.006 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/how_to/extend_tvm/bring_your_own_datatypes.rst.txt b/docs/_sources/how_to/extend_tvm/bring_your_own_datatypes.rst.txt
index 0f9307dea6..607799c347 100644
--- a/docs/_sources/how_to/extend_tvm/bring_your_own_datatypes.rst.txt
+++ b/docs/_sources/how_to/extend_tvm/bring_your_own_datatypes.rst.txt
@@ -463,7 +463,7 @@ First let us define two helper functions to get the mobilenet model and a cat im
.. code-block:: none
- Downloading /workspace/.mxnet/models/mobilenet0.25-9f83e440.zipf00e4e20-651a-4595-b06a-fae9a812d3ab from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/mobilenet0.25-9f83e440.zip...
+ Downloading /workspace/.mxnet/models/mobilenet0.25-9f83e440.zip156d0c7d-0023-4dbc-9bc2-6c10836a16c5 from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/mobilenet0.25-9f83e440.zip...
diff --git a/docs/_sources/how_to/extend_tvm/sg_execution_times.rst.txt b/docs/_sources/how_to/extend_tvm/sg_execution_times.rst.txt
index 4faa54ddf6..703d67279a 100644
--- a/docs/_sources/how_to/extend_tvm/sg_execution_times.rst.txt
+++ b/docs/_sources/how_to/extend_tvm/sg_execution_times.rst.txt
@@ -5,14 +5,14 @@
Computation times
=================
-**00:54.808** total execution time for **how_to_extend_tvm** files:
+**00:53.388** total execution time for **how_to_extend_tvm** files:
+-------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_extend_tvm_bring_your_own_datatypes.py` (``bring_your_own_datatypes.py``) | 00:50.656 | 0.0 MB |
+| :ref:`sphx_glr_how_to_extend_tvm_bring_your_own_datatypes.py` (``bring_your_own_datatypes.py``) | 00:49.641 | 0.0 MB |
+-------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_extend_tvm_use_pass_instrument.py` (``use_pass_instrument.py``) | 00:03.027 | 0.0 MB |
+| :ref:`sphx_glr_how_to_extend_tvm_use_pass_instrument.py` (``use_pass_instrument.py``) | 00:02.694 | 0.0 MB |
+-------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_extend_tvm_use_pass_infra.py` (``use_pass_infra.py``) | 00:01.118 | 0.0 MB |
+| :ref:`sphx_glr_how_to_extend_tvm_use_pass_infra.py` (``use_pass_infra.py``) | 00:01.046 | 0.0 MB |
+-------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_extend_tvm_low_level_custom_pass.py` (``low_level_custom_pass.py``) | 00:00.008 | 0.0 MB |
+| :ref:`sphx_glr_how_to_extend_tvm_low_level_custom_pass.py` (``low_level_custom_pass.py``) | 00:00.007 | 0.0 MB |
+-------------------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/how_to/extend_tvm/use_pass_instrument.rst.txt b/docs/_sources/how_to/extend_tvm/use_pass_instrument.rst.txt
index 39e65c4ede..5e7b8668d5 100644
--- a/docs/_sources/how_to/extend_tvm/use_pass_instrument.rst.txt
+++ b/docs/_sources/how_to/extend_tvm/use_pass_instrument.rst.txt
@@ -220,10 +220,10 @@ profile the execution time of each passes.
.. code-block:: none
Printing results of timing profile...
- InferType: 27356us [27356us] (48.95%; 48.95%)
- FoldScaleAxis: 28531us [7us] (51.05%; 51.05%)
- FoldConstant: 28524us [1702us] (51.04%; 99.97%)
- InferType: 26821us [26821us] (47.99%; 94.03%)
+ InferType: 21921us [21921us] (48.72%; 48.72%)
+ FoldScaleAxis: 23073us [7us] (51.28%; 51.28%)
+ FoldConstant: 23066us [1735us] (51.27%; 99.97%)
+ InferType: 21331us [21331us] (47.41%; 92.48%)
@@ -262,10 +262,10 @@ Refer to following sections and :py:func:`tvm.instrument.pass_instrument` for th
.. code-block:: none
Printing results of timing profile...
- InferType: 26704us [26704us] (48.31%; 48.31%)
- FoldScaleAxis: 28571us [6us] (51.69%; 51.69%)
- FoldConstant: 28564us [1746us] (51.68%; 99.98%)
- InferType: 26818us [26818us] (48.52%; 93.89%)
+ InferType: 21382us [21382us] (47.49%; 47.49%)
+ FoldScaleAxis: 23639us [5us] (52.51%; 52.51%)
+ FoldConstant: 23635us [1672us] (52.50%; 99.98%)
+ InferType: 21962us [21962us] (48.78%; 92.92%)
diff --git a/docs/_sources/how_to/optimize_operators/opt_conv_cuda.rst.txt b/docs/_sources/how_to/optimize_operators/opt_conv_cuda.rst.txt
index f85954b386..9cde50a0da 100644
--- a/docs/_sources/how_to/optimize_operators/opt_conv_cuda.rst.txt
+++ b/docs/_sources/how_to/optimize_operators/opt_conv_cuda.rst.txt
@@ -331,7 +331,7 @@ latency of convolution.
.. code-block:: none
- Convolution: 52.545440 ms
+ Convolution: 54.152862 ms
diff --git a/docs/_sources/how_to/optimize_operators/opt_conv_tensorcore.rst.txt b/docs/_sources/how_to/optimize_operators/opt_conv_tensorcore.rst.txt
index 1b91a8a782..26c4416ddd 100644
--- a/docs/_sources/how_to/optimize_operators/opt_conv_tensorcore.rst.txt
+++ b/docs/_sources/how_to/optimize_operators/opt_conv_tensorcore.rst.txt
@@ -598,7 +598,7 @@ be able to run on our build server
.. code-block:: none
- conv2d with tensor core: 13.357066 ms
+ conv2d with tensor core: 6.998237 ms
diff --git a/docs/_sources/how_to/optimize_operators/opt_gemm.rst.txt b/docs/_sources/how_to/optimize_operators/opt_gemm.rst.txt
index 5c12732581..1855a1b6c4 100644
--- a/docs/_sources/how_to/optimize_operators/opt_gemm.rst.txt
+++ b/docs/_sources/how_to/optimize_operators/opt_gemm.rst.txt
@@ -134,8 +134,8 @@ Then we write a baseline implementation, the simplest way to write a matrix mult
.. code-block:: none
- Numpy running time: 0.020222
- Baseline: 3.602864
+ Numpy running time: 0.017828
+ Baseline: 3.281359
@@ -227,7 +227,7 @@ fill 32 * 32 * sizeof(float) which is 4KB in the cache whose total size is 32KB
.. code-block:: none
- Opt1: 0.332328
+ Opt1: 0.307110
@@ -318,7 +318,7 @@ In this tutorial, we chose to vectorize the inner loop row data since it is cach
.. code-block:: none
- Opt2: 0.358482
+ Opt2: 0.330430
@@ -406,7 +406,7 @@ the access pattern for A matrix is more cache friendly.
.. code-block:: none
- Opt3: 0.136659
+ Opt3: 0.113476
@@ -523,7 +523,7 @@ flattening.
.. code-block:: none
- Opt4: 0.112330
+ Opt4: 0.109412
@@ -635,7 +635,7 @@ write to C when all the block results are ready.
.. code-block:: none
- Opt5: 0.113374
+ Opt5: 0.111865
@@ -748,7 +748,7 @@ Furthermore, we can also utilize multi-core processors to do the thread-level pa
.. code-block:: none
- Opt6: 0.148179
+ Opt6: 0.146553
diff --git a/docs/_sources/how_to/optimize_operators/sg_execution_times.rst.txt b/docs/_sources/how_to/optimize_operators/sg_execution_times.rst.txt
index 0812942ec8..732bcfde3a 100644
--- a/docs/_sources/how_to/optimize_operators/sg_execution_times.rst.txt
+++ b/docs/_sources/how_to/optimize_operators/sg_execution_times.rst.txt
@@ -5,12 +5,12 @@
Computation times
=================
-**00:36.875** total execution time for **how_to_optimize_operators** files:
+**00:34.451** total execution time for **how_to_optimize_operators** files:
+-----------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_optimize_operators_opt_gemm.py` (``opt_gemm.py``) | 00:34.125 | 0.0 MB |
+| :ref:`sphx_glr_how_to_optimize_operators_opt_gemm.py` (``opt_gemm.py``) | 00:31.797 | 0.0 MB |
+-----------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_optimize_operators_opt_conv_tensorcore.py` (``opt_conv_tensorcore.py``) | 00:01.630 | 0.0 MB |
+| :ref:`sphx_glr_how_to_optimize_operators_opt_conv_tensorcore.py` (``opt_conv_tensorcore.py``) | 00:01.529 | 0.0 MB |
+-----------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_optimize_operators_opt_conv_cuda.py` (``opt_conv_cuda.py``) | 00:01.120 | 0.0 MB |
+| :ref:`sphx_glr_how_to_optimize_operators_opt_conv_cuda.py` (``opt_conv_cuda.py``) | 00:01.126 | 0.0 MB |
+-----------------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/how_to/tune_with_autoscheduler/sg_execution_times.rst.txt b/docs/_sources/how_to/tune_with_autoscheduler/sg_execution_times.rst.txt
index 1b65ac0142..43998b9681 100644
--- a/docs/_sources/how_to/tune_with_autoscheduler/sg_execution_times.rst.txt
+++ b/docs/_sources/how_to/tune_with_autoscheduler/sg_execution_times.rst.txt
@@ -5,18 +5,18 @@
Computation times
=================
-**10:53.147** total execution time for **how_to_tune_with_autoscheduler** files:
+**10:00.162** total execution time for **how_to_tune_with_autoscheduler** files:
+----------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_conv2d_layer_cuda.py` (``tune_conv2d_layer_cuda.py``) | 06:47.248 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_conv2d_layer_cuda.py` (``tune_conv2d_layer_cuda.py``) | 06:12.870 | 0.0 MB |
+----------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_x86.py` (``tune_network_x86.py``) | 01:48.279 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_x86.py` (``tune_network_x86.py``) | 01:40.793 | 0.0 MB |
+----------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_cuda.py` (``tune_network_cuda.py``) | 01:11.065 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_cuda.py` (``tune_network_cuda.py``) | 01:06.524 | 0.0 MB |
+----------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_sparse_x86.py` (``tune_sparse_x86.py``) | 00:36.619 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_sparse_x86.py` (``tune_sparse_x86.py``) | 00:32.697 | 0.0 MB |
+----------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_arm.py` (``tune_network_arm.py``) | 00:15.794 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_arm.py` (``tune_network_arm.py``) | 00:13.965 | 0.0 MB |
+----------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_mali.py` (``tune_network_mali.py``) | 00:14.143 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_mali.py` (``tune_network_mali.py``) | 00:13.313 | 0.0 MB |
+----------------------------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/how_to/tune_with_autoscheduler/tune_conv2d_layer_cuda.rst.txt b/docs/_sources/how_to/tune_with_autoscheduler/tune_conv2d_layer_cuda.rst.txt
index 8845644a76..532dd759e9 100644
--- a/docs/_sources/how_to/tune_with_autoscheduler/tune_conv2d_layer_cuda.rst.txt
+++ b/docs/_sources/how_to/tune_with_autoscheduler/tune_conv2d_layer_cuda.rst.txt
@@ -243,32 +243,1316 @@ cooperative fetching, unrolling and operator fusion.
@T.prim_func
def main(data: T.Buffer((1, 512, 7, 7), "float32"), kernel: T.Buffer((512, 512, 3, 3), "float32"), bias: T.Buffer((1, 512, 1, 1), "float32"), compute: T.Buffer((1, 512, 7, 7), "float32")):
T.func_attr({"from_legacy_te_schedule": True, "global_symbol": "main", "tir.noalias": True})
- blockIdx_x = T.launch_thread("blockIdx.x", 16)
- conv2d_nchw = T.allocate([2], "float32", "local")
- pad_temp_shared = T.allocate([784], "float32", "shared")
- kernel_shared = T.allocate([512], "float32", "shared")
- threadIdx_x = T.launch_thread("threadIdx.x", 784)
- conv2d_nchw_1 = T.Buffer((2,), data=conv2d_nchw, scope="local", align=8)
- for ff_inner_init in range(2):
- conv2d_nchw_1[ff_inner_init] = T.float32(0)
- for rc_outer_outer, ry_outer_outer, rx_outer_outer in T.grid(32, 3, 3):
- pad_temp_shared_1 = T.Buffer((784,), data=pad_temp_shared, scope="shared")
- with T.launch_thread("threadIdx.x", 784) as threadIdx_x_1:
- for ax0_ax1_fused_ax2_fused_ax3_fused_inner_s in range(4):
- if T.likely(threadIdx_x_1 < 196):
- data_1 = T.Buffer((25088,), data=data.data)
- pad_temp_shared_1[threadIdx_x_1 * 4 + ax0_ax1_fused_ax2_fused_ax3_fused_inner_s] = T.if_then_else(1 <= (threadIdx_x_1 * 4 + ax0_ax1_fused_ax2_fused_ax3_fused_inner_s) % 49 // 7 + ry_outer_outer and (threadIdx_x_1 * 4 + ax0_ax1_fused_ax2_fused_ax3_fused_inner_s) % 49 // 7 + ry_outer_outer < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 4 + ax0_ax1_fused_ax2_fused_ax3_fused_inner_s) % 7 and rx_outer_outer + (threadIdx_x_1 * 4 + ax0_ax1_fused_ax2_fused_ax3_fused_i [...]
- kernel_shared_1 = T.Buffer((512,), data=kernel_shared, scope="shared")
- with T.launch_thread("threadIdx.x", 784) as threadIdx_x_1:
- if T.likely(threadIdx_x_1 < 512):
- kernel_1 = T.Buffer((2359296,), data=kernel.data)
- kernel_shared_1[threadIdx_x_1] = kernel_1[blockIdx_x * 147456 + threadIdx_x_1 // 16 * 4608 + rc_outer_outer * 144 + threadIdx_x_1 % 16 * 9 + ry_outer_outer * 3 + rx_outer_outer]
- for rc_outer_inner, ff_inner in T.grid(16, 2):
- conv2d_nchw_1[ff_inner] = conv2d_nchw_1[ff_inner] + pad_temp_shared_1[rc_outer_inner * 49 + threadIdx_x % 49] * kernel_shared_1[threadIdx_x // 49 * 32 + ff_inner * 16 + rc_outer_inner]
+ blockIdx_x = T.launch_thread("blockIdx.x", 32)
+ conv2d_nchw = T.allocate([14], "float32", "local")
+ pad_temp_shared = T.allocate([2592], "float32", "shared")
+ kernel_shared = T.allocate([4608], "float32", "shared")
+ threadIdx_x = T.launch_thread("threadIdx.x", 56)
+ conv2d_nchw_1 = T.Buffer((4,), data=conv2d_nchw, scope="local", align=8)
+ conv2d_nchw_1[0] = T.float32(0)
+ conv2d_nchw_1[2] = T.float32(0)
+ conv2d_nchw_1[4] = T.float32(0)
+ conv2d_nchw_1[6] = T.float32(0)
+ conv2d_nchw_1[8] = T.float32(0)
+ conv2d_nchw_1[10] = T.float32(0)
+ conv2d_nchw_1[12] = T.float32(0)
+ conv2d_nchw_1[1] = T.float32(0)
+ conv2d_nchw_1[3] = T.float32(0)
+ conv2d_nchw_1[5] = T.float32(0)
+ conv2d_nchw_1[7] = T.float32(0)
+ conv2d_nchw_1[9] = T.float32(0)
+ conv2d_nchw_1[11] = T.float32(0)
+ conv2d_nchw_1[13] = T.float32(0)
+ for rc_outer_outer in range(16):
+ cse_var_2: T.int32 = rc_outer_outer * 1568
+ cse_var_1: T.int32 = rc_outer_outer * 288
+ threadIdx_x_1 = T.env_thread("threadIdx.x")
+ pad_temp_shared_1 = T.Buffer((2592,), data=pad_temp_shared, scope="shared")
+ data_1 = T.Buffer((25088,), data=data.data)
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1] = T.if_then_else(9 <= threadIdx_x_1 and 1 <= threadIdx_x_1 % 9 and threadIdx_x_1 % 9 < 8, data_1[cse_var_2 + threadIdx_x_1 // 9 * 7 + threadIdx_x_1 % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 56] = T.if_then_else(9 <= (threadIdx_x_1 + 56) % 81 and (threadIdx_x_1 + 56) % 81 < 72 and 1 <= (threadIdx_x_1 + 2) % 9 and (threadIdx_x_1 + 2) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 56) // 81 * 49 + (threadIdx_x_1 + 56) % 81 // 9 * 7 + (threadIdx_x_1 + 2) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 112] = T.if_then_else(9 <= (threadIdx_x_1 + 31) % 81 and (threadIdx_x_1 + 31) % 81 < 72 and 1 <= (threadIdx_x_1 + 4) % 9 and (threadIdx_x_1 + 4) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 112) // 81 * 49 + (threadIdx_x_1 + 31) % 81 // 9 * 7 + (threadIdx_x_1 + 4) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 168] = T.if_then_else(3 <= threadIdx_x_1 and 1 <= (threadIdx_x_1 + 6) % 9 and (threadIdx_x_1 + 6) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 168) // 81 * 49 + (threadIdx_x_1 + 6) // 9 * 7 + (threadIdx_x_1 + 6) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 224] = T.if_then_else(9 <= (threadIdx_x_1 + 62) % 81 and (threadIdx_x_1 + 62) % 81 < 72 and 1 <= (threadIdx_x_1 + 8) % 9 and (threadIdx_x_1 + 8) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 224) // 81 * 49 + (threadIdx_x_1 + 62) % 81 // 9 * 7 + (threadIdx_x_1 + 8) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 280] = T.if_then_else(9 <= (threadIdx_x_1 + 37) % 81 and (threadIdx_x_1 + 37) % 81 < 72 and 1 <= (threadIdx_x_1 + 1) % 9 and (threadIdx_x_1 + 1) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 280) // 81 * 49 + (threadIdx_x_1 + 37) % 81 // 9 * 7 + (threadIdx_x_1 + 1) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 336] = T.if_then_else(1 <= (threadIdx_x_1 + 3) % 9 and (threadIdx_x_1 + 3) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 336) // 81 * 49 + (threadIdx_x_1 + 12) // 9 * 7 + (threadIdx_x_1 + 3) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 392] = T.if_then_else(9 <= (threadIdx_x_1 + 68) % 81 and (threadIdx_x_1 + 68) % 81 < 72 and 1 <= (threadIdx_x_1 + 5) % 9 and (threadIdx_x_1 + 5) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 392) // 81 * 49 + (threadIdx_x_1 + 68) % 81 // 9 * 7 + (threadIdx_x_1 + 5) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 448] = T.if_then_else(9 <= (threadIdx_x_1 + 43) % 81 and (threadIdx_x_1 + 43) % 81 < 72 and 1 <= (threadIdx_x_1 + 7) % 9 and (threadIdx_x_1 + 7) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 448) // 81 * 49 + (threadIdx_x_1 + 43) % 81 // 9 * 7 + (threadIdx_x_1 + 7) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 504] = T.if_then_else(threadIdx_x_1 < 54 and 1 <= threadIdx_x_1 % 9 and threadIdx_x_1 % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 504) // 81 * 49 + threadIdx_x_1 // 9 * 7 + threadIdx_x_1 % 9 + 6], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 560] = T.if_then_else(9 <= (threadIdx_x_1 + 74) % 81 and (threadIdx_x_1 + 74) % 81 < 72 and 1 <= (threadIdx_x_1 + 2) % 9 and (threadIdx_x_1 + 2) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 560) // 81 * 49 + (threadIdx_x_1 + 74) % 81 // 9 * 7 + (threadIdx_x_1 + 2) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 616] = T.if_then_else(9 <= (threadIdx_x_1 + 49) % 81 and (threadIdx_x_1 + 49) % 81 < 72 and 1 <= (threadIdx_x_1 + 4) % 9 and (threadIdx_x_1 + 4) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 616) // 81 * 49 + (threadIdx_x_1 + 49) % 81 // 9 * 7 + (threadIdx_x_1 + 4) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 672] = T.if_then_else(threadIdx_x_1 < 48 and 1 <= (threadIdx_x_1 + 6) % 9 and (threadIdx_x_1 + 6) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 672) // 81 * 49 + (threadIdx_x_1 + 24) // 9 * 7 + (threadIdx_x_1 + 6) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 728] = T.if_then_else(9 <= (threadIdx_x_1 + 80) % 81 and (threadIdx_x_1 + 80) % 81 < 72 and 1 <= (threadIdx_x_1 + 8) % 9 and (threadIdx_x_1 + 8) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 728) // 81 * 49 + (threadIdx_x_1 + 80) % 81 // 9 * 7 + (threadIdx_x_1 + 8) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 784] = T.if_then_else(9 <= (threadIdx_x_1 + 55) % 81 and (threadIdx_x_1 + 55) % 81 < 72 and 1 <= (threadIdx_x_1 + 1) % 9 and (threadIdx_x_1 + 1) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 784) // 81 * 49 + (threadIdx_x_1 + 55) % 81 // 9 * 7 + (threadIdx_x_1 + 1) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 840] = T.if_then_else(9 <= (threadIdx_x_1 + 30) % 81 and (threadIdx_x_1 + 30) % 81 < 72 and 1 <= (threadIdx_x_1 + 3) % 9 and (threadIdx_x_1 + 3) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 840) // 81 * 49 + (threadIdx_x_1 + 30) % 81 // 9 * 7 + (threadIdx_x_1 + 3) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 896] = T.if_then_else(4 <= threadIdx_x_1 and 1 <= (threadIdx_x_1 + 5) % 9 and (threadIdx_x_1 + 5) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 896) // 81 * 49 + (threadIdx_x_1 + 5) // 9 * 7 + (threadIdx_x_1 + 5) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 952] = T.if_then_else(9 <= (threadIdx_x_1 + 61) % 81 and (threadIdx_x_1 + 61) % 81 < 72 and 1 <= (threadIdx_x_1 + 7) % 9 and (threadIdx_x_1 + 7) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 952) // 81 * 49 + (threadIdx_x_1 + 61) % 81 // 9 * 7 + (threadIdx_x_1 + 7) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 1008] = T.if_then_else(1 <= (threadIdx_x_1 // 9 + 4) % 9 and (threadIdx_x_1 + 36) % 81 < 72 and 1 <= threadIdx_x_1 % 9 and threadIdx_x_1 % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 1008) // 81 * 49 + (threadIdx_x_1 // 9 + 4) % 9 * 7 + threadIdx_x_1 % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 1064] = T.if_then_else(1 <= (threadIdx_x_1 + 2) % 9 and (threadIdx_x_1 + 2) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 1064) // 81 * 49 + (threadIdx_x_1 + 11) // 9 * 7 + (threadIdx_x_1 + 2) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 1120] = T.if_then_else(9 <= (threadIdx_x_1 + 67) % 81 and (threadIdx_x_1 + 67) % 81 < 72 and 1 <= (threadIdx_x_1 + 4) % 9 and (threadIdx_x_1 + 4) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 1120) // 81 * 49 + (threadIdx_x_1 + 67) % 81 // 9 * 7 + (threadIdx_x_1 + 4) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 1176] = T.if_then_else(9 <= (threadIdx_x_1 + 42) % 81 and (threadIdx_x_1 + 42) % 81 < 72 and 1 <= (threadIdx_x_1 + 6) % 9 and (threadIdx_x_1 + 6) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 1176) // 81 * 49 + (threadIdx_x_1 + 42) % 81 // 9 * 7 + (threadIdx_x_1 + 6) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 1232] = T.if_then_else(threadIdx_x_1 < 55 and 1 <= (threadIdx_x_1 + 8) % 9 and (threadIdx_x_1 + 8) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 1232) // 81 * 49 + (threadIdx_x_1 + 17) // 9 * 7 + (threadIdx_x_1 + 8) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 1288] = T.if_then_else(9 <= (threadIdx_x_1 + 73) % 81 and (threadIdx_x_1 + 73) % 81 < 72 and 1 <= (threadIdx_x_1 + 1) % 9 and (threadIdx_x_1 + 1) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 1288) // 81 * 49 + (threadIdx_x_1 + 73) % 81 // 9 * 7 + (threadIdx_x_1 + 1) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 1344] = T.if_then_else(9 <= (threadIdx_x_1 + 48) % 81 and (threadIdx_x_1 + 48) % 81 < 72 and 1 <= (threadIdx_x_1 + 3) % 9 and (threadIdx_x_1 + 3) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 1344) // 81 * 49 + (threadIdx_x_1 + 48) % 81 // 9 * 7 + (threadIdx_x_1 + 3) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 1400] = T.if_then_else(threadIdx_x_1 < 49 and 1 <= (threadIdx_x_1 + 5) % 9 and (threadIdx_x_1 + 5) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 1400) // 81 * 49 + (threadIdx_x_1 + 23) // 9 * 7 + (threadIdx_x_1 + 5) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 1456] = T.if_then_else(9 <= (threadIdx_x_1 + 79) % 81 and (threadIdx_x_1 + 79) % 81 < 72 and 1 <= (threadIdx_x_1 + 7) % 9 and (threadIdx_x_1 + 7) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 1456) // 81 * 49 + (threadIdx_x_1 + 79) % 81 // 9 * 7 + (threadIdx_x_1 + 7) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 1512] = T.if_then_else(1 <= (threadIdx_x_1 // 9 + 6) % 9 and (threadIdx_x_1 + 54) % 81 < 72 and 1 <= threadIdx_x_1 % 9 and threadIdx_x_1 % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 1512) // 81 * 49 + (threadIdx_x_1 // 9 + 6) % 9 * 7 + threadIdx_x_1 % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 1568] = T.if_then_else(9 <= (threadIdx_x_1 + 29) % 81 and (threadIdx_x_1 + 29) % 81 < 72 and 1 <= (threadIdx_x_1 + 2) % 9 and (threadIdx_x_1 + 2) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 1568) // 81 * 49 + (threadIdx_x_1 + 29) % 81 // 9 * 7 + (threadIdx_x_1 + 2) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 1624] = T.if_then_else(5 <= threadIdx_x_1 and 1 <= (threadIdx_x_1 + 4) % 9 and (threadIdx_x_1 + 4) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 1624) // 81 * 49 + (threadIdx_x_1 + 4) // 9 * 7 + (threadIdx_x_1 + 4) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 1680] = T.if_then_else(9 <= (threadIdx_x_1 + 60) % 81 and (threadIdx_x_1 + 60) % 81 < 72 and 1 <= (threadIdx_x_1 + 6) % 9 and (threadIdx_x_1 + 6) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 1680) // 81 * 49 + (threadIdx_x_1 + 60) % 81 // 9 * 7 + (threadIdx_x_1 + 6) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 1736] = T.if_then_else(9 <= (threadIdx_x_1 + 35) % 81 and (threadIdx_x_1 + 35) % 81 < 72 and 1 <= (threadIdx_x_1 + 8) % 9 and (threadIdx_x_1 + 8) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 1736) // 81 * 49 + (threadIdx_x_1 + 35) % 81 // 9 * 7 + (threadIdx_x_1 + 8) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 1792] = T.if_then_else(1 <= (threadIdx_x_1 + 1) % 9 and (threadIdx_x_1 + 1) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 1792) // 81 * 49 + (threadIdx_x_1 + 10) // 9 * 7 + (threadIdx_x_1 + 1) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 1848] = T.if_then_else(9 <= (threadIdx_x_1 + 66) % 81 and (threadIdx_x_1 + 66) % 81 < 72 and 1 <= (threadIdx_x_1 + 3) % 9 and (threadIdx_x_1 + 3) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 1848) // 81 * 49 + (threadIdx_x_1 + 66) % 81 // 9 * 7 + (threadIdx_x_1 + 3) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 1904] = T.if_then_else(9 <= (threadIdx_x_1 + 41) % 81 and (threadIdx_x_1 + 41) % 81 < 72 and 1 <= (threadIdx_x_1 + 5) % 9 and (threadIdx_x_1 + 5) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 1904) // 81 * 49 + (threadIdx_x_1 + 41) % 81 // 9 * 7 + (threadIdx_x_1 + 5) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 1960] = T.if_then_else(1 <= (threadIdx_x_1 + 7) % 9 and (threadIdx_x_1 + 7) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 1960) // 81 * 49 + (threadIdx_x_1 + 16) // 9 * 7 + (threadIdx_x_1 + 7) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 2016] = T.if_then_else(1 <= (threadIdx_x_1 // 9 + 8) % 9 and (threadIdx_x_1 + 72) % 81 < 72 and 1 <= threadIdx_x_1 % 9 and threadIdx_x_1 % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 2016) // 81 * 49 + (threadIdx_x_1 // 9 + 8) % 9 * 7 + threadIdx_x_1 % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 2072] = T.if_then_else(9 <= (threadIdx_x_1 + 47) % 81 and (threadIdx_x_1 + 47) % 81 < 72 and 1 <= (threadIdx_x_1 + 2) % 9 and (threadIdx_x_1 + 2) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 2072) // 81 * 49 + (threadIdx_x_1 + 47) % 81 // 9 * 7 + (threadIdx_x_1 + 2) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 2128] = T.if_then_else(threadIdx_x_1 < 50 and 1 <= (threadIdx_x_1 + 4) % 9 and (threadIdx_x_1 + 4) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 2128) // 81 * 49 + (threadIdx_x_1 + 22) // 9 * 7 + (threadIdx_x_1 + 4) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 2184] = T.if_then_else(9 <= (threadIdx_x_1 + 78) % 81 and (threadIdx_x_1 + 78) % 81 < 72 and 1 <= (threadIdx_x_1 + 6) % 9 and (threadIdx_x_1 + 6) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 2184) // 81 * 49 + (threadIdx_x_1 + 78) % 81 // 9 * 7 + (threadIdx_x_1 + 6) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 2240] = T.if_then_else(9 <= (threadIdx_x_1 + 53) % 81 and (threadIdx_x_1 + 53) % 81 < 72 and 1 <= (threadIdx_x_1 + 8) % 9 and (threadIdx_x_1 + 8) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 2240) // 81 * 49 + (threadIdx_x_1 + 53) % 81 // 9 * 7 + (threadIdx_x_1 + 8) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 2296] = T.if_then_else(9 <= (threadIdx_x_1 + 28) % 81 and (threadIdx_x_1 + 28) % 81 < 72 and 1 <= (threadIdx_x_1 + 1) % 9 and (threadIdx_x_1 + 1) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 2296) // 81 * 49 + (threadIdx_x_1 + 28) % 81 // 9 * 7 + (threadIdx_x_1 + 1) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 2352] = T.if_then_else(6 <= threadIdx_x_1 and 1 <= (threadIdx_x_1 + 3) % 9 and (threadIdx_x_1 + 3) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 2352) // 81 * 49 + (threadIdx_x_1 + 3) // 9 * 7 + (threadIdx_x_1 + 3) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 2408] = T.if_then_else(9 <= (threadIdx_x_1 + 59) % 81 and (threadIdx_x_1 + 59) % 81 < 72 and 1 <= (threadIdx_x_1 + 5) % 9 and (threadIdx_x_1 + 5) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 2408) // 81 * 49 + (threadIdx_x_1 + 59) % 81 // 9 * 7 + (threadIdx_x_1 + 5) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 2464] = T.if_then_else(9 <= (threadIdx_x_1 + 34) % 81 and (threadIdx_x_1 + 34) % 81 < 72 and 1 <= (threadIdx_x_1 + 7) % 9 and (threadIdx_x_1 + 7) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 2464) // 81 * 49 + (threadIdx_x_1 + 34) % 81 // 9 * 7 + (threadIdx_x_1 + 7) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 2520] = T.if_then_else(1 <= threadIdx_x_1 % 9 and threadIdx_x_1 % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 2520) // 81 * 49 + threadIdx_x_1 // 9 * 7 + threadIdx_x_1 % 9 - 1], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ if T.likely(threadIdx_x_1 < 16):
+ pad_temp_shared_1[threadIdx_x_1 + 2576] = T.if_then_else(threadIdx_x_1 < 7 and 1 <= (threadIdx_x_1 + 2) % 9 and (threadIdx_x_1 + 2) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 2576) // 81 * 49 + (threadIdx_x_1 + 65) // 9 * 7 + threadIdx_x_1 - 6], T.float32(0))
+ threadIdx_x_2 = T.env_thread("threadIdx.x")
+ kernel_shared_1 = T.Buffer((4608,), data=kernel_shared, scope="shared")
+ kernel_1 = T.Buffer((2359296,), data=kernel.data)
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2] = kernel_1[blockIdx_x * 73728 + cse_var_1 + threadIdx_x_2]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 56] = kernel_1[blockIdx_x * 73728 + cse_var_1 + threadIdx_x_2 + 56]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 112] = kernel_1[blockIdx_x * 73728 + cse_var_1 + threadIdx_x_2 + 112]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 168] = kernel_1[blockIdx_x * 73728 + cse_var_1 + threadIdx_x_2 + 168]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 224] = kernel_1[blockIdx_x * 73728 + cse_var_1 + threadIdx_x_2 + 224]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 280] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 280) // 288 * 4608 + cse_var_1 + (threadIdx_x_2 + 280) % 288]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 336] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 336) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 48]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 392] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 392) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 104]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 448] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 448) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 160]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 504] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 504) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 216]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 560] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 560) // 288 * 4608 + cse_var_1 + (threadIdx_x_2 + 272) % 288]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 616] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 616) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 40]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 672] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 672) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 96]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 728] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 728) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 152]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 784] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 784) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 208]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 840] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 840) // 288 * 4608 + cse_var_1 + (threadIdx_x_2 + 264) % 288]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 896] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 896) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 32]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 952] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 952) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 88]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1008] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1008) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 144]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1064] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1064) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 200]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1120] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1120) // 288 * 4608 + cse_var_1 + (threadIdx_x_2 + 256) % 288]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1176] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1176) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 24]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1232] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1232) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 80]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1288] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1288) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 136]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1344] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1344) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 192]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1400] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1400) // 288 * 4608 + cse_var_1 + (threadIdx_x_2 + 248) % 288]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1456] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1456) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 16]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1512] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1512) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 72]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1568] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1568) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 128]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1624] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1624) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 184]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1680] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1680) // 288 * 4608 + cse_var_1 + (threadIdx_x_2 + 240) % 288]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1736] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1736) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 8]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1792] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1792) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 64]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1848] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1848) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 120]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1904] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1904) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 176]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1960] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1960) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 232]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 2016] = kernel_1[blockIdx_x * 73728 + cse_var_1 + threadIdx_x_2 + 32256]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 2072] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2072) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 56]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 2128] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2128) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 112]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 2184] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2184) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 168]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 2240] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2240) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 224]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 2296] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2296) // 288 * 4608 + cse_var_1 + (threadIdx_x_2 + 280) % 288]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 2352] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2352) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 48]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 2408] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2408) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 104]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 2464] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2464) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 160]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 2520] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2520) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 216]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 2576] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2576) // 288 * 4608 + cse_var_1 + (threadIdx_x_2 + 272) % 288]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 2632] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2632) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 40]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 2688] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2688) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 96]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 2744] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2744) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 152]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 2800] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2800) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 208]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 2856] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2856) // 288 * 4608 + cse_var_1 + (threadIdx_x_2 + 264) % 288]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 2912] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2912) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 32]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 2968] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2968) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 88]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 3024] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3024) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 144]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 3080] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3080) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 200]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 3136] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3136) // 288 * 4608 + cse_var_1 + (threadIdx_x_2 + 256) % 288]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 3192] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3192) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 24]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 3248] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3248) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 80]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 3304] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3304) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 136]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 3360] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3360) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 192]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 3416] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3416) // 288 * 4608 + cse_var_1 + (threadIdx_x_2 + 248) % 288]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 3472] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3472) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 16]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 3528] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3528) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 72]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 3584] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3584) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 128]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 3640] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3640) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 184]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 3696] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3696) // 288 * 4608 + cse_var_1 + (threadIdx_x_2 + 240) % 288]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 3752] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3752) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 8]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 3808] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3808) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 64]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 3864] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3864) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 120]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 3920] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3920) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 176]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 3976] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3976) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 232]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 4032] = kernel_1[blockIdx_x * 73728 + cse_var_1 + threadIdx_x_2 + 64512]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 4088] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 4088) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 56]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 4144] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 4144) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 112]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 4200] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 4200) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 168]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 4256] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 4256) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 224]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 4312] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 4312) // 288 * 4608 + cse_var_1 + (threadIdx_x_2 + 280) % 288]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 4368] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 4368) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 48]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 4424] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 4424) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 104]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 4480] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 4480) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 160]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 4536] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 4536) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 216]
+ with T.launch_thread(threadIdx_x_2, 56):
+ if T.likely(threadIdx_x_2 < 16):
+ kernel_shared_1[threadIdx_x_2 + 4592] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 4592) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 272]
+ for rc_outer_inner in range(4):
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 9] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 18] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 27] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 36] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 45] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 54] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 288]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 9] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 288]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 18] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 288]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 27] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 288]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 36] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 288]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 45] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 288]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 54] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 288]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 1] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 1]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 10] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 1]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 19] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 1]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 28] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 1]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 37] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 1]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 46] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 1]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 55] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 1]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 1] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 289]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 10] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 289]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 19] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 289]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 28] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 289]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 37] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 289]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 46] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 289]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 55] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 289]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 2] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 2]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 11] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 2]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 20] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 2]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 29] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 2]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 38] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 2]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 47] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 2]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 56] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 2]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 2] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 290]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 11] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 290]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 20] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 290]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 29] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 290]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 38] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 290]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 47] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 290]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 56] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 290]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 81] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 9]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 90] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 9]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 99] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 9]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 108] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 9]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 117] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 9]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 126] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 9]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 135] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 9]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 81] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 297]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 90] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 297]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 99] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 297]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 108] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 297]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 117] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 297]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 126] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 297]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 135] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 297]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 82] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 10]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 91] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 10]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 100] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 10]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 109] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 10]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 118] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 10]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 127] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 10]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 136] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 10]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 82] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 298]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 91] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 298]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 100] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 298]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 109] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 298]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 118] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 298]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 127] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 298]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 136] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 298]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 83] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 11]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 92] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 11]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 101] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 11]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 110] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 11]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 119] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 11]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 128] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 11]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 137] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 11]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 83] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 299]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 92] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 299]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 101] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 299]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 110] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 299]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 119] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 299]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 128] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 299]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 137] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 299]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 162] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 18]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 171] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 18]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 180] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 18]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 189] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 18]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 198] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 18]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 207] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 18]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 216] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 18]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 162] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 306]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 171] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 306]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 180] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 306]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 189] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 306]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 198] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 306]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 207] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 306]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 216] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 306]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 163] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 19]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 172] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 19]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 181] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 19]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 190] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 19]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 199] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 19]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 208] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 19]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 217] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 19]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 163] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 307]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 172] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 307]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 181] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 307]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 190] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 307]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 199] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 307]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 208] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 307]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 217] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 307]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 164] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 20]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 173] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 20]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 182] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 20]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 191] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 20]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 200] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 20]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 209] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 20]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 218] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 20]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 164] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 308]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 173] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 308]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 182] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 308]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 191] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 308]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 200] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 308]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 209] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 308]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 218] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 308]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 243] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 27]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 252] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 27]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 261] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 27]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 270] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 27]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 279] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 27]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 288] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 27]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 297] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 27]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 243] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 315]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 252] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 315]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 261] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 315]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 270] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 315]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 279] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 315]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 288] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 315]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 297] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 315]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 244] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 28]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 253] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 28]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 262] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 28]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 271] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 28]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 280] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 28]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 289] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 28]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 298] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 28]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 244] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 316]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 253] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 316]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 262] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 316]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 271] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 316]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 280] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 316]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 289] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 316]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 298] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 316]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 245] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 29]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 254] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 29]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 263] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 29]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 272] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 29]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 281] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 29]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 290] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 29]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 299] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 29]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 245] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 317]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 254] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 317]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 263] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 317]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 272] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 317]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 281] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 317]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 290] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 317]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 299] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 317]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 324] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 36]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 333] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 36]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 342] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 36]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 351] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 36]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 360] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 36]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 369] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 36]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 378] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 36]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 324] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 324]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 333] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 324]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 342] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 324]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 351] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 324]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 360] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 324]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 369] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 324]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 378] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 324]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 325] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 37]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 334] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 37]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 343] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 37]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 352] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 37]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 361] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 37]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 370] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 37]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 379] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 37]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 325] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 325]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 334] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 325]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 343] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 325]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 352] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 325]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 361] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 325]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 370] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 325]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 379] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 325]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 326] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 38]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 335] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 38]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 344] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 38]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 353] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 38]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 362] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 38]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 371] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 38]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 380] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 38]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 326] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 326]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 335] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 326]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 344] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 326]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 353] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 326]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 362] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 326]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 371] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 326]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 380] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 326]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 405] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 45]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 414] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 45]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 423] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 45]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 432] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 45]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 441] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 45]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 450] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 45]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 459] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 45]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 405] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 333]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 414] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 333]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 423] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 333]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 432] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 333]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 441] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 333]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 450] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 333]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 459] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 333]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 406] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 46]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 415] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 46]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 424] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 46]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 433] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 46]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 442] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 46]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 451] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 46]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 460] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 46]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 406] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 334]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 415] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 334]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 424] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 334]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 433] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 334]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 442] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 334]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 451] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 334]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 460] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 334]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 407] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 47]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 416] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 47]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 425] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 47]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 434] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 47]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 443] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 47]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 452] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 47]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 461] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 47]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 407] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 335]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 416] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 335]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 425] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 335]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 434] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 335]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 443] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 335]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 452] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 335]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 461] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 335]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 486] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 54]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 495] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 54]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 504] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 54]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 513] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 54]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 522] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 54]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 531] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 54]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 540] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 54]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 486] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 342]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 495] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 342]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 504] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 342]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 513] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 342]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 522] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 342]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 531] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 342]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 540] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 342]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 487] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 55]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 496] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 55]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 505] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 55]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 514] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 55]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 523] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 55]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 532] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 55]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 541] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 55]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 487] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 343]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 496] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 343]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 505] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 343]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 514] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 343]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 523] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 343]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 532] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 343]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 541] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 343]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 488] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 56]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 497] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 56]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 506] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 56]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 515] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 56]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 524] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 56]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 533] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 56]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 542] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 56]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 488] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 344]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 497] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 344]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 506] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 344]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 515] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 344]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 524] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 344]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 533] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 344]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 542] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 344]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 567] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 63]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 576] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 63]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 585] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 63]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 594] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 63]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 603] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 63]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 612] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 63]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 621] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 63]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 567] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 351]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 576] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 351]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 585] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 351]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 594] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 351]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 603] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 351]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 612] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 351]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 621] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 351]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 568] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 64]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 577] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 64]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 586] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 64]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 595] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 64]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 604] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 64]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 613] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 64]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 622] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 64]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 568] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 352]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 577] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 352]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 586] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 352]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 595] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 352]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 604] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 352]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 613] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 352]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 622] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 352]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 569] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 65]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 578] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 65]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 587] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 65]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 596] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 65]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 605] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 65]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 614] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 65]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 623] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 65]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 569] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 353]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 578] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 353]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 587] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 353]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 596] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 353]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 605] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 353]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 614] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 353]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 623] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 353]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 9] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 3]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 18] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 3]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 27] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 3]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 36] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 3]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 45] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 3]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 54] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 3]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 63] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 3]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 9] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 291]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 18] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 291]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 27] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 291]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 36] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 291]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 45] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 291]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 54] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 291]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 63] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 291]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 10] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 4]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 19] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 4]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 28] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 4]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 37] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 4]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 46] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 4]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 55] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 4]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 64] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 4]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 10] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 292]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 19] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 292]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 28] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 292]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 37] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 292]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 46] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 292]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 55] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 292]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 64] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 292]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 11] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 5]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 20] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 5]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 29] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 5]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 38] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 5]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 47] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 5]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 56] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 5]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 65] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 5]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 11] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 293]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 20] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 293]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 29] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 293]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 38] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 293]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 47] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 293]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 56] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 293]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 65] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 293]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 90] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 12]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 99] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 12]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 108] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 12]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 117] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 12]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 126] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 12]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 135] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 12]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 144] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 12]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 90] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 300]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 99] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 300]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 108] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 300]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 117] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 300]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 126] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 300]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 135] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 300]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 144] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 300]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 91] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 13]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 100] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 13]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 109] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 13]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 118] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 13]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 127] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 13]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 136] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 13]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 145] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 13]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 91] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 301]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 100] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 301]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 109] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 301]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 118] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 301]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 127] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 301]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 136] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 301]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 145] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 301]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 92] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 14]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 101] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 14]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 110] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 14]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 119] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 14]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 128] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 14]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 137] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 14]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 146] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 14]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 92] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 302]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 101] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 302]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 110] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 302]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 119] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 302]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 128] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 302]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 137] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 302]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 146] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 302]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 171] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 21]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 180] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 21]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 189] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 21]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 198] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 21]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 207] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 21]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 216] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 21]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 225] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 21]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 171] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 309]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 180] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 309]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 189] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 309]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 198] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 309]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 207] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 309]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 216] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 309]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 225] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 309]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 172] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 22]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 181] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 22]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 190] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 22]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 199] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 22]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 208] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 22]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 217] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 22]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 226] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 22]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 172] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 310]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 181] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 310]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 190] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 310]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 199] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 310]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 208] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 310]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 217] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 310]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 226] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 310]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 173] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 23]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 182] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 23]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 191] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 23]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 200] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 23]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 209] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 23]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 218] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 23]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 227] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 23]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 173] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 311]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 182] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 311]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 191] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 311]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 200] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 311]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 209] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 311]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 218] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 311]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 227] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 311]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 252] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 30]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 261] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 30]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 270] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 30]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 279] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 30]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 288] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 30]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 297] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 30]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 306] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 30]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 252] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 318]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 261] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 318]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 270] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 318]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 279] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 318]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 288] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 318]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 297] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 318]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 306] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 318]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 253] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 31]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 262] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 31]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 271] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 31]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 280] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 31]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 289] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 31]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 298] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 31]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 307] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 31]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 253] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 319]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 262] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 319]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 271] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 319]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 280] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 319]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 289] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 319]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 298] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 319]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 307] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 319]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 254] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 32]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 263] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 32]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 272] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 32]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 281] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 32]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 290] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 32]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 299] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 32]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 308] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 32]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 254] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 320]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 263] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 320]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 272] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 320]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 281] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 320]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 290] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 320]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 299] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 320]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 308] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 320]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 333] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 39]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 342] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 39]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 351] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 39]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 360] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 39]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 369] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 39]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 378] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 39]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 387] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 39]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 333] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 327]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 342] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 327]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 351] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 327]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 360] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 327]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 369] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 327]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 378] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 327]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 387] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 327]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 334] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 40]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 343] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 40]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 352] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 40]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 361] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 40]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 370] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 40]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 379] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 40]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 388] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 40]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 334] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 328]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 343] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 328]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 352] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 328]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 361] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 328]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 370] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 328]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 379] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 328]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 388] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 328]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 335] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 41]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 344] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 41]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 353] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 41]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 362] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 41]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 371] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 41]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 380] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 41]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 389] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 41]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 335] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 329]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 344] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 329]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 353] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 329]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 362] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 329]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 371] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 329]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 380] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 329]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 389] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 329]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 414] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 48]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 423] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 48]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 432] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 48]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 441] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 48]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 450] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 48]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 459] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 48]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 468] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 48]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 414] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 336]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 423] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 336]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 432] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 336]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 441] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 336]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 450] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 336]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 459] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 336]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 468] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 336]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 415] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 49]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 424] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 49]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 433] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 49]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 442] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 49]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 451] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 49]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 460] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 49]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 469] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 49]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 415] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 337]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 424] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 337]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 433] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 337]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 442] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 337]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 451] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 337]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 460] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 337]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 469] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 337]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 416] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 50]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 425] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 50]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 434] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 50]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 443] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 50]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 452] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 50]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 461] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 50]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 470] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 50]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 416] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 338]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 425] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 338]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 434] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 338]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 443] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 338]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 452] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 338]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 461] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 338]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 470] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 338]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 495] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 57]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 504] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 57]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 513] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 57]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 522] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 57]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 531] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 57]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 540] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 57]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 549] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 57]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 495] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 345]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 504] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 345]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 513] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 345]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 522] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 345]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 531] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 345]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 540] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 345]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 549] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 345]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 496] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 58]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 505] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 58]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 514] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 58]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 523] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 58]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 532] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 58]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 541] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 58]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 550] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 58]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 496] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 346]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 505] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 346]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 514] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 346]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 523] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 346]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 532] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 346]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 541] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 346]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 550] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 346]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 497] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 59]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 506] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 59]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 515] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 59]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 524] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 59]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 533] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 59]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 542] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 59]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 551] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 59]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 497] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 347]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 506] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 347]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 515] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 347]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 524] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 347]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 533] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 347]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 542] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 347]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 551] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 347]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 576] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 66]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 585] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 66]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 594] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 66]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 603] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 66]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 612] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 66]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 621] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 66]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 630] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 66]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 576] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 354]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 585] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 354]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 594] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 354]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 603] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 354]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 612] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 354]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 621] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 354]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 630] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 354]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 577] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 67]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 586] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 67]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 595] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 67]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 604] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 67]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 613] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 67]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 622] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 67]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 631] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 67]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 577] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 355]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 586] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 355]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 595] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 355]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 604] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 355]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 613] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 355]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 622] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 355]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 631] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 355]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 578] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 68]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 587] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 68]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 596] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 68]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 605] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 68]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 614] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 68]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 623] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 68]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 632] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 68]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 578] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 356]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 587] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 356]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 596] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 356]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 605] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 356]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 614] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 356]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 623] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 356]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 632] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 356]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 18] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 6]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 27] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 6]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 36] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 6]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 45] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 6]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 54] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 6]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 63] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 6]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 72] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 6]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 18] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 294]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 27] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 294]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 36] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 294]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 45] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 294]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 54] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 294]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 63] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 294]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 72] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 294]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 19] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 7]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 28] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 7]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 37] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 7]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 46] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 7]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 55] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 7]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 64] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 7]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 73] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 7]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 19] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 295]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 28] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 295]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 37] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 295]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 46] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 295]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 55] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 295]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 64] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 295]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 73] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 295]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 20] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 8]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 29] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 8]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 38] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 8]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 47] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 8]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 56] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 8]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 65] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 8]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 74] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 8]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 20] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 296]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 29] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 296]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 38] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 296]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 47] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 296]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 56] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 296]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 65] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 296]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 74] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 296]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 99] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 15]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 108] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 15]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 117] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 15]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 126] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 15]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 135] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 15]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 144] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 15]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 153] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 15]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 99] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 303]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 108] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 303]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 117] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 303]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 126] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 303]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 135] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 303]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 144] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 303]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 153] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 303]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 100] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 16]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 109] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 16]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 118] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 16]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 127] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 16]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 136] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 16]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 145] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 16]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 154] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 16]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 100] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 304]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 109] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 304]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 118] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 304]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 127] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 304]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 136] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 304]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 145] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 304]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 154] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 304]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 101] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 17]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 110] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 17]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 119] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 17]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 128] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 17]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 137] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 17]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 146] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 17]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 155] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 17]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 101] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 305]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 110] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 305]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 119] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 305]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 128] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 305]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 137] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 305]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 146] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 305]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 155] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 305]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 180] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 24]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 189] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 24]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 198] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 24]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 207] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 24]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 216] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 24]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 225] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 24]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 234] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 24]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 180] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 312]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 189] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 312]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 198] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 312]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 207] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 312]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 216] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 312]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 225] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 312]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 234] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 312]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 181] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 25]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 190] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 25]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 199] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 25]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 208] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 25]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 217] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 25]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 226] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 25]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 235] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 25]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 181] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 313]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 190] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 313]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 199] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 313]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 208] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 313]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 217] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 313]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 226] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 313]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 235] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 313]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 182] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 26]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 191] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 26]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 200] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 26]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 209] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 26]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 218] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 26]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 227] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 26]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 236] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 26]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 182] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 314]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 191] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 314]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 200] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 314]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 209] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 314]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 218] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 314]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 227] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 314]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 236] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 314]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 261] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 33]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 270] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 33]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 279] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 33]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 288] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 33]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 297] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 33]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 306] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 33]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 315] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 33]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 261] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 321]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 270] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 321]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 279] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 321]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 288] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 321]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 297] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 321]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 306] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 321]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 315] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 321]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 262] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 34]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 271] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 34]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 280] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 34]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 289] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 34]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 298] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 34]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 307] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 34]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 316] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 34]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 262] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 322]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 271] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 322]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 280] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 322]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 289] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 322]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 298] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 322]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 307] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 322]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 316] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 322]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 263] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 35]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 272] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 35]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 281] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 35]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 290] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 35]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 299] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 35]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 308] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 35]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 317] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 35]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 263] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 323]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 272] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 323]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 281] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 323]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 290] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 323]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 299] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 323]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 308] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 323]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 317] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 323]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 342] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 42]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 351] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 42]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 360] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 42]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 369] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 42]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 378] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 42]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 387] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 42]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 396] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 42]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 342] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 330]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 351] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 330]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 360] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 330]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 369] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 330]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 378] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 330]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 387] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 330]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 396] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 330]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 343] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 43]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 352] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 43]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 361] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 43]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 370] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 43]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 379] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 43]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 388] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 43]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 397] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 43]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 343] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 331]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 352] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 331]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 361] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 331]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 370] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 331]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 379] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 331]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 388] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 331]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 397] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 331]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 344] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 44]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 353] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 44]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 362] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 44]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 371] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 44]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 380] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 44]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 389] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 44]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 398] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 44]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 344] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 332]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 353] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 332]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 362] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 332]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 371] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 332]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 380] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 332]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 389] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 332]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 398] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 332]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 423] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 51]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 432] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 51]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 441] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 51]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 450] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 51]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 459] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 51]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 468] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 51]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 477] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 51]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 423] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 339]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 432] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 339]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 441] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 339]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 450] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 339]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 459] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 339]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 468] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 339]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 477] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 339]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 424] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 52]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 433] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 52]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 442] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 52]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 451] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 52]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 460] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 52]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 469] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 52]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 478] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 52]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 424] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 340]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 433] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 340]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 442] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 340]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 451] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 340]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 460] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 340]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 469] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 340]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 478] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 340]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 425] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 53]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 434] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 53]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 443] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 53]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 452] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 53]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 461] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 53]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 470] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 53]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 479] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 53]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 425] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 341]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 434] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 341]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 443] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 341]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 452] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 341]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 461] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 341]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 470] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 341]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 479] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 341]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 504] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 60]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 513] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 60]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 522] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 60]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 531] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 60]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 540] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 60]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 549] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 60]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 558] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 60]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 504] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 348]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 513] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 348]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 522] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 348]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 531] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 348]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 540] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 348]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 549] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 348]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 558] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 348]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 505] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 61]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 514] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 61]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 523] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 61]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 532] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 61]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 541] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 61]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 550] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 61]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 559] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 61]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 505] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 349]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 514] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 349]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 523] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 349]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 532] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 349]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 541] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 349]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 550] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 349]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 559] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 349]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 506] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 62]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 515] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 62]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 524] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 62]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 533] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 62]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 542] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 62]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 551] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 62]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 560] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 62]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 506] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 350]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 515] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 350]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 524] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 350]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 533] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 350]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 542] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 350]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 551] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 350]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 560] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 350]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 585] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 69]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 594] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 69]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 603] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 69]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 612] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 69]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 621] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 69]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 630] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 69]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 639] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 69]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 585] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 357]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 594] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 357]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 603] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 357]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 612] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 357]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 621] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 357]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 630] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 357]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 639] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 357]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 586] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 70]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 595] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 70]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 604] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 70]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 613] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 70]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 622] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 70]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 631] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 70]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 640] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 70]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 586] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 358]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 595] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 358]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 604] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 358]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 613] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 358]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 622] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 358]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 631] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 358]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 640] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 358]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 587] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 71]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 596] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 71]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 605] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 71]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 614] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 71]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 623] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 71]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 632] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 71]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 641] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 71]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 587] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 359]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 596] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 359]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 605] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 359]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 614] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 359]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 623] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 359]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 632] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 359]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 641] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 359]
for i1_inner in range(2):
compute_1 = T.Buffer((25088,), data=compute.data)
bias_1 = T.Buffer((512,), data=bias.data)
- compute_1[blockIdx_x * 1568 + threadIdx_x // 49 * 98 + i1_inner * 49 + threadIdx_x % 49] = T.max(conv2d_nchw_1[i1_inner] + bias_1[blockIdx_x * 32 + threadIdx_x // 49 * 2 + i1_inner], T.float32(0))
+ compute_1[blockIdx_x * 784 + threadIdx_x // 7 * 98 + i1_inner * 49 + threadIdx_x % 7] = T.max(conv2d_nchw_1[i1_inner] + bias_1[blockIdx_x * 16 + threadIdx_x // 7 * 2 + i1_inner], T.float32(0))
+ compute_1[blockIdx_x * 784 + threadIdx_x // 7 * 98 + i1_inner * 49 + threadIdx_x % 7 + 7] = T.max(conv2d_nchw_1[i1_inner + 2] + bias_1[blockIdx_x * 16 + threadIdx_x // 7 * 2 + i1_inner], T.float32(0))
+ compute_1[blockIdx_x * 784 + threadIdx_x // 7 * 98 + i1_inner * 49 + threadIdx_x % 7 + 14] = T.max(conv2d_nchw_1[i1_inner + 4] + bias_1[blockIdx_x * 16 + threadIdx_x // 7 * 2 + i1_inner], T.float32(0))
+ compute_1[blockIdx_x * 784 + threadIdx_x // 7 * 98 + i1_inner * 49 + threadIdx_x % 7 + 21] = T.max(conv2d_nchw_1[i1_inner + 6] + bias_1[blockIdx_x * 16 + threadIdx_x // 7 * 2 + i1_inner], T.float32(0))
+ compute_1[blockIdx_x * 784 + threadIdx_x // 7 * 98 + i1_inner * 49 + threadIdx_x % 7 + 28] = T.max(conv2d_nchw_1[i1_inner + 8] + bias_1[blockIdx_x * 16 + threadIdx_x // 7 * 2 + i1_inner], T.float32(0))
+ compute_1[blockIdx_x * 784 + threadIdx_x // 7 * 98 + i1_inner * 49 + threadIdx_x % 7 + 35] = T.max(conv2d_nchw_1[i1_inner + 10] + bias_1[blockIdx_x * 16 + threadIdx_x // 7 * 2 + i1_inner], T.float32(0))
+ compute_1[blockIdx_x * 784 + threadIdx_x // 7 * 98 + i1_inner * 49 + threadIdx_x % 7 + 42] = T.max(conv2d_nchw_1[i1_inner + 12] + bias_1[blockIdx_x * 16 + threadIdx_x // 7 * 2 + i1_inner], T.float32(0))
@@ -318,7 +1602,7 @@ We build the binary and check its correctness and performance.
.. code-block:: none
- Execution time of this operator: 0.405 ms
+ Execution time of this operator: 0.234 ms
@@ -368,32 +1652,32 @@ They can be used for debugging and learning the behavior of the auto-scheduler.
conv2d_nchw_nn_o_o_o_o, conv2d_nchw_nn_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_nn_o_o_o_i, factor=1)
conv2d_nchw_ff_o_i, conv2d_nchw_ff_i = s[conv2d_nchw].split(conv2d_nchw_ff, factor=2)
conv2d_nchw_ff_o_o_i, conv2d_nchw_ff_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_i, factor=1)
- conv2d_nchw_ff_o_o_o_i, conv2d_nchw_ff_o_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_o_i, factor=16)
+ conv2d_nchw_ff_o_o_o_i, conv2d_nchw_ff_o_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_o_i, factor=8)
conv2d_nchw_ff_o_o_o_o, conv2d_nchw_ff_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_o_o_i, factor=1)
conv2d_nchw_yy_o_i, conv2d_nchw_yy_i = s[conv2d_nchw].split(conv2d_nchw_yy, factor=1)
conv2d_nchw_yy_o_o_i, conv2d_nchw_yy_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_i, factor=1)
- conv2d_nchw_yy_o_o_o_i, conv2d_nchw_yy_o_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_o_i, factor=7)
- conv2d_nchw_yy_o_o_o_o, conv2d_nchw_yy_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_o_o_i, factor=1)
+ conv2d_nchw_yy_o_o_o_i, conv2d_nchw_yy_o_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_o_i, factor=1)
+ conv2d_nchw_yy_o_o_o_o, conv2d_nchw_yy_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_o_o_i, factor=7)
conv2d_nchw_xx_o_i, conv2d_nchw_xx_i = s[conv2d_nchw].split(conv2d_nchw_xx, factor=1)
conv2d_nchw_xx_o_o_i, conv2d_nchw_xx_o_i = s[conv2d_nchw].split(conv2d_nchw_xx_o_i, factor=1)
conv2d_nchw_xx_o_o_o_i, conv2d_nchw_xx_o_o_i = s[conv2d_nchw].split(conv2d_nchw_xx_o_o_i, factor=7)
conv2d_nchw_xx_o_o_o_o, conv2d_nchw_xx_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_xx_o_o_o_i, factor=1)
- conv2d_nchw_rc_o_i, conv2d_nchw_rc_i = s[conv2d_nchw].split(conv2d_nchw_rc, factor=1)
- conv2d_nchw_rc_o_o, conv2d_nchw_rc_o_i = s[conv2d_nchw].split(conv2d_nchw_rc_o_i, factor=16)
+ conv2d_nchw_rc_o_i, conv2d_nchw_rc_i = s[conv2d_nchw].split(conv2d_nchw_rc, factor=8)
+ conv2d_nchw_rc_o_o, conv2d_nchw_rc_o_i = s[conv2d_nchw].split(conv2d_nchw_rc_o_i, factor=4)
conv2d_nchw_ry_o_i, conv2d_nchw_ry_i = s[conv2d_nchw].split(conv2d_nchw_ry, factor=1)
- conv2d_nchw_ry_o_o, conv2d_nchw_ry_o_i = s[conv2d_nchw].split(conv2d_nchw_ry_o_i, factor=1)
- conv2d_nchw_rx_o_i, conv2d_nchw_rx_i = s[conv2d_nchw].split(conv2d_nchw_rx, factor=1)
+ conv2d_nchw_ry_o_o, conv2d_nchw_ry_o_i = s[conv2d_nchw].split(conv2d_nchw_ry_o_i, factor=3)
+ conv2d_nchw_rx_o_i, conv2d_nchw_rx_i = s[conv2d_nchw].split(conv2d_nchw_rx, factor=3)
conv2d_nchw_rx_o_o, conv2d_nchw_rx_o_i = s[conv2d_nchw].split(conv2d_nchw_rx_o_i, factor=1)
s[conv2d_nchw].reorder(conv2d_nchw_nn_o_o_o_o, conv2d_nchw_ff_o_o_o_o, conv2d_nchw_yy_o_o_o_o, conv2d_nchw_xx_o_o_o_o, conv2d_nchw_nn_o_o_o_i, conv2d_nchw_ff_o_o_o_i, conv2d_nchw_yy_o_o_o_i, conv2d_nchw_xx_o_o_o_i, conv2d_nchw_nn_o_o_i, conv2d_nchw_ff_o_o_i, conv2d_nchw_yy_o_o_i, conv2d_nchw_xx_o_o_i, conv2d_nchw_rc_o_o, conv2d_nchw_ry_o_o, conv2d_nchw_rx_o_o, conv2d_nchw_rc_o_i, conv2d_nchw_ry_o_i, conv2d_nchw_rx_o_i, conv2d_nchw_nn_o_i, conv2d_nchw_ff_o_i, conv2d_nchw_yy_o_i, conv2 [...]
compute_i0_o_i, compute_i0_i = s[compute].split(compute_i0, factor=1)
compute_i0_o_o_i, compute_i0_o_i = s[compute].split(compute_i0_o_i, factor=1)
compute_i0_o_o_o, compute_i0_o_o_i = s[compute].split(compute_i0_o_o_i, factor=1)
compute_i1_o_i, compute_i1_i = s[compute].split(compute_i1, factor=2)
- compute_i1_o_o_i, compute_i1_o_i = s[compute].split(compute_i1_o_i, factor=16)
+ compute_i1_o_o_i, compute_i1_o_i = s[compute].split(compute_i1_o_i, factor=8)
compute_i1_o_o_o, compute_i1_o_o_i = s[compute].split(compute_i1_o_o_i, factor=1)
compute_i2_o_i, compute_i2_i = s[compute].split(compute_i2, factor=1)
- compute_i2_o_o_i, compute_i2_o_i = s[compute].split(compute_i2_o_i, factor=7)
- compute_i2_o_o_o, compute_i2_o_o_i = s[compute].split(compute_i2_o_o_i, factor=1)
+ compute_i2_o_o_i, compute_i2_o_i = s[compute].split(compute_i2_o_i, factor=1)
+ compute_i2_o_o_o, compute_i2_o_o_i = s[compute].split(compute_i2_o_o_i, factor=7)
compute_i3_o_i, compute_i3_i = s[compute].split(compute_i3, factor=1)
compute_i3_o_o_i, compute_i3_o_i = s[compute].split(compute_i3_o_i, factor=7)
compute_i3_o_o_o, compute_i3_o_o_i = s[compute].split(compute_i3_o_o_i, factor=1)
@@ -415,14 +1699,14 @@ They can be used for debugging and learning the behavior of the auto-scheduler.
kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused = s[kernel_shared].fuse(kernel_shared_ax0, kernel_shared_ax1, kernel_shared_ax2, kernel_shared_ax3)
kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_i = s[kernel_shared].split(kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused, factor=1)
s[kernel_shared].vectorize(kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_i)
- kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_o, kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i = s[kernel_shared].split(kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, factor=784)
+ kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_o, kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i = s[kernel_shared].split(kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, factor=56)
s[kernel_shared].bind(kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i, te.thread_axis("threadIdx.x"))
pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused = s[pad_temp_shared].fuse(pad_temp_shared_ax0, pad_temp_shared_ax1, pad_temp_shared_ax2, pad_temp_shared_ax3)
- pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_i = s[pad_temp_shared].split(pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused, factor=4)
+ pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_i = s[pad_temp_shared].split(pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused, factor=1)
s[pad_temp_shared].vectorize(pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_i)
- pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_o, pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i = s[pad_temp_shared].split(pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, factor=784)
+ pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_o, pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i = s[pad_temp_shared].split(pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, factor=56)
s[pad_temp_shared].bind(pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i, te.thread_axis("threadIdx.x"))
- s[conv2d_nchw].pragma(conv2d_nchw_nn_o_o_o_o, "auto_unroll_max_step", 0)
+ s[conv2d_nchw].pragma(conv2d_nchw_nn_o_o_o_o, "auto_unroll_max_step", 1024)
s[conv2d_nchw].pragma(conv2d_nchw_nn_o_o_o_o, "unroll_explicit", True)
CUDA source code:
@@ -440,36 +1724,1180 @@ They can be used for debugging and learning the behavior of the auto-scheduler.
#define int64_t long long
#define uint64_t unsigned long long
#endif
- extern "C" __global__ void __launch_bounds__(784) default_function_kernel0(float* __restrict__ data, float* __restrict__ kernel, float* __restrict__ compute, float* __restrict__ bias) {
- float conv2d_nchw[2];
- __shared__ float pad_temp_shared[784];
- __shared__ float kernel_shared[512];
- for (int ff_inner_init = 0; ff_inner_init < 2; ++ff_inner_init) {
- conv2d_nchw[ff_inner_init] = 0.000000e+00f;
- }
- for (int rc_outer_outer = 0; rc_outer_outer < 32; ++rc_outer_outer) {
- for (int ry_outer_outer = 0; ry_outer_outer < 3; ++ry_outer_outer) {
- for (int rx_outer_outer = 0; rx_outer_outer < 3; ++rx_outer_outer) {
- __syncthreads();
- for (int ax0_ax1_fused_ax2_fused_ax3_fused_inner_s = 0; ax0_ax1_fused_ax2_fused_ax3_fused_inner_s < 4; ++ax0_ax1_fused_ax2_fused_ax3_fused_inner_s) {
- if (((int)threadIdx.x) < 196) {
- pad_temp_shared[((((int)threadIdx.x) * 4) + ax0_ax1_fused_ax2_fused_ax3_fused_inner_s)] = (((((1 <= (((((((int)threadIdx.x) * 4) + ax0_ax1_fused_ax2_fused_ax3_fused_inner_s) % 49) / 7) + ry_outer_outer)) && ((((((((int)threadIdx.x) * 4) + ax0_ax1_fused_ax2_fused_ax3_fused_inner_s) % 49) / 7) + ry_outer_outer) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 4) + ax0_ax1_fused_ax2_fused_ax3_fused_inner_s) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 4) + a [...]
- }
- }
- if (((int)threadIdx.x) < 512) {
- kernel_shared[((int)threadIdx.x)] = kernel[((((((((int)blockIdx.x) * 147456) + ((((int)threadIdx.x) >> 4) * 4608)) + (rc_outer_outer * 144)) + ((((int)threadIdx.x) & 15) * 9)) + (ry_outer_outer * 3)) + rx_outer_outer)];
- }
- __syncthreads();
- for (int rc_outer_inner = 0; rc_outer_inner < 16; ++rc_outer_inner) {
- for (int ff_inner = 0; ff_inner < 2; ++ff_inner) {
- conv2d_nchw[ff_inner] = (conv2d_nchw[ff_inner] + (pad_temp_shared[((rc_outer_inner * 49) + (((int)threadIdx.x) % 49))] * kernel_shared[((((((int)threadIdx.x) / 49) * 32) + (ff_inner * 16)) + rc_outer_inner)]));
- }
- }
- }
+ extern "C" __global__ void __launch_bounds__(56) default_function_kernel0(float* __restrict__ data, float* __restrict__ kernel, float* __restrict__ compute, float* __restrict__ bias) {
+ float conv2d_nchw[14];
+ __shared__ float pad_temp_shared[2592];
+ __shared__ float kernel_shared[4608];
+ conv2d_nchw[0] = 0.000000e+00f;
+ conv2d_nchw[2] = 0.000000e+00f;
+ conv2d_nchw[4] = 0.000000e+00f;
+ conv2d_nchw[6] = 0.000000e+00f;
+ conv2d_nchw[8] = 0.000000e+00f;
+ conv2d_nchw[10] = 0.000000e+00f;
+ conv2d_nchw[12] = 0.000000e+00f;
+ conv2d_nchw[1] = 0.000000e+00f;
+ conv2d_nchw[3] = 0.000000e+00f;
+ conv2d_nchw[5] = 0.000000e+00f;
+ conv2d_nchw[7] = 0.000000e+00f;
+ conv2d_nchw[9] = 0.000000e+00f;
+ conv2d_nchw[11] = 0.000000e+00f;
+ conv2d_nchw[13] = 0.000000e+00f;
+ for (int rc_outer_outer = 0; rc_outer_outer < 16; ++rc_outer_outer) {
+ __syncthreads();
+ pad_temp_shared[((int)threadIdx.x)] = ((((9 <= ((int)threadIdx.x)) && (1 <= (((int)threadIdx.x) % 9))) && ((((int)threadIdx.x) % 9) < 8)) ? data[((((rc_outer_outer * 1568) + ((((int)threadIdx.x) / 9) * 7)) + (((int)threadIdx.x) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 56)] = (((((9 <= ((((int)threadIdx.x) + 56) % 81)) && (((((int)threadIdx.x) + 56) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 2) % 9))) && (((((int)threadIdx.x) + 2) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 56) / 81) * 49)) + ((((((int)threadIdx.x) + 56) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 2) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 112)] = (((((9 <= ((((int)threadIdx.x) + 31) % 81)) && (((((int)threadIdx.x) + 31) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 4) % 9))) && (((((int)threadIdx.x) + 4) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 112) / 81) * 49)) + ((((((int)threadIdx.x) + 31) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 4) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 168)] = ((((3 <= ((int)threadIdx.x)) && (1 <= ((((int)threadIdx.x) + 6) % 9))) && (((((int)threadIdx.x) + 6) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 168) / 81) * 49)) + (((((int)threadIdx.x) + 6) / 9) * 7)) + ((((int)threadIdx.x) + 6) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 224)] = (((((9 <= ((((int)threadIdx.x) + 62) % 81)) && (((((int)threadIdx.x) + 62) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 8) % 9))) && (((((int)threadIdx.x) + 8) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 224) / 81) * 49)) + ((((((int)threadIdx.x) + 62) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 8) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 280)] = (((((9 <= ((((int)threadIdx.x) + 37) % 81)) && (((((int)threadIdx.x) + 37) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 1) % 9))) && (((((int)threadIdx.x) + 1) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 280) / 81) * 49)) + ((((((int)threadIdx.x) + 37) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 1) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 336)] = (((1 <= ((((int)threadIdx.x) + 3) % 9)) && (((((int)threadIdx.x) + 3) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 336) / 81) * 49)) + (((((int)threadIdx.x) + 12) / 9) * 7)) + ((((int)threadIdx.x) + 3) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 392)] = (((((9 <= ((((int)threadIdx.x) + 68) % 81)) && (((((int)threadIdx.x) + 68) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 5) % 9))) && (((((int)threadIdx.x) + 5) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 392) / 81) * 49)) + ((((((int)threadIdx.x) + 68) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 5) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 448)] = (((((9 <= ((((int)threadIdx.x) + 43) % 81)) && (((((int)threadIdx.x) + 43) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 7) % 9))) && (((((int)threadIdx.x) + 7) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 448) / 81) * 49)) + ((((((int)threadIdx.x) + 43) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 7) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 504)] = ((((((int)threadIdx.x) < 54) && (1 <= (((int)threadIdx.x) % 9))) && ((((int)threadIdx.x) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 504) / 81) * 49)) + ((((int)threadIdx.x) / 9) * 7)) + (((int)threadIdx.x) % 9)) + 6)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 560)] = (((((9 <= ((((int)threadIdx.x) + 74) % 81)) && (((((int)threadIdx.x) + 74) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 2) % 9))) && (((((int)threadIdx.x) + 2) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 560) / 81) * 49)) + ((((((int)threadIdx.x) + 74) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 2) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 616)] = (((((9 <= ((((int)threadIdx.x) + 49) % 81)) && (((((int)threadIdx.x) + 49) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 4) % 9))) && (((((int)threadIdx.x) + 4) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 616) / 81) * 49)) + ((((((int)threadIdx.x) + 49) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 4) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 672)] = ((((((int)threadIdx.x) < 48) && (1 <= ((((int)threadIdx.x) + 6) % 9))) && (((((int)threadIdx.x) + 6) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 672) / 81) * 49)) + (((((int)threadIdx.x) + 24) / 9) * 7)) + ((((int)threadIdx.x) + 6) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 728)] = (((((9 <= ((((int)threadIdx.x) + 80) % 81)) && (((((int)threadIdx.x) + 80) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 8) % 9))) && (((((int)threadIdx.x) + 8) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 728) / 81) * 49)) + ((((((int)threadIdx.x) + 80) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 8) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 784)] = (((((9 <= ((((int)threadIdx.x) + 55) % 81)) && (((((int)threadIdx.x) + 55) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 1) % 9))) && (((((int)threadIdx.x) + 1) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 784) / 81) * 49)) + ((((((int)threadIdx.x) + 55) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 1) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 840)] = (((((9 <= ((((int)threadIdx.x) + 30) % 81)) && (((((int)threadIdx.x) + 30) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 3) % 9))) && (((((int)threadIdx.x) + 3) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 840) / 81) * 49)) + ((((((int)threadIdx.x) + 30) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 3) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 896)] = ((((4 <= ((int)threadIdx.x)) && (1 <= ((((int)threadIdx.x) + 5) % 9))) && (((((int)threadIdx.x) + 5) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 896) / 81) * 49)) + (((((int)threadIdx.x) + 5) / 9) * 7)) + ((((int)threadIdx.x) + 5) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 952)] = (((((9 <= ((((int)threadIdx.x) + 61) % 81)) && (((((int)threadIdx.x) + 61) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 7) % 9))) && (((((int)threadIdx.x) + 7) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 952) / 81) * 49)) + ((((((int)threadIdx.x) + 61) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 7) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 1008)] = (((((1 <= (((((int)threadIdx.x) / 9) + 4) % 9)) && (((((int)threadIdx.x) + 36) % 81) < 72)) && (1 <= (((int)threadIdx.x) % 9))) && ((((int)threadIdx.x) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 1008) / 81) * 49)) + ((((((int)threadIdx.x) / 9) + 4) % 9) * 7)) + (((int)threadIdx.x) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 1064)] = (((1 <= ((((int)threadIdx.x) + 2) % 9)) && (((((int)threadIdx.x) + 2) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 1064) / 81) * 49)) + (((((int)threadIdx.x) + 11) / 9) * 7)) + ((((int)threadIdx.x) + 2) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 1120)] = (((((9 <= ((((int)threadIdx.x) + 67) % 81)) && (((((int)threadIdx.x) + 67) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 4) % 9))) && (((((int)threadIdx.x) + 4) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 1120) / 81) * 49)) + ((((((int)threadIdx.x) + 67) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 4) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 1176)] = (((((9 <= ((((int)threadIdx.x) + 42) % 81)) && (((((int)threadIdx.x) + 42) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 6) % 9))) && (((((int)threadIdx.x) + 6) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 1176) / 81) * 49)) + ((((((int)threadIdx.x) + 42) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 6) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 1232)] = ((((((int)threadIdx.x) < 55) && (1 <= ((((int)threadIdx.x) + 8) % 9))) && (((((int)threadIdx.x) + 8) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 1232) / 81) * 49)) + (((((int)threadIdx.x) + 17) / 9) * 7)) + ((((int)threadIdx.x) + 8) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 1288)] = (((((9 <= ((((int)threadIdx.x) + 73) % 81)) && (((((int)threadIdx.x) + 73) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 1) % 9))) && (((((int)threadIdx.x) + 1) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 1288) / 81) * 49)) + ((((((int)threadIdx.x) + 73) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 1) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 1344)] = (((((9 <= ((((int)threadIdx.x) + 48) % 81)) && (((((int)threadIdx.x) + 48) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 3) % 9))) && (((((int)threadIdx.x) + 3) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 1344) / 81) * 49)) + ((((((int)threadIdx.x) + 48) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 3) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 1400)] = ((((((int)threadIdx.x) < 49) && (1 <= ((((int)threadIdx.x) + 5) % 9))) && (((((int)threadIdx.x) + 5) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 1400) / 81) * 49)) + (((((int)threadIdx.x) + 23) / 9) * 7)) + ((((int)threadIdx.x) + 5) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 1456)] = (((((9 <= ((((int)threadIdx.x) + 79) % 81)) && (((((int)threadIdx.x) + 79) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 7) % 9))) && (((((int)threadIdx.x) + 7) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 1456) / 81) * 49)) + ((((((int)threadIdx.x) + 79) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 7) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 1512)] = (((((1 <= (((((int)threadIdx.x) / 9) + 6) % 9)) && (((((int)threadIdx.x) + 54) % 81) < 72)) && (1 <= (((int)threadIdx.x) % 9))) && ((((int)threadIdx.x) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 1512) / 81) * 49)) + ((((((int)threadIdx.x) / 9) + 6) % 9) * 7)) + (((int)threadIdx.x) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 1568)] = (((((9 <= ((((int)threadIdx.x) + 29) % 81)) && (((((int)threadIdx.x) + 29) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 2) % 9))) && (((((int)threadIdx.x) + 2) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 1568) / 81) * 49)) + ((((((int)threadIdx.x) + 29) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 2) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 1624)] = ((((5 <= ((int)threadIdx.x)) && (1 <= ((((int)threadIdx.x) + 4) % 9))) && (((((int)threadIdx.x) + 4) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 1624) / 81) * 49)) + (((((int)threadIdx.x) + 4) / 9) * 7)) + ((((int)threadIdx.x) + 4) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 1680)] = (((((9 <= ((((int)threadIdx.x) + 60) % 81)) && (((((int)threadIdx.x) + 60) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 6) % 9))) && (((((int)threadIdx.x) + 6) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 1680) / 81) * 49)) + ((((((int)threadIdx.x) + 60) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 6) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 1736)] = (((((9 <= ((((int)threadIdx.x) + 35) % 81)) && (((((int)threadIdx.x) + 35) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 8) % 9))) && (((((int)threadIdx.x) + 8) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 1736) / 81) * 49)) + ((((((int)threadIdx.x) + 35) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 8) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 1792)] = (((1 <= ((((int)threadIdx.x) + 1) % 9)) && (((((int)threadIdx.x) + 1) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 1792) / 81) * 49)) + (((((int)threadIdx.x) + 10) / 9) * 7)) + ((((int)threadIdx.x) + 1) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 1848)] = (((((9 <= ((((int)threadIdx.x) + 66) % 81)) && (((((int)threadIdx.x) + 66) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 3) % 9))) && (((((int)threadIdx.x) + 3) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 1848) / 81) * 49)) + ((((((int)threadIdx.x) + 66) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 3) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 1904)] = (((((9 <= ((((int)threadIdx.x) + 41) % 81)) && (((((int)threadIdx.x) + 41) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 5) % 9))) && (((((int)threadIdx.x) + 5) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 1904) / 81) * 49)) + ((((((int)threadIdx.x) + 41) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 5) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 1960)] = (((1 <= ((((int)threadIdx.x) + 7) % 9)) && (((((int)threadIdx.x) + 7) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 1960) / 81) * 49)) + (((((int)threadIdx.x) + 16) / 9) * 7)) + ((((int)threadIdx.x) + 7) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 2016)] = (((((1 <= (((((int)threadIdx.x) / 9) + 8) % 9)) && (((((int)threadIdx.x) + 72) % 81) < 72)) && (1 <= (((int)threadIdx.x) % 9))) && ((((int)threadIdx.x) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 2016) / 81) * 49)) + ((((((int)threadIdx.x) / 9) + 8) % 9) * 7)) + (((int)threadIdx.x) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 2072)] = (((((9 <= ((((int)threadIdx.x) + 47) % 81)) && (((((int)threadIdx.x) + 47) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 2) % 9))) && (((((int)threadIdx.x) + 2) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 2072) / 81) * 49)) + ((((((int)threadIdx.x) + 47) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 2) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 2128)] = ((((((int)threadIdx.x) < 50) && (1 <= ((((int)threadIdx.x) + 4) % 9))) && (((((int)threadIdx.x) + 4) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 2128) / 81) * 49)) + (((((int)threadIdx.x) + 22) / 9) * 7)) + ((((int)threadIdx.x) + 4) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 2184)] = (((((9 <= ((((int)threadIdx.x) + 78) % 81)) && (((((int)threadIdx.x) + 78) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 6) % 9))) && (((((int)threadIdx.x) + 6) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 2184) / 81) * 49)) + ((((((int)threadIdx.x) + 78) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 6) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 2240)] = (((((9 <= ((((int)threadIdx.x) + 53) % 81)) && (((((int)threadIdx.x) + 53) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 8) % 9))) && (((((int)threadIdx.x) + 8) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 2240) / 81) * 49)) + ((((((int)threadIdx.x) + 53) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 8) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 2296)] = (((((9 <= ((((int)threadIdx.x) + 28) % 81)) && (((((int)threadIdx.x) + 28) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 1) % 9))) && (((((int)threadIdx.x) + 1) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 2296) / 81) * 49)) + ((((((int)threadIdx.x) + 28) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 1) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 2352)] = ((((6 <= ((int)threadIdx.x)) && (1 <= ((((int)threadIdx.x) + 3) % 9))) && (((((int)threadIdx.x) + 3) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 2352) / 81) * 49)) + (((((int)threadIdx.x) + 3) / 9) * 7)) + ((((int)threadIdx.x) + 3) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 2408)] = (((((9 <= ((((int)threadIdx.x) + 59) % 81)) && (((((int)threadIdx.x) + 59) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 5) % 9))) && (((((int)threadIdx.x) + 5) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 2408) / 81) * 49)) + ((((((int)threadIdx.x) + 59) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 5) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 2464)] = (((((9 <= ((((int)threadIdx.x) + 34) % 81)) && (((((int)threadIdx.x) + 34) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 7) % 9))) && (((((int)threadIdx.x) + 7) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 2464) / 81) * 49)) + ((((((int)threadIdx.x) + 34) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 7) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 2520)] = (((1 <= (((int)threadIdx.x) % 9)) && ((((int)threadIdx.x) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 2520) / 81) * 49)) + ((((int)threadIdx.x) / 9) * 7)) + (((int)threadIdx.x) % 9)) - 1)] : 0.000000e+00f);
+ if (((int)threadIdx.x) < 16) {
+ pad_temp_shared[(((int)threadIdx.x) + 2576)] = ((((((int)threadIdx.x) < 7) && (1 <= ((((int)threadIdx.x) + 2) % 9))) && (((((int)threadIdx.x) + 2) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 2576) / 81) * 49)) + (((((int)threadIdx.x) + 65) / 9) * 7)) + ((int)threadIdx.x)) - 6)] : 0.000000e+00f);
+ }
+ kernel_shared[((int)threadIdx.x)] = kernel[(((((int)blockIdx.x) * 73728) + (rc_outer_outer * 288)) + ((int)threadIdx.x))];
+ kernel_shared[(((int)threadIdx.x) + 56)] = kernel[((((((int)blockIdx.x) * 73728) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 56)];
+ kernel_shared[(((int)threadIdx.x) + 112)] = kernel[((((((int)blockIdx.x) * 73728) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 112)];
+ kernel_shared[(((int)threadIdx.x) + 168)] = kernel[((((((int)blockIdx.x) * 73728) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 168)];
+ kernel_shared[(((int)threadIdx.x) + 224)] = kernel[((((((int)blockIdx.x) * 73728) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 224)];
+ kernel_shared[(((int)threadIdx.x) + 280)] = kernel[((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 280) / 288) * 4608)) + (rc_outer_outer * 288)) + ((((int)threadIdx.x) + 280) % 288))];
+ kernel_shared[(((int)threadIdx.x) + 336)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 336) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 48)];
+ kernel_shared[(((int)threadIdx.x) + 392)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 392) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 104)];
+ kernel_shared[(((int)threadIdx.x) + 448)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 448) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 160)];
+ kernel_shared[(((int)threadIdx.x) + 504)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 504) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 216)];
+ kernel_shared[(((int)threadIdx.x) + 560)] = kernel[((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 560) / 288) * 4608)) + (rc_outer_outer * 288)) + ((((int)threadIdx.x) + 272) % 288))];
+ kernel_shared[(((int)threadIdx.x) + 616)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 616) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 40)];
+ kernel_shared[(((int)threadIdx.x) + 672)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 672) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 96)];
+ kernel_shared[(((int)threadIdx.x) + 728)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 728) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 152)];
+ kernel_shared[(((int)threadIdx.x) + 784)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 784) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 208)];
+ kernel_shared[(((int)threadIdx.x) + 840)] = kernel[((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 840) / 288) * 4608)) + (rc_outer_outer * 288)) + ((((int)threadIdx.x) + 264) % 288))];
+ kernel_shared[(((int)threadIdx.x) + 896)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 896) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 32)];
+ kernel_shared[(((int)threadIdx.x) + 952)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 952) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 88)];
+ kernel_shared[(((int)threadIdx.x) + 1008)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1008) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 144)];
+ kernel_shared[(((int)threadIdx.x) + 1064)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1064) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 200)];
+ kernel_shared[(((int)threadIdx.x) + 1120)] = kernel[((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1120) / 288) * 4608)) + (rc_outer_outer * 288)) + ((((int)threadIdx.x) + 256) % 288))];
+ kernel_shared[(((int)threadIdx.x) + 1176)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1176) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 24)];
+ kernel_shared[(((int)threadIdx.x) + 1232)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1232) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 80)];
+ kernel_shared[(((int)threadIdx.x) + 1288)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1288) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 136)];
+ kernel_shared[(((int)threadIdx.x) + 1344)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1344) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 192)];
+ kernel_shared[(((int)threadIdx.x) + 1400)] = kernel[((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1400) / 288) * 4608)) + (rc_outer_outer * 288)) + ((((int)threadIdx.x) + 248) % 288))];
+ kernel_shared[(((int)threadIdx.x) + 1456)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1456) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 16)];
+ kernel_shared[(((int)threadIdx.x) + 1512)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1512) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 72)];
+ kernel_shared[(((int)threadIdx.x) + 1568)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1568) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 128)];
+ kernel_shared[(((int)threadIdx.x) + 1624)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1624) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 184)];
+ kernel_shared[(((int)threadIdx.x) + 1680)] = kernel[((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1680) / 288) * 4608)) + (rc_outer_outer * 288)) + ((((int)threadIdx.x) + 240) % 288))];
+ kernel_shared[(((int)threadIdx.x) + 1736)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1736) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 8)];
+ kernel_shared[(((int)threadIdx.x) + 1792)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1792) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 64)];
+ kernel_shared[(((int)threadIdx.x) + 1848)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1848) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 120)];
+ kernel_shared[(((int)threadIdx.x) + 1904)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1904) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 176)];
+ kernel_shared[(((int)threadIdx.x) + 1960)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1960) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 232)];
+ kernel_shared[(((int)threadIdx.x) + 2016)] = kernel[((((((int)blockIdx.x) * 73728) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 32256)];
+ kernel_shared[(((int)threadIdx.x) + 2072)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2072) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 56)];
+ kernel_shared[(((int)threadIdx.x) + 2128)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2128) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 112)];
+ kernel_shared[(((int)threadIdx.x) + 2184)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2184) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 168)];
+ kernel_shared[(((int)threadIdx.x) + 2240)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2240) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 224)];
+ kernel_shared[(((int)threadIdx.x) + 2296)] = kernel[((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2296) / 288) * 4608)) + (rc_outer_outer * 288)) + ((((int)threadIdx.x) + 280) % 288))];
+ kernel_shared[(((int)threadIdx.x) + 2352)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2352) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 48)];
+ kernel_shared[(((int)threadIdx.x) + 2408)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2408) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 104)];
+ kernel_shared[(((int)threadIdx.x) + 2464)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2464) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 160)];
+ kernel_shared[(((int)threadIdx.x) + 2520)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2520) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 216)];
+ kernel_shared[(((int)threadIdx.x) + 2576)] = kernel[((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2576) / 288) * 4608)) + (rc_outer_outer * 288)) + ((((int)threadIdx.x) + 272) % 288))];
+ kernel_shared[(((int)threadIdx.x) + 2632)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2632) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 40)];
+ kernel_shared[(((int)threadIdx.x) + 2688)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2688) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 96)];
+ kernel_shared[(((int)threadIdx.x) + 2744)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2744) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 152)];
+ kernel_shared[(((int)threadIdx.x) + 2800)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2800) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 208)];
+ kernel_shared[(((int)threadIdx.x) + 2856)] = kernel[((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2856) / 288) * 4608)) + (rc_outer_outer * 288)) + ((((int)threadIdx.x) + 264) % 288))];
+ kernel_shared[(((int)threadIdx.x) + 2912)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2912) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 32)];
+ kernel_shared[(((int)threadIdx.x) + 2968)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2968) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 88)];
+ kernel_shared[(((int)threadIdx.x) + 3024)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3024) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 144)];
+ kernel_shared[(((int)threadIdx.x) + 3080)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3080) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 200)];
+ kernel_shared[(((int)threadIdx.x) + 3136)] = kernel[((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3136) / 288) * 4608)) + (rc_outer_outer * 288)) + ((((int)threadIdx.x) + 256) % 288))];
+ kernel_shared[(((int)threadIdx.x) + 3192)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3192) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 24)];
+ kernel_shared[(((int)threadIdx.x) + 3248)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3248) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 80)];
+ kernel_shared[(((int)threadIdx.x) + 3304)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3304) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 136)];
+ kernel_shared[(((int)threadIdx.x) + 3360)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3360) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 192)];
+ kernel_shared[(((int)threadIdx.x) + 3416)] = kernel[((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3416) / 288) * 4608)) + (rc_outer_outer * 288)) + ((((int)threadIdx.x) + 248) % 288))];
+ kernel_shared[(((int)threadIdx.x) + 3472)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3472) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 16)];
+ kernel_shared[(((int)threadIdx.x) + 3528)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3528) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 72)];
+ kernel_shared[(((int)threadIdx.x) + 3584)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3584) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 128)];
+ kernel_shared[(((int)threadIdx.x) + 3640)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3640) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 184)];
+ kernel_shared[(((int)threadIdx.x) + 3696)] = kernel[((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3696) / 288) * 4608)) + (rc_outer_outer * 288)) + ((((int)threadIdx.x) + 240) % 288))];
+ kernel_shared[(((int)threadIdx.x) + 3752)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3752) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 8)];
+ kernel_shared[(((int)threadIdx.x) + 3808)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3808) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 64)];
+ kernel_shared[(((int)threadIdx.x) + 3864)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3864) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 120)];
+ kernel_shared[(((int)threadIdx.x) + 3920)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3920) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 176)];
+ kernel_shared[(((int)threadIdx.x) + 3976)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3976) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 232)];
+ kernel_shared[(((int)threadIdx.x) + 4032)] = kernel[((((((int)blockIdx.x) * 73728) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 64512)];
+ kernel_shared[(((int)threadIdx.x) + 4088)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 4088) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 56)];
+ kernel_shared[(((int)threadIdx.x) + 4144)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 4144) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 112)];
+ kernel_shared[(((int)threadIdx.x) + 4200)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 4200) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 168)];
+ kernel_shared[(((int)threadIdx.x) + 4256)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 4256) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 224)];
+ kernel_shared[(((int)threadIdx.x) + 4312)] = kernel[((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 4312) / 288) * 4608)) + (rc_outer_outer * 288)) + ((((int)threadIdx.x) + 280) % 288))];
+ kernel_shared[(((int)threadIdx.x) + 4368)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 4368) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 48)];
+ kernel_shared[(((int)threadIdx.x) + 4424)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 4424) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 104)];
+ kernel_shared[(((int)threadIdx.x) + 4480)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 4480) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 160)];
+ kernel_shared[(((int)threadIdx.x) + 4536)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 4536) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 216)];
+ if (((int)threadIdx.x) < 16) {
+ kernel_shared[(((int)threadIdx.x) + 4592)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 4592) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 272)];
+ }
+ __syncthreads();
+ for (int rc_outer_inner = 0; rc_outer_inner < 4; ++rc_outer_inner) {
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((rc_outer_inner * 648) + (((int)threadIdx.x) % 7))] * kernel_shared[(((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72))]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[(((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72))]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[(((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72))]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[(((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72))]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[(((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72))]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[(((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72))]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[(((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72))]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((rc_outer_inner * 648) + (((int)threadIdx.x) % 7))] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 288)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 288)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 288)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 288)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 288)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 288)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 288)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 1)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 1)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 1)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 28)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 1)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 37)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 1)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 46)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 1)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 1)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 289)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 289)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 289)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 28)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 289)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 37)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 289)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 46)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 289)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 289)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 2)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 2)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 2)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 29)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 2)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 38)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 2)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 47)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 2)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 56)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 2)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 290)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 290)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 290)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 29)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 290)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 38)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 290)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 47)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 290)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 56)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 290)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 81)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 9)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 9)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 9)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 9)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 9)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 9)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 9)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 81)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 297)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 297)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 297)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 297)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 297)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 297)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 297)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 82)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 10)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 10)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 10)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 109)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 10)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 118)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 10)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 10)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 136)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 10)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 82)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 298)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 298)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 298)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 109)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 298)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 118)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 298)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 298)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 136)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 298)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 83)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 11)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 11)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 11)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 110)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 11)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 119)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 11)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 11)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 137)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 11)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 83)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 299)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 299)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 299)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 110)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 299)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 119)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 299)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 299)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 137)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 299)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 162)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 18)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 18)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 18)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 18)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 18)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 18)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 18)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 162)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 306)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 306)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 306)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 306)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 306)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 306)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 306)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 163)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 19)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 19)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 19)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 19)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 199)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 19)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 208)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 19)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 217)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 19)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 163)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 307)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 307)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 307)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 307)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 199)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 307)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 208)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 307)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 217)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 307)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 164)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 20)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 20)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 20)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 20)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 200)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 20)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 209)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 20)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 218)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 20)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 164)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 308)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 308)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 308)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 308)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 200)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 308)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 209)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 308)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 218)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 308)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 243)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 27)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 27)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 27)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 27)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 27)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 27)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 27)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 243)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 315)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 315)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 315)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 315)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 315)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 315)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 315)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 244)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 28)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 28)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 28)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 271)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 28)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 280)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 28)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 289)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 28)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 298)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 28)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 244)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 316)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 316)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 316)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 271)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 316)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 280)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 316)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 289)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 316)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 298)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 316)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 245)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 29)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 29)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 29)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 272)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 29)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 281)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 29)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 290)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 29)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 299)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 29)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 245)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 317)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 317)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 317)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 272)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 317)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 281)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 317)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 290)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 317)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 299)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 317)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 324)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 36)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 333)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 36)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 342)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 36)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 351)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 36)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 360)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 36)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 369)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 36)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 36)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 324)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 324)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 333)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 324)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 342)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 324)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 351)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 324)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 360)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 324)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 369)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 324)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 324)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 325)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 37)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 334)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 37)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 343)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 37)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 352)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 37)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 361)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 37)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 370)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 37)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 379)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 37)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 325)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 325)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 334)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 325)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 343)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 325)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 352)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 325)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 361)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 325)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 370)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 325)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 379)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 325)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 326)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 38)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 335)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 38)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 344)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 38)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 353)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 38)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 362)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 38)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 371)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 38)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 380)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 38)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 326)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 326)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 335)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 326)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 344)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 326)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 353)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 326)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 362)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 326)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 371)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 326)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 380)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 326)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 405)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 45)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 414)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 45)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 423)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 45)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 432)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 45)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 45)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 450)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 45)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 459)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 45)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 405)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 333)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 414)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 333)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 423)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 333)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 432)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 333)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 333)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 450)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 333)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 459)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 333)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 406)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 46)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 415)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 46)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 424)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 46)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 433)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 46)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 442)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 46)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 451)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 46)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 460)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 46)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 406)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 334)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 415)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 334)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 424)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 334)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 433)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 334)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 442)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 334)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 451)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 334)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 460)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 334)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 407)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 47)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 416)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 47)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 425)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 47)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 434)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 47)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 443)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 47)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 452)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 47)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 461)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 47)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 407)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 335)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 416)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 335)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 425)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 335)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 434)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 335)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 443)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 335)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 452)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 335)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 461)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 335)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 486)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 54)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 495)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 54)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 504)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 54)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 513)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 54)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 522)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 54)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 531)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 54)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 540)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 54)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 486)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 342)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 495)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 342)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 504)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 342)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 513)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 342)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 522)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 342)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 531)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 342)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 540)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 342)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 487)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 55)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 496)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 55)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 505)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 55)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 514)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 55)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 523)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 55)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 532)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 55)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 541)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 55)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 487)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 343)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 496)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 343)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 505)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 343)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 514)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 343)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 523)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 343)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 532)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 343)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 541)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 343)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 488)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 56)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 497)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 56)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 506)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 56)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 515)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 56)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 524)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 56)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 533)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 56)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 542)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 56)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 488)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 344)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 497)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 344)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 506)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 344)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 515)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 344)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 524)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 344)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 533)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 344)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 542)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 344)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 567)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 63)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 576)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 63)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 585)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 63)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 594)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 63)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 603)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 63)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 612)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 63)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 621)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 63)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 567)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 351)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 576)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 351)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 585)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 351)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 594)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 351)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 603)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 351)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 612)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 351)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 621)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 351)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 568)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 64)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 577)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 64)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 586)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 64)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 595)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 64)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 604)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 64)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 613)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 64)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 622)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 64)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 568)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 352)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 577)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 352)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 586)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 352)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 595)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 352)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 604)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 352)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 613)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 352)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 622)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 352)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 569)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 65)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 578)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 65)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 587)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 65)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 596)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 65)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 605)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 65)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 614)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 65)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 623)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 65)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 569)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 353)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 578)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 353)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 587)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 353)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 596)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 353)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 605)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 353)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 614)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 353)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 623)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 353)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 3)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 3)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 3)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 3)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 3)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 3)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 3)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 291)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 291)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 291)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 291)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 291)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 291)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 291)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 4)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 4)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 28)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 4)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 37)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 4)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 46)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 4)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 4)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 4)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 292)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 292)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 28)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 292)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 37)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 292)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 46)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 292)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 292)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 292)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 5)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 5)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 29)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 5)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 38)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 5)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 47)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 5)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 56)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 5)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 5)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 293)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 293)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 29)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 293)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 38)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 293)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 47)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 293)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 56)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 293)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 293)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 12)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 12)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 12)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 12)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 12)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 12)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 144)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 12)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 300)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 300)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 300)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 300)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 300)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 300)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 144)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 300)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 13)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 13)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 109)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 13)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 118)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 13)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 13)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 136)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 13)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 145)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 13)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 301)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 301)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 109)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 301)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 118)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 301)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 301)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 136)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 301)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 145)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 301)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 14)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 14)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 110)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 14)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 119)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 14)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 14)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 137)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 14)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 146)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 14)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 302)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 302)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 110)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 302)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 119)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 302)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 302)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 137)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 302)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 146)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 302)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 21)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 21)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 21)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 21)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 21)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 21)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 225)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 21)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 309)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 309)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 309)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 309)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 309)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 309)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 225)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 309)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 22)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 22)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 22)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 199)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 22)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 208)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 22)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 217)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 22)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 226)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 22)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 310)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 310)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 310)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 199)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 310)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 208)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 310)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 217)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 310)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 226)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 310)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 23)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 23)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 23)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 200)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 23)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 209)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 23)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 218)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 23)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 227)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 23)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 311)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 311)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 311)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 200)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 311)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 209)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 311)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 218)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 311)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 227)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 311)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 30)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 30)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 30)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 30)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 30)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 30)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 306)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 30)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 318)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 318)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 318)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 318)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 318)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 318)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 306)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 318)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 31)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 31)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 271)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 31)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 280)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 31)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 289)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 31)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 298)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 31)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 307)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 31)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 319)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 319)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 271)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 319)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 280)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 319)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 289)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 319)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 298)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 319)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 307)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 319)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 32)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 32)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 272)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 32)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 281)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 32)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 290)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 32)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 299)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 32)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 308)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 32)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 320)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 320)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 272)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 320)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 281)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 320)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 290)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 320)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 299)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 320)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 308)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 320)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 333)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 39)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 342)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 39)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 351)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 39)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 360)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 39)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 369)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 39)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 39)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 387)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 39)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 333)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 327)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 342)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 327)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 351)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 327)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 360)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 327)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 369)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 327)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 327)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 387)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 327)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 334)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 40)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 343)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 40)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 352)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 40)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 361)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 40)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 370)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 40)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 379)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 40)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 388)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 40)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 334)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 328)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 343)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 328)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 352)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 328)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 361)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 328)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 370)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 328)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 379)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 328)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 388)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 328)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 335)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 41)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 344)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 41)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 353)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 41)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 362)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 41)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 371)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 41)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 380)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 41)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 389)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 41)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 335)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 329)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 344)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 329)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 353)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 329)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 362)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 329)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 371)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 329)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 380)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 329)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 389)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 329)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 414)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 48)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 423)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 48)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 432)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 48)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 48)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 450)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 48)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 459)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 48)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 468)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 48)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 414)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 336)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 423)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 336)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 432)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 336)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 336)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 450)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 336)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 459)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 336)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 468)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 336)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 415)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 49)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 424)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 49)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 433)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 49)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 442)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 49)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 451)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 49)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 460)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 49)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 469)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 49)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 415)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 337)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 424)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 337)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 433)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 337)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 442)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 337)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 451)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 337)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 460)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 337)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 469)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 337)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 416)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 50)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 425)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 50)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 434)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 50)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 443)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 50)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 452)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 50)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 461)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 50)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 470)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 50)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 416)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 338)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 425)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 338)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 434)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 338)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 443)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 338)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 452)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 338)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 461)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 338)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 470)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 338)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 495)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 57)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 504)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 57)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 513)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 57)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 522)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 57)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 531)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 57)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 540)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 57)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 549)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 57)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 495)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 345)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 504)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 345)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 513)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 345)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 522)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 345)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 531)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 345)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 540)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 345)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 549)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 345)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 496)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 58)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 505)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 58)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 514)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 58)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 523)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 58)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 532)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 58)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 541)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 58)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 550)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 58)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 496)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 346)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 505)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 346)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 514)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 346)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 523)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 346)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 532)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 346)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 541)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 346)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 550)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 346)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 497)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 59)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 506)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 59)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 515)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 59)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 524)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 59)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 533)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 59)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 542)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 59)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 551)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 59)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 497)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 347)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 506)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 347)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 515)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 347)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 524)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 347)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 533)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 347)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 542)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 347)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 551)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 347)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 576)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 66)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 585)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 66)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 594)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 66)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 603)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 66)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 612)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 66)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 621)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 66)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 630)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 66)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 576)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 354)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 585)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 354)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 594)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 354)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 603)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 354)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 612)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 354)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 621)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 354)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 630)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 354)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 577)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 67)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 586)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 67)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 595)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 67)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 604)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 67)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 613)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 67)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 622)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 67)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 631)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 67)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 577)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 355)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 586)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 355)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 595)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 355)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 604)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 355)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 613)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 355)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 622)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 355)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 631)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 355)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 578)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 68)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 587)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 68)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 596)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 68)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 605)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 68)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 614)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 68)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 623)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 68)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 632)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 68)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 578)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 356)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 587)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 356)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 596)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 356)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 605)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 356)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 614)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 356)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 623)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 356)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 632)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 356)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 6)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 6)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 6)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 6)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 6)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 6)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 72)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 6)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 294)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 294)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 294)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 294)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 294)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 294)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 72)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 294)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 7)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 28)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 7)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 37)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 7)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 46)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 7)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 7)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 7)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 73)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 7)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 295)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 28)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 295)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 37)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 295)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 46)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 295)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 295)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 295)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 73)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 295)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 8)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 29)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 8)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 38)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 8)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 47)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 8)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 56)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 8)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 8)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 74)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 8)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 296)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 29)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 296)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 38)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 296)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 47)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 296)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 56)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 296)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 296)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 74)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 296)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 15)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 15)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 15)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 15)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 15)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 144)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 15)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 153)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 15)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 303)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 303)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 303)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 303)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 303)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 144)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 303)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 153)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 303)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 16)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 109)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 16)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 118)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 16)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 16)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 136)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 16)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 145)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 16)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 154)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 16)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 304)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 109)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 304)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 118)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 304)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 304)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 136)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 304)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 145)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 304)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 154)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 304)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 17)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 110)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 17)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 119)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 17)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 17)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 137)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 17)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 146)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 17)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 155)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 17)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 305)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 110)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 305)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 119)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 305)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 305)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 137)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 305)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 146)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 305)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 155)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 305)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 24)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 24)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 24)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 24)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 24)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 225)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 24)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 234)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 24)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 312)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 312)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 312)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 312)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 312)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 225)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 312)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 234)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 312)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 25)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 25)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 199)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 25)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 208)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 25)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 217)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 25)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 226)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 25)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 235)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 25)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 313)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 313)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 199)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 313)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 208)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 313)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 217)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 313)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 226)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 313)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 235)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 313)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 26)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 26)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 200)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 26)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 209)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 26)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 218)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 26)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 227)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 26)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 236)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 26)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 314)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 314)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 200)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 314)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 209)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 314)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 218)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 314)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 227)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 314)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 236)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 314)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 33)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 33)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 33)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 33)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 33)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 306)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 33)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 33)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 321)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 321)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 321)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 321)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 321)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 306)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 321)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 321)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 34)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 271)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 34)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 280)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 34)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 289)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 34)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 298)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 34)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 307)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 34)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 316)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 34)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 322)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 271)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 322)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 280)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 322)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 289)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 322)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 298)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 322)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 307)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 322)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 316)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 322)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 35)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 272)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 35)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 281)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 35)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 290)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 35)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 299)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 35)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 308)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 35)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 317)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 35)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 323)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 272)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 323)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 281)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 323)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 290)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 323)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 299)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 323)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 308)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 323)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 317)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 323)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 342)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 42)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 351)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 42)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 360)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 42)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 369)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 42)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 42)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 387)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 42)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 396)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 42)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 342)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 330)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 351)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 330)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 360)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 330)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 369)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 330)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 330)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 387)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 330)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 396)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 330)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 343)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 43)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 352)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 43)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 361)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 43)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 370)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 43)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 379)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 43)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 388)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 43)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 397)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 43)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 343)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 331)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 352)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 331)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 361)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 331)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 370)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 331)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 379)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 331)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 388)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 331)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 397)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 331)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 344)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 44)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 353)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 44)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 362)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 44)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 371)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 44)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 380)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 44)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 389)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 44)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 398)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 44)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 344)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 332)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 353)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 332)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 362)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 332)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 371)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 332)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 380)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 332)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 389)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 332)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 398)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 332)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 423)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 51)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 432)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 51)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 51)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 450)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 51)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 459)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 51)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 468)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 51)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 477)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 51)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 423)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 339)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 432)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 339)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 339)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 450)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 339)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 459)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 339)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 468)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 339)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 477)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 339)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 424)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 52)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 433)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 52)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 442)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 52)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 451)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 52)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 460)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 52)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 469)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 52)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 478)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 52)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 424)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 340)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 433)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 340)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 442)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 340)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 451)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 340)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 460)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 340)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 469)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 340)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 478)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 340)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 425)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 53)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 434)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 53)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 443)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 53)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 452)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 53)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 461)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 53)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 470)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 53)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 479)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 53)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 425)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 341)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 434)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 341)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 443)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 341)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 452)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 341)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 461)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 341)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 470)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 341)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 479)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 341)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 504)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 60)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 513)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 60)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 522)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 60)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 531)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 60)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 540)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 60)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 549)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 60)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 558)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 60)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 504)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 348)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 513)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 348)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 522)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 348)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 531)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 348)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 540)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 348)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 549)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 348)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 558)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 348)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 505)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 61)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 514)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 61)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 523)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 61)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 532)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 61)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 541)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 61)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 550)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 61)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 559)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 61)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 505)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 349)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 514)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 349)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 523)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 349)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 532)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 349)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 541)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 349)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 550)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 349)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 559)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 349)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 506)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 62)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 515)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 62)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 524)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 62)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 533)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 62)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 542)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 62)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 551)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 62)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 560)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 62)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 506)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 350)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 515)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 350)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 524)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 350)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 533)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 350)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 542)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 350)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 551)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 350)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 560)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 350)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 585)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 69)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 594)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 69)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 603)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 69)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 612)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 69)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 621)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 69)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 630)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 69)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 639)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 69)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 585)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 357)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 594)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 357)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 603)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 357)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 612)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 357)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 621)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 357)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 630)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 357)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 639)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 357)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 586)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 70)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 595)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 70)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 604)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 70)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 613)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 70)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 622)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 70)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 631)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 70)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 640)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 70)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 586)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 358)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 595)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 358)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 604)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 358)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 613)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 358)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 622)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 358)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 631)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 358)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 640)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 358)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 587)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 71)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 596)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 71)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 605)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 71)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 614)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 71)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 623)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 71)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 632)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 71)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 641)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 71)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 587)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 359)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 596)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 359)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 605)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 359)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 614)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 359)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 623)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 359)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 632)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 359)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 641)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 359)]));
}
}
for (int i1_inner = 0; i1_inner < 2; ++i1_inner) {
- compute[((((((int)blockIdx.x) * 1568) + ((((int)threadIdx.x) / 49) * 98)) + (i1_inner * 49)) + (((int)threadIdx.x) % 49))] = max((conv2d_nchw[i1_inner] + bias[(((((int)blockIdx.x) * 32) + ((((int)threadIdx.x) / 49) * 2)) + i1_inner)]), 0.000000e+00f);
+ compute[((((((int)blockIdx.x) * 784) + ((((int)threadIdx.x) / 7) * 98)) + (i1_inner * 49)) + (((int)threadIdx.x) % 7))] = max((conv2d_nchw[i1_inner] + bias[(((((int)blockIdx.x) * 16) + ((((int)threadIdx.x) / 7) * 2)) + i1_inner)]), 0.000000e+00f);
+ compute[(((((((int)blockIdx.x) * 784) + ((((int)threadIdx.x) / 7) * 98)) + (i1_inner * 49)) + (((int)threadIdx.x) % 7)) + 7)] = max((conv2d_nchw[(i1_inner + 2)] + bias[(((((int)blockIdx.x) * 16) + ((((int)threadIdx.x) / 7) * 2)) + i1_inner)]), 0.000000e+00f);
+ compute[(((((((int)blockIdx.x) * 784) + ((((int)threadIdx.x) / 7) * 98)) + (i1_inner * 49)) + (((int)threadIdx.x) % 7)) + 14)] = max((conv2d_nchw[(i1_inner + 4)] + bias[(((((int)blockIdx.x) * 16) + ((((int)threadIdx.x) / 7) * 2)) + i1_inner)]), 0.000000e+00f);
+ compute[(((((((int)blockIdx.x) * 784) + ((((int)threadIdx.x) / 7) * 98)) + (i1_inner * 49)) + (((int)threadIdx.x) % 7)) + 21)] = max((conv2d_nchw[(i1_inner + 6)] + bias[(((((int)blockIdx.x) * 16) + ((((int)threadIdx.x) / 7) * 2)) + i1_inner)]), 0.000000e+00f);
+ compute[(((((((int)blockIdx.x) * 784) + ((((int)threadIdx.x) / 7) * 98)) + (i1_inner * 49)) + (((int)threadIdx.x) % 7)) + 28)] = max((conv2d_nchw[(i1_inner + 8)] + bias[(((((int)blockIdx.x) * 16) + ((((int)threadIdx.x) / 7) * 2)) + i1_inner)]), 0.000000e+00f);
+ compute[(((((((int)blockIdx.x) * 784) + ((((int)threadIdx.x) / 7) * 98)) + (i1_inner * 49)) + (((int)threadIdx.x) % 7)) + 35)] = max((conv2d_nchw[(i1_inner + 10)] + bias[(((((int)blockIdx.x) * 16) + ((((int)threadIdx.x) / 7) * 2)) + i1_inner)]), 0.000000e+00f);
+ compute[(((((((int)blockIdx.x) * 784) + ((((int)threadIdx.x) / 7) * 98)) + (i1_inner * 49)) + (((int)threadIdx.x) % 7)) + 42)] = max((conv2d_nchw[(i1_inner + 12)] + bias[(((((int)blockIdx.x) * 16) + ((((int)threadIdx.x) / 7) * 2)) + i1_inner)]), 0.000000e+00f);
}
}
@@ -529,7 +2957,7 @@ In the example below we resume the status and do more 5 trials.
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 6 minutes 47.248 seconds)
+ **Total running time of the script:** ( 6 minutes 12.870 seconds)
.. _sphx_glr_download_how_to_tune_with_autoscheduler_tune_conv2d_layer_cuda.py:
diff --git a/docs/_sources/how_to/tune_with_autoscheduler/tune_network_cuda.rst.txt b/docs/_sources/how_to/tune_with_autoscheduler/tune_network_cuda.rst.txt
index 7fac8ad392..e1b5c09b2f 100644
--- a/docs/_sources/how_to/tune_with_autoscheduler/tune_network_cuda.rst.txt
+++ b/docs/_sources/how_to/tune_with_autoscheduler/tune_network_cuda.rst.txt
@@ -647,7 +647,7 @@ so we can read the log file and load the best schedules.
Evaluate inference time cost...
Execution time summary:
mean (ms) median (ms) max (ms) min (ms) std (ms)
- 7.9360 7.9369 7.9395 7.9317 0.0032
+ 7.8932 7.8938 7.9052 7.8806 0.0101
@@ -675,7 +675,7 @@ Other Tips
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 11.065 seconds)
+ **Total running time of the script:** ( 1 minutes 6.524 seconds)
.. _sphx_glr_download_how_to_tune_with_autoscheduler_tune_network_cuda.py:
diff --git a/docs/_sources/how_to/tune_with_autoscheduler/tune_network_x86.rst.txt b/docs/_sources/how_to/tune_with_autoscheduler/tune_network_x86.rst.txt
index eb2bfdadf2..b599f43616 100644
--- a/docs/_sources/how_to/tune_with_autoscheduler/tune_network_x86.rst.txt
+++ b/docs/_sources/how_to/tune_with_autoscheduler/tune_network_x86.rst.txt
@@ -666,7 +666,7 @@ so we can read the log file and load the best schedules.
Evaluate inference time cost...
Execution time summary:
mean (ms) median (ms) max (ms) min (ms) std (ms)
- 764.1290 765.7904 765.8056 760.7911 2.3603
+ 746.5936 746.7246 747.6161 745.4401 0.8932
@@ -694,7 +694,7 @@ Other Tips
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 48.279 seconds)
+ **Total running time of the script:** ( 1 minutes 40.793 seconds)
.. _sphx_glr_download_how_to_tune_with_autoscheduler_tune_network_x86.py:
diff --git a/docs/_sources/how_to/tune_with_autoscheduler/tune_sparse_x86.rst.txt b/docs/_sources/how_to/tune_with_autoscheduler/tune_sparse_x86.rst.txt
index 2b2967c98e..01a98d9e86 100644
--- a/docs/_sources/how_to/tune_with_autoscheduler/tune_sparse_x86.rst.txt
+++ b/docs/_sources/how_to/tune_with_autoscheduler/tune_sparse_x86.rst.txt
@@ -389,12 +389,12 @@ layout transformation, parallelization, vectorization, unrolling, and operator f
@T.prim_func
def main(placeholder: T.Buffer((128, 256), "float32"), placeholder_1: T.Buffer((4916, 16, 1), "float32"), placeholder_2: T.Buffer((4916,), "int32"), placeholder_3: T.Buffer((33,), "int32"), placeholder_4: T.Buffer((128, 512), "float32"), compute: T.Buffer((128, 512), "float32")):
T.func_attr({"from_legacy_te_schedule": True, "global_symbol": "main", "tir.noalias": True})
- for i0_outer_i1_outer_fused in T.parallel(32):
- compute_1 = T.allocate([2048], "float32", "global")
- compute_2 = T.Buffer((2048,), data=compute_1)
- for i_outer_inner in range(8):
+ for i0_outer_i1_outer_fused in T.parallel(128):
+ compute_1 = T.allocate([512], "float32", "global")
+ compute_2 = T.Buffer((512,), data=compute_1)
+ for nb_j_inner in range(2):
for i_inner_init in range(16):
- cse_var_1: T.int32 = i_outer_inner * 256 + i_inner_init * 16
+ cse_var_1: T.int32 = i_inner_init * 32 + nb_j_inner * 16
compute_2[cse_var_1] = T.float32(0)
compute_2[cse_var_1 + 1] = T.float32(0)
compute_2[cse_var_1 + 2] = T.float32(0)
@@ -411,64 +411,52 @@ layout transformation, parallelization, vectorization, unrolling, and operator f
compute_2[cse_var_1 + 13] = T.float32(0)
compute_2[cse_var_1 + 14] = T.float32(0)
compute_2[cse_var_1 + 15] = T.float32(0)
- for elem_idx, i_inner in T.grid(placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused], 16):
+ for elem_idx, i_inner in T.grid(T.Let(placeholder_5[cse_var_2 + 1] - placeholder_5[cse_var_2], where={cse_var_2: i0_outer_i1_outer_fused % 16 * 2 + nb_j_inner}), 16):
+ cse_var_2 = T.int32()
placeholder_5 = T.Buffer((33,), "int32", data=placeholder_3.data)
+ cse_var_21: T.int32 = elem_idx * 16
+ cse_var_20: T.int32 = i_inner * 32 + nb_j_inner * 16
+ cse_var_19: T.int32 = i0_outer_i1_outer_fused % 16 * 2 + nb_j_inner
+ cse_var_18: T.int32 = i0_outer_i1_outer_fused // 16 * 4096 + i_inner * 256
+ cse_var_17: T.int32 = cse_var_20 + 9
+ cse_var_16: T.int32 = cse_var_20 + 8
+ cse_var_15: T.int32 = cse_var_20 + 7
+ cse_var_14: T.int32 = cse_var_20 + 6
+ cse_var_13: T.int32 = cse_var_20 + 5
+ cse_var_12: T.int32 = cse_var_20 + 4
+ cse_var_11: T.int32 = cse_var_20 + 3
+ cse_var_10: T.int32 = cse_var_20 + 2
+ cse_var_9: T.int32 = cse_var_20 + 15
+ cse_var_8: T.int32 = cse_var_20 + 14
+ cse_var_7: T.int32 = cse_var_20 + 13
+ cse_var_6: T.int32 = cse_var_20 + 12
+ cse_var_5: T.int32 = cse_var_20 + 11
+ cse_var_4: T.int32 = cse_var_20 + 10
+ cse_var_3: T.int32 = cse_var_20 + 1
placeholder_6 = T.Buffer((78656,), data=placeholder_1.data)
placeholder_7 = T.Buffer((32768,), data=placeholder.data)
placeholder_8 = T.Buffer((4916,), "int32", data=placeholder_2.data)
- if T.likely(elem_idx < placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused]):
- cse_var_2: T.int32 = i_outer_inner * 256 + i_inner * 16
- compute_2[cse_var_2] = compute_2[cse_var_2] + placeholder_6[placeholder_5[i0_outer_i1_outer_fused] * 16 + elem_idx * 16] * T.max(placeholder_7[i_outer_inner * 4096 + i_inner * 256 + placeholder_8[placeholder_5[i0_outer_i1_outer_fused] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused]):
- cse_var_3: T.int32 = i_outer_inner * 256 + i_inner * 16 + 1
- compute_2[cse_var_3] = compute_2[cse_var_3] + placeholder_6[placeholder_5[i0_outer_i1_outer_fused] * 16 + elem_idx * 16 + 1] * T.max(placeholder_7[i_outer_inner * 4096 + i_inner * 256 + placeholder_8[placeholder_5[i0_outer_i1_outer_fused] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused]):
- cse_var_4: T.int32 = i_outer_inner * 256 + i_inner * 16 + 2
- compute_2[cse_var_4] = compute_2[cse_var_4] + placeholder_6[placeholder_5[i0_outer_i1_outer_fused] * 16 + elem_idx * 16 + 2] * T.max(placeholder_7[i_outer_inner * 4096 + i_inner * 256 + placeholder_8[placeholder_5[i0_outer_i1_outer_fused] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused]):
- cse_var_5: T.int32 = i_outer_inner * 256 + i_inner * 16 + 3
- compute_2[cse_var_5] = compute_2[cse_var_5] + placeholder_6[placeholder_5[i0_outer_i1_outer_fused] * 16 + elem_idx * 16 + 3] * T.max(placeholder_7[i_outer_inner * 4096 + i_inner * 256 + placeholder_8[placeholder_5[i0_outer_i1_outer_fused] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused]):
- cse_var_6: T.int32 = i_outer_inner * 256 + i_inner * 16 + 4
- compute_2[cse_var_6] = compute_2[cse_var_6] + placeholder_6[placeholder_5[i0_outer_i1_outer_fused] * 16 + elem_idx * 16 + 4] * T.max(placeholder_7[i_outer_inner * 4096 + i_inner * 256 + placeholder_8[placeholder_5[i0_outer_i1_outer_fused] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused]):
- cse_var_7: T.int32 = i_outer_inner * 256 + i_inner * 16 + 5
- compute_2[cse_var_7] = compute_2[cse_var_7] + placeholder_6[placeholder_5[i0_outer_i1_outer_fused] * 16 + elem_idx * 16 + 5] * T.max(placeholder_7[i_outer_inner * 4096 + i_inner * 256 + placeholder_8[placeholder_5[i0_outer_i1_outer_fused] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused]):
- cse_var_8: T.int32 = i_outer_inner * 256 + i_inner * 16 + 6
- compute_2[cse_var_8] = compute_2[cse_var_8] + placeholder_6[placeholder_5[i0_outer_i1_outer_fused] * 16 + elem_idx * 16 + 6] * T.max(placeholder_7[i_outer_inner * 4096 + i_inner * 256 + placeholder_8[placeholder_5[i0_outer_i1_outer_fused] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused]):
- cse_var_9: T.int32 = i_outer_inner * 256 + i_inner * 16 + 7
- compute_2[cse_var_9] = compute_2[cse_var_9] + placeholder_6[placeholder_5[i0_outer_i1_outer_fused] * 16 + elem_idx * 16 + 7] * T.max(placeholder_7[i_outer_inner * 4096 + i_inner * 256 + placeholder_8[placeholder_5[i0_outer_i1_outer_fused] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused]):
- cse_var_10: T.int32 = i_outer_inner * 256 + i_inner * 16 + 8
- compute_2[cse_var_10] = compute_2[cse_var_10] + placeholder_6[placeholder_5[i0_outer_i1_outer_fused] * 16 + elem_idx * 16 + 8] * T.max(placeholder_7[i_outer_inner * 4096 + i_inner * 256 + placeholder_8[placeholder_5[i0_outer_i1_outer_fused] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused]):
- cse_var_11: T.int32 = i_outer_inner * 256 + i_inner * 16 + 9
- compute_2[cse_var_11] = compute_2[cse_var_11] + placeholder_6[placeholder_5[i0_outer_i1_outer_fused] * 16 + elem_idx * 16 + 9] * T.max(placeholder_7[i_outer_inner * 4096 + i_inner * 256 + placeholder_8[placeholder_5[i0_outer_i1_outer_fused] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused]):
- cse_var_12: T.int32 = i_outer_inner * 256 + i_inner * 16 + 10
- compute_2[cse_var_12] = compute_2[cse_var_12] + placeholder_6[placeholder_5[i0_outer_i1_outer_fused] * 16 + elem_idx * 16 + 10] * T.max(placeholder_7[i_outer_inner * 4096 + i_inner * 256 + placeholder_8[placeholder_5[i0_outer_i1_outer_fused] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused]):
- cse_var_13: T.int32 = i_outer_inner * 256 + i_inner * 16 + 11
- compute_2[cse_var_13] = compute_2[cse_var_13] + placeholder_6[placeholder_5[i0_outer_i1_outer_fused] * 16 + elem_idx * 16 + 11] * T.max(placeholder_7[i_outer_inner * 4096 + i_inner * 256 + placeholder_8[placeholder_5[i0_outer_i1_outer_fused] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused]):
- cse_var_14: T.int32 = i_outer_inner * 256 + i_inner * 16 + 12
- compute_2[cse_var_14] = compute_2[cse_var_14] + placeholder_6[placeholder_5[i0_outer_i1_outer_fused] * 16 + elem_idx * 16 + 12] * T.max(placeholder_7[i_outer_inner * 4096 + i_inner * 256 + placeholder_8[placeholder_5[i0_outer_i1_outer_fused] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused]):
- cse_var_15: T.int32 = i_outer_inner * 256 + i_inner * 16 + 13
- compute_2[cse_var_15] = compute_2[cse_var_15] + placeholder_6[placeholder_5[i0_outer_i1_outer_fused] * 16 + elem_idx * 16 + 13] * T.max(placeholder_7[i_outer_inner * 4096 + i_inner * 256 + placeholder_8[placeholder_5[i0_outer_i1_outer_fused] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused]):
- cse_var_16: T.int32 = i_outer_inner * 256 + i_inner * 16 + 14
- compute_2[cse_var_16] = compute_2[cse_var_16] + placeholder_6[placeholder_5[i0_outer_i1_outer_fused] * 16 + elem_idx * 16 + 14] * T.max(placeholder_7[i_outer_inner * 4096 + i_inner * 256 + placeholder_8[placeholder_5[i0_outer_i1_outer_fused] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused]):
- cse_var_17: T.int32 = i_outer_inner * 256 + i_inner * 16 + 15
- compute_2[cse_var_17] = compute_2[cse_var_17] + placeholder_6[placeholder_5[i0_outer_i1_outer_fused] * 16 + elem_idx * 16 + 15] * T.max(placeholder_7[i_outer_inner * 4096 + i_inner * 256 + placeholder_8[placeholder_5[i0_outer_i1_outer_fused] + elem_idx]], T.float32(0))
- for i0_inner in range(128):
- cse_var_18: T.int32 = i0_inner * 512 + i0_outer_i1_outer_fused * 16
+ compute_2[cse_var_20] = compute_2[cse_var_20] + placeholder_6[placeholder_5[cse_var_19] * 16 + cse_var_21] * T.max(placeholder_7[cse_var_18 + placeholder_8[placeholder_5[cse_var_19] + elem_idx]], T.float32(0))
+ compute_2[cse_var_3] = compute_2[cse_var_3] + placeholder_6[placeholder_5[cse_var_19] * 16 + cse_var_21 + 1] * T.max(placeholder_7[cse_var_18 + placeholder_8[placeholder_5[cse_var_19] + elem_idx]], T.float32(0))
+ compute_2[cse_var_10] = compute_2[cse_var_10] + placeholder_6[placeholder_5[cse_var_19] * 16 + cse_var_21 + 2] * T.max(placeholder_7[cse_var_18 + placeholder_8[placeholder_5[cse_var_19] + elem_idx]], T.float32(0))
+ compute_2[cse_var_11] = compute_2[cse_var_11] + placeholder_6[placeholder_5[cse_var_19] * 16 + cse_var_21 + 3] * T.max(placeholder_7[cse_var_18 + placeholder_8[placeholder_5[cse_var_19] + elem_idx]], T.float32(0))
+ compute_2[cse_var_12] = compute_2[cse_var_12] + placeholder_6[placeholder_5[cse_var_19] * 16 + cse_var_21 + 4] * T.max(placeholder_7[cse_var_18 + placeholder_8[placeholder_5[cse_var_19] + elem_idx]], T.float32(0))
+ compute_2[cse_var_13] = compute_2[cse_var_13] + placeholder_6[placeholder_5[cse_var_19] * 16 + cse_var_21 + 5] * T.max(placeholder_7[cse_var_18 + placeholder_8[placeholder_5[cse_var_19] + elem_idx]], T.float32(0))
+ compute_2[cse_var_14] = compute_2[cse_var_14] + placeholder_6[placeholder_5[cse_var_19] * 16 + cse_var_21 + 6] * T.max(placeholder_7[cse_var_18 + placeholder_8[placeholder_5[cse_var_19] + elem_idx]], T.float32(0))
+ compute_2[cse_var_15] = compute_2[cse_var_15] + placeholder_6[placeholder_5[cse_var_19] * 16 + cse_var_21 + 7] * T.max(placeholder_7[cse_var_18 + placeholder_8[placeholder_5[cse_var_19] + elem_idx]], T.float32(0))
+ compute_2[cse_var_16] = compute_2[cse_var_16] + placeholder_6[placeholder_5[cse_var_19] * 16 + cse_var_21 + 8] * T.max(placeholder_7[cse_var_18 + placeholder_8[placeholder_5[cse_var_19] + elem_idx]], T.float32(0))
+ compute_2[cse_var_17] = compute_2[cse_var_17] + placeholder_6[placeholder_5[cse_var_19] * 16 + cse_var_21 + 9] * T.max(placeholder_7[cse_var_18 + placeholder_8[placeholder_5[cse_var_19] + elem_idx]], T.float32(0))
+ compute_2[cse_var_4] = compute_2[cse_var_4] + placeholder_6[placeholder_5[cse_var_19] * 16 + cse_var_21 + 10] * T.max(placeholder_7[cse_var_18 + placeholder_8[placeholder_5[cse_var_19] + elem_idx]], T.float32(0))
+ compute_2[cse_var_5] = compute_2[cse_var_5] + placeholder_6[placeholder_5[cse_var_19] * 16 + cse_var_21 + 11] * T.max(placeholder_7[cse_var_18 + placeholder_8[placeholder_5[cse_var_19] + elem_idx]], T.float32(0))
+ compute_2[cse_var_6] = compute_2[cse_var_6] + placeholder_6[placeholder_5[cse_var_19] * 16 + cse_var_21 + 12] * T.max(placeholder_7[cse_var_18 + placeholder_8[placeholder_5[cse_var_19] + elem_idx]], T.float32(0))
+ compute_2[cse_var_7] = compute_2[cse_var_7] + placeholder_6[placeholder_5[cse_var_19] * 16 + cse_var_21 + 13] * T.max(placeholder_7[cse_var_18 + placeholder_8[placeholder_5[cse_var_19] + elem_idx]], T.float32(0))
+ compute_2[cse_var_8] = compute_2[cse_var_8] + placeholder_6[placeholder_5[cse_var_19] * 16 + cse_var_21 + 14] * T.max(placeholder_7[cse_var_18 + placeholder_8[placeholder_5[cse_var_19] + elem_idx]], T.float32(0))
+ compute_2[cse_var_9] = compute_2[cse_var_9] + placeholder_6[placeholder_5[cse_var_19] * 16 + cse_var_21 + 15] * T.max(placeholder_7[cse_var_18 + placeholder_8[placeholder_5[cse_var_19] + elem_idx]], T.float32(0))
+ for i0_inner in range(16):
+ cse_var_22: T.int32 = i0_outer_i1_outer_fused // 16 * 8192 + i0_inner * 512 + i0_outer_i1_outer_fused % 16 * 32
compute_3 = T.Buffer((65536,), data=compute.data)
placeholder_5 = T.Buffer((65536,), data=placeholder_4.data)
- compute_3[cse_var_18:cse_var_18 + 16] = T.max(compute_2[i0_inner * 16:i0_inner * 16 + 16] + placeholder_5[cse_var_18:cse_var_18 + 16], T.Broadcast(T.float32(0), 16))
+ compute_3[cse_var_22:cse_var_22 + 32] = T.max(compute_2[i0_inner * 32:i0_inner * 32 + 32] + placeholder_5[cse_var_22:cse_var_22 + 32], T.Broadcast(T.float32(0), 32))
@@ -518,7 +506,7 @@ We build the binary and check its correctness and performance.
.. code-block:: none
- Execution time of this operator: 1.700 ms
+ Execution time of this operator: 1.670 ms
diff --git a/docs/_sources/how_to/tune_with_autotvm/sg_execution_times.rst.txt b/docs/_sources/how_to/tune_with_autotvm/sg_execution_times.rst.txt
index d68b89f5da..4e29a1294d 100644
--- a/docs/_sources/how_to/tune_with_autotvm/sg_execution_times.rst.txt
+++ b/docs/_sources/how_to/tune_with_autotvm/sg_execution_times.rst.txt
@@ -5,16 +5,16 @@
Computation times
=================
-**00:35.388** total execution time for **how_to_tune_with_autotvm** files:
+**00:45.727** total execution time for **how_to_tune_with_autotvm** files:
+--------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autotvm_tune_conv2d_cuda.py` (``tune_conv2d_cuda.py``) | 00:35.352 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autotvm_tune_conv2d_cuda.py` (``tune_conv2d_cuda.py``) | 00:45.695 | 0.0 MB |
+--------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autotvm_tune_relay_x86.py` (``tune_relay_x86.py``) | 00:00.021 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autotvm_tune_relay_x86.py` (``tune_relay_x86.py``) | 00:00.019 | 0.0 MB |
+--------------------------------------------------------------------------------------------------+-----------+--------+
| :ref:`sphx_glr_how_to_tune_with_autotvm_tune_relay_cuda.py` (``tune_relay_cuda.py``) | 00:00.005 | 0.0 MB |
+--------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autotvm_tune_relay_arm.py` (``tune_relay_arm.py``) | 00:00.005 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autotvm_tune_relay_arm.py` (``tune_relay_arm.py``) | 00:00.004 | 0.0 MB |
+--------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autotvm_tune_relay_mobile_gpu.py` (``tune_relay_mobile_gpu.py``) | 00:00.005 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autotvm_tune_relay_mobile_gpu.py` (``tune_relay_mobile_gpu.py``) | 00:00.004 | 0.0 MB |
+--------------------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/how_to/tune_with_autotvm/tune_conv2d_cuda.rst.txt b/docs/_sources/how_to/tune_with_autotvm/tune_conv2d_cuda.rst.txt
index e51ceac2e1..28b7b73b57 100644
--- a/docs/_sources/how_to/tune_with_autotvm/tune_conv2d_cuda.rst.txt
+++ b/docs/_sources/how_to/tune_with_autotvm/tune_conv2d_cuda.rst.txt
@@ -390,501 +390,162 @@ for this template
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 1, 1, 16]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 64, 1]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,3314684
- No: 2 GFLOPS: 0.00/0.00 result: Traceback (most recent call last):
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
- func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
- func = build(s, args, target=target, runtime=runtime)
- File "/workspace/python/tvm/driver/build_module.py", line 227, in build
- input_mod = lower(inputs, args, name=name, binds=binds)
- File "/workspace/python/tvm/driver/build_module.py", line 134, in lower
- return ffi.lower_schedule(inp, args, name, binds, simple_mode)
- File "tvm/_ffi/_cython/./packed_func.pxi", line 331, in tvm._ffi._cy3.core.PackedFuncBase.__call__
- File "tvm/_ffi/_cython/./packed_func.pxi", line 276, in tvm._ffi._cy3.core.FuncCall
- File "tvm/_ffi/_cython/./base.pxi", line 181, in tvm._ffi._cy3.core.CHECK_CALL
- tvm._ffi.base.TVMError: Traceback (most recent call last):
- 24: TVMFuncCall
- at ../src/runtime/c_runtime_api.cc:477
- 23: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 22: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 21: operator()
- at ../include/tvm/runtime/packed_func.h:1734
- 20: unpack_call<tvm::IRModule, 5, tvm::<lambda(tvm::te::Schedule, const tvm::runtime::Array<tvm::runtime::ObjectRef>&, const tvm::runtime::String&, const tvm::runtime::Map<tvm::te::Tensor, tvm::tir::Buffer>&, bool)> >
- at ../include/tvm/runtime/packed_func.h:1674
- 19: run<>
- at ../include/tvm/runtime/packed_func.h:1634
- 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1649
- 13: operator()
- at ../src/driver/driver_api.cc:402
- 12: tvm::LowerSchedule(tvm::te::Schedule, tvm::runtime::Array<tvm::runtime::ObjectRef, void> const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::unordered_map<tvm::te::Tensor, tvm::tir::Buffer, std::hash<tvm::te::Tensor>, std::equal_to<tvm::te::Tensor>, std::allocator<std::pair<tvm::te::Tensor const, tvm::tir::Buffer> > > const&, tvm::GlobalVarSupply, bool)
- at ../src/driver/driver_api.cc:388
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:283
- 10: tvm::transform::Pass::operator()(tvm::IRModule) const
- at ../src/ir/transform.cc:258
- 9: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 8: tvm::transform::SequentialNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:451
- 7: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 6: tvm::tir::transform::PrimFuncPassNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/tir/ir/transform.cc:100
- 5: tvm::runtime::TypedPackedFunc<tvm::tir::PrimFunc (tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext)>::operator()(tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext) const
- at ../include/tvm/runtime/packed_func.h:1753
- 4: tvm::tir::PrimFunc tvm::runtime::detail::typed_packed_call_dispatcher<tvm::tir::PrimFunc>::run<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::runtime::PackedFunc const&, tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&)
- at ../include/tvm/runtime/packed_func.h:1697
- 3: tvm::runtime::TVMRetValue tvm::runtime::PackedFunc::operator()<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&) const
- at ../include/tvm/runtime/packed_func.h:1621
- 2: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 1: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 0: operator()
- at ../src/runtime/c_runtime_api.cc:534
- File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
- raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel
-
- Traceback (most recent call last):
- 24: TVMFuncCall
- at ../src/runtime/c_runtime_api.cc:477
- 23: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 22: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 21: operator()
- at ../include/tvm/runtime/packed_func.h:1734
- 20: unpack_call<tvm::IRModule, 5, tvm::<lambda(tvm::te::Schedule, const tvm::runtime::Array<tvm::runtime::ObjectRef>&, const tvm::runtime::String&, const tvm::runtime::Map<tvm::te::Tensor, tvm::tir::Buffer>&, bool)> >
- at ../include/tvm/runtime/packed_func.h:1674
- 19: run<>
- at ../include/tvm/runtime/packed_func.h:1634
- 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1649
- 13: operator()
- at ../src/driver/driver_api.cc:402
- 12: tvm::LowerSchedule(tvm::te::Schedule, tvm::runtime::Array<tvm::runtime::ObjectRef, void> const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::unordered_map<tvm::te::Tensor, tvm::tir::Buffer, std::hash<tvm::te::Tensor>, std::equal_to<tvm::te::Tensor>, std::allocator<std::pair<tvm::te::Tensor const, tvm::tir::Buffer> > > const&, tvm::GlobalVarSupply, bool)
- at ../src/driver/driver_api.cc:388
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:283
- 10: tvm::transform::Pass::operator()(tvm::IRModule) const
- at ../src/ir/transform.cc:258
- 9: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 8: tvm::transform::SequentialNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:451
- 7: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 6: tvm::tir::transform::PrimFuncPassNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/tir/ir/transform.cc:100
- 5: tvm::runtime::TypedPackedFunc<tvm::tir::PrimFunc (tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext)>::operator()(tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext) const
- at ../include/tvm/runtime/packed_func.h:1753
- 4: tvm::tir::PrimFunc tvm::runtime::detail::typed_packed_call_dispatcher<tvm::tir::PrimFunc>::run<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::runtime::PackedFunc const&, tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&)
- at ../include/tvm/runtime/packed_func.h:1697
- 3: tvm::runtime::TVMRetValue tvm::runtime::PackedFunc::operator()<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&) const
- at ../include/tvm/runtime/packed_func.h:1621
- 2: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 1: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 0: operator()
- at ../src/runtime/c_runtime_api.cc:534
- File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
- raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 64, 1, 2]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 64, 2]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,2769201
- No: 3 GFLOPS: 0.00/0.00 result: Traceback (most recent call last):
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
- func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
- func = build(s, args, target=target, runtime=runtime)
- File "/workspace/python/tvm/driver/build_module.py", line 227, in build
- input_mod = lower(inputs, args, name=name, binds=binds)
- File "/workspace/python/tvm/driver/build_module.py", line 134, in lower
- return ffi.lower_schedule(inp, args, name, binds, simple_mode)
- File "tvm/_ffi/_cython/./packed_func.pxi", line 331, in tvm._ffi._cy3.core.PackedFuncBase.__call__
- File "tvm/_ffi/_cython/./packed_func.pxi", line 276, in tvm._ffi._cy3.core.FuncCall
- File "tvm/_ffi/_cython/./base.pxi", line 181, in tvm._ffi._cy3.core.CHECK_CALL
- tvm._ffi.base.TVMError: Traceback (most recent call last):
- 24: TVMFuncCall
- at ../src/runtime/c_runtime_api.cc:477
- 23: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 22: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 21: operator()
- at ../include/tvm/runtime/packed_func.h:1734
- 20: unpack_call<tvm::IRModule, 5, tvm::<lambda(tvm::te::Schedule, const tvm::runtime::Array<tvm::runtime::ObjectRef>&, const tvm::runtime::String&, const tvm::runtime::Map<tvm::te::Tensor, tvm::tir::Buffer>&, bool)> >
- at ../include/tvm/runtime/packed_func.h:1674
- 19: run<>
- at ../include/tvm/runtime/packed_func.h:1634
- 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1649
- 13: operator()
- at ../src/driver/driver_api.cc:402
- 12: tvm::LowerSchedule(tvm::te::Schedule, tvm::runtime::Array<tvm::runtime::ObjectRef, void> const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::unordered_map<tvm::te::Tensor, tvm::tir::Buffer, std::hash<tvm::te::Tensor>, std::equal_to<tvm::te::Tensor>, std::allocator<std::pair<tvm::te::Tensor const, tvm::tir::Buffer> > > const&, tvm::GlobalVarSupply, bool)
- at ../src/driver/driver_api.cc:388
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:283
- 10: tvm::transform::Pass::operator()(tvm::IRModule) const
- at ../src/ir/transform.cc:258
- 9: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 8: tvm::transform::SequentialNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:451
- 7: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 6: tvm::tir::transform::PrimFuncPassNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/tir/ir/transform.cc:100
- 5: tvm::runtime::TypedPackedFunc<tvm::tir::PrimFunc (tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext)>::operator()(tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext) const
- at ../include/tvm/runtime/packed_func.h:1753
- 4: tvm::tir::PrimFunc tvm::runtime::detail::typed_packed_call_dispatcher<tvm::tir::PrimFunc>::run<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::runtime::PackedFunc const&, tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&)
- at ../include/tvm/runtime/packed_func.h:1697
- 3: tvm::runtime::TVMRetValue tvm::runtime::PackedFunc::operator()<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&) const
- at ../include/tvm/runtime/packed_func.h:1621
- 2: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 1: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 0: operator()
- at ../src/runtime/c_runtime_api.cc:534
- File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
- raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel
-
- Traceback (most recent call last):
- 24: TVMFuncCall
- at ../src/runtime/c_runtime_api.cc:477
- 23: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 22: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 21: operator()
- at ../include/tvm/runtime/packed_func.h:1734
- 20: unpack_call<tvm::IRModule, 5, tvm::<lambda(tvm::te::Schedule, const tvm::runtime::Array<tvm::runtime::ObjectRef>&, const tvm::runtime::String&, const tvm::runtime::Map<tvm::te::Tensor, tvm::tir::Buffer>&, bool)> >
- at ../include/tvm/runtime/packed_func.h:1674
- 19: run<>
- at ../include/tvm/runtime/packed_func.h:1634
- 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1649
- 13: operator()
- at ../src/driver/driver_api.cc:402
- 12: tvm::LowerSchedule(tvm::te::Schedule, tvm::runtime::Array<tvm::runtime::ObjectRef, void> const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::unordered_map<tvm::te::Tensor, tvm::tir::Buffer, std::hash<tvm::te::Tensor>, std::equal_to<tvm::te::Tensor>, std::allocator<std::pair<tvm::te::Tensor const, tvm::tir::Buffer> > > const&, tvm::GlobalVarSupply, bool)
- at ../src/driver/driver_api.cc:388
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:283
- 10: tvm::transform::Pass::operator()(tvm::IRModule) const
- at ../src/ir/transform.cc:258
- 9: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 8: tvm::transform::SequentialNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:451
- 7: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 6: tvm::tir::transform::PrimFuncPassNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/tir/ir/transform.cc:100
- 5: tvm::runtime::TypedPackedFunc<tvm::tir::PrimFunc (tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext)>::operator()(tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext) const
- at ../include/tvm/runtime/packed_func.h:1753
- 4: tvm::tir::PrimFunc tvm::runtime::detail::typed_packed_call_dispatcher<tvm::tir::PrimFunc>::run<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::runtime::PackedFunc const&, tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&)
- at ../include/tvm/runtime/packed_func.h:1697
- 3: tvm::runtime::TVMRetValue tvm::runtime::PackedFunc::operator()<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&) const
- at ../include/tvm/runtime/packed_func.h:1621
- 2: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 1: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 0: operator()
- at ../src/runtime/c_runtime_api.cc:534
- File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
- raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 1, 4, 32]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 4, 32]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,8280114
- No: 4 GFLOPS: 0.00/0.00 result: Traceback (most recent call last):
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
- func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
- func = build(s, args, target=target, runtime=runtime)
- File "/workspace/python/tvm/driver/build_module.py", line 227, in build
- input_mod = lower(inputs, args, name=name, binds=binds)
- File "/workspace/python/tvm/driver/build_module.py", line 134, in lower
- return ffi.lower_schedule(inp, args, name, binds, simple_mode)
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 1, 1, 256]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 512, 1]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,4681596
+ No: 2 GFLOPS: 42.72/42.72 result: MeasureResult(costs=(0.005419384523809524,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.903918981552124, timestamp=1679085229.6098208) [('tile_f', [-1, 16, 32, 1]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 1, 2]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,231264
+ No: 3 GFLOPS: 0.00/42.72 result: Traceback (most recent call last):
+ File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 742, in __call__
+ yield remote, remote.load_module(os.path.split(build_result.filename)[1])
+ File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 706, in run_through_rpc
+ costs = time_f(*args).results
+ File "/workspace/python/tvm/runtime/module.py", line 357, in evaluator
+ blob = feval(*args)
File "tvm/_ffi/_cython/./packed_func.pxi", line 331, in tvm._ffi._cy3.core.PackedFuncBase.__call__
- File "tvm/_ffi/_cython/./packed_func.pxi", line 276, in tvm._ffi._cy3.core.FuncCall
+ File "tvm/_ffi/_cython/./packed_func.pxi", line 262, in tvm._ffi._cy3.core.FuncCall
+ File "tvm/_ffi/_cython/./packed_func.pxi", line 251, in tvm._ffi._cy3.core.FuncCall3
File "tvm/_ffi/_cython/./base.pxi", line 181, in tvm._ffi._cy3.core.CHECK_CALL
tvm._ffi.base.TVMError: Traceback (most recent call last):
- 24: TVMFuncCall
+ 4: TVMFuncCall
at ../src/runtime/c_runtime_api.cc:477
- 23: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
+ 3: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
at ../include/tvm/runtime/packed_func.h:1217
- 22: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 21: operator()
- at ../include/tvm/runtime/packed_func.h:1734
- 20: unpack_call<tvm::IRModule, 5, tvm::<lambda(tvm::te::Schedule, const tvm::runtime::Array<tvm::runtime::ObjectRef>&, const tvm::runtime::String&, const tvm::runtime::Map<tvm::te::Tensor, tvm::tir::Buffer>&, bool)> >
- at ../include/tvm/runtime/packed_func.h:1674
- 19: run<>
- at ../include/tvm/runtime/packed_func.h:1634
- 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1649
- 13: operator()
- at ../src/driver/driver_api.cc:402
- 12: tvm::LowerSchedule(tvm::te::Schedule, tvm::runtime::Array<tvm::runtime::ObjectRef, void> const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::unordered_map<tvm::te::Tensor, tvm::tir::Buffer, std::hash<tvm::te::Tensor>, std::equal_to<tvm::te::Tensor>, std::allocator<std::pair<tvm::te::Tensor const, tvm::tir::Buffer> > > const&, tvm::GlobalVarSupply, bool)
- at ../src/driver/driver_api.cc:388
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:283
- 10: tvm::transform::Pass::operator()(tvm::IRModule) const
- at ../src/ir/transform.cc:258
- 9: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 8: tvm::transform::SequentialNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:451
- 7: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 6: tvm::tir::transform::PrimFuncPassNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/tir/ir/transform.cc:100
- 5: tvm::runtime::TypedPackedFunc<tvm::tir::PrimFunc (tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext)>::operator()(tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext) const
- at ../include/tvm/runtime/packed_func.h:1753
- 4: tvm::tir::PrimFunc tvm::runtime::detail::typed_packed_call_dispatcher<tvm::tir::PrimFunc>::run<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::runtime::PackedFunc const&, tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&)
- at ../include/tvm/runtime/packed_func.h:1697
- 3: tvm::runtime::TVMRetValue tvm::runtime::PackedFunc::operator()<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&) const
- at ../include/tvm/runtime/packed_func.h:1621
- 2: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 1: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 0: operator()
- at ../src/runtime/c_runtime_api.cc:534
- File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
- raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel
+ 2: tvm::runtime::RPCWrappedFunc::operator()(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
+ at ../src/runtime/rpc/rpc_module.cc:129
+ 1: tvm::runtime::RPCClientSession::CallFunc(void*, TVMValue const*, int const*, int, std::function<void (tvm::runtime::TVMArgs)> const&)
+ at ../src/runtime/rpc/rpc_endpoint.cc:1012
+ 0: tvm::runtime::RPCEndpoint::CallFunc(void*, TVMValue const*, int const*, int, std::function<void (tvm::runtime::TVMArgs)>)
+ at ../src/runtime/rpc/rpc_endpoint.cc:804
+ File "../src/runtime/rpc/rpc_endpoint.cc", line 804
+ TVMError:
+ ---------------------------------------------------------------
+ An error occurred during the execution of TVM.
+ For more information, please see: https://tvm.apache.org/docs/errors.html
+ ---------------------------------------------------------------
+ Check failed: (code == RPCCode::kReturn) is false: code=kShutdown
+
+ During handling of the above exception, another exception occurred:
Traceback (most recent call last):
- 24: TVMFuncCall
- at ../src/runtime/c_runtime_api.cc:477
- 23: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 22: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 21: operator()
- at ../include/tvm/runtime/packed_func.h:1734
- 20: unpack_call<tvm::IRModule, 5, tvm::<lambda(tvm::te::Schedule, const tvm::runtime::Array<tvm::runtime::ObjectRef>&, const tvm::runtime::String&, const tvm::runtime::Map<tvm::te::Tensor, tvm::tir::Buffer>&, bool)> >
- at ../include/tvm/runtime/packed_func.h:1674
- 19: run<>
- at ../include/tvm/runtime/packed_func.h:1634
- 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1649
- 13: operator()
- at ../src/driver/driver_api.cc:402
- 12: tvm::LowerSchedule(tvm::te::Schedule, tvm::runtime::Array<tvm::runtime::ObjectRef, void> const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::unordered_map<tvm::te::Tensor, tvm::tir::Buffer, std::hash<tvm::te::Tensor>, std::equal_to<tvm::te::Tensor>, std::allocator<std::pair<tvm::te::Tensor const, tvm::tir::Buffer> > > const&, tvm::GlobalVarSupply, bool)
- at ../src/driver/driver_api.cc:388
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:283
- 10: tvm::transform::Pass::operator()(tvm::IRModule) const
- at ../src/ir/transform.cc:258
- 9: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 8: tvm::transform::SequentialNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:451
- 7: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 6: tvm::tir::transform::PrimFuncPassNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/tir/ir/transform.cc:100
- 5: tvm::runtime::TypedPackedFunc<tvm::tir::PrimFunc (tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext)>::operator()(tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext) const
- at ../include/tvm/runtime/packed_func.h:1753
- 4: tvm::tir::PrimFunc tvm::runtime::detail::typed_packed_call_dispatcher<tvm::tir::PrimFunc>::run<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::runtime::PackedFunc const&, tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&)
- at ../include/tvm/runtime/packed_func.h:1697
- 3: tvm::runtime::TVMRetValue tvm::runtime::PackedFunc::operator()<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&) const
- at ../include/tvm/runtime/packed_func.h:1621
- 2: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 1: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 0: operator()
- at ../src/runtime/c_runtime_api.cc:534
- File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
- raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 512, 1, 1]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 8, 16]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,4583049
- No: 5 GFLOPS: 10.92/10.92 result: MeasureResult(costs=(0.02120587783333333,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.8181979656219482, timestamp=1679084112.4073758) [('tile_f', [-1, 1, 2, 32]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 2, 8]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,6681370
- No: 6 GFLOPS: 0.00/10.92 result: Traceback (most recent call last):
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
- func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
- func = build(s, args, target=target, runtime=runtime)
- File "/workspace/python/tvm/driver/build_module.py", line 227, in build
- input_mod = lower(inputs, args, name=name, binds=binds)
- File "/workspace/python/tvm/driver/build_module.py", line 134, in lower
- return ffi.lower_schedule(inp, args, name, binds, simple_mode)
- File "tvm/_ffi/_cython/./packed_func.pxi", line 331, in tvm._ffi._cy3.core.PackedFuncBase.__call__
- File "tvm/_ffi/_cython/./packed_func.pxi", line 276, in tvm._ffi._cy3.core.FuncCall
- File "tvm/_ffi/_cython/./base.pxi", line 181, in tvm._ffi._cy3.core.CHECK_CALL
+ File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 706, in run_through_rpc
+ costs = time_f(*args).results
+ File "/usr/lib/python3.7/contextlib.py", line 130, in __exit__
+ self.gen.throw(type, value, traceback)
+ File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 746, in __call__
+ remote.remove(build_result.filename)
+ File "/workspace/python/tvm/rpc/client.py", line 144, in remove
+ self._remote_funcs["remove"] = self.get_function("tvm.rpc.server.remove")
+ File "/workspace/python/tvm/rpc/client.py", line 72, in get_function
+ return self._sess.get_function(name)
+ File "/workspace/python/tvm/runtime/module.py", line 171, in get_function
+ self.handle, c_str(name), ctypes.c_int(query_imports), ctypes.byref(ret_handle)
+ File "/workspace/python/tvm/_ffi/base.py", line 348, in check_call
+ raise get_last_ffi_error()
tvm._ffi.base.TVMError: Traceback (most recent call last):
- 24: TVMFuncCall
- at ../src/runtime/c_runtime_api.cc:477
- 23: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 22: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 21: operator()
- at ../include/tvm/runtime/packed_func.h:1734
- 20: unpack_call<tvm::IRModule, 5, tvm::<lambda(tvm::te::Schedule, const tvm::runtime::Array<tvm::runtime::ObjectRef>&, const tvm::runtime::String&, const tvm::runtime::Map<tvm::te::Tensor, tvm::tir::Buffer>&, bool)> >
- at ../include/tvm/runtime/packed_func.h:1674
- 19: run<>
- at ../include/tvm/runtime/packed_func.h:1634
- 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1649
- 13: operator()
- at ../src/driver/driver_api.cc:402
- 12: tvm::LowerSchedule(tvm::te::Schedule, tvm::runtime::Array<tvm::runtime::ObjectRef, void> const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::unordered_map<tvm::te::Tensor, tvm::tir::Buffer, std::hash<tvm::te::Tensor>, std::equal_to<tvm::te::Tensor>, std::allocator<std::pair<tvm::te::Tensor const, tvm::tir::Buffer> > > const&, tvm::GlobalVarSupply, bool)
- at ../src/driver/driver_api.cc:388
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:283
- 10: tvm::transform::Pass::operator()(tvm::IRModule) const
- at ../src/ir/transform.cc:258
- 9: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 8: tvm::transform::SequentialNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:451
- 7: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 6: tvm::tir::transform::PrimFuncPassNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/tir/ir/transform.cc:100
- 5: tvm::runtime::TypedPackedFunc<tvm::tir::PrimFunc (tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext)>::operator()(tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext) const
- at ../include/tvm/runtime/packed_func.h:1753
- 4: tvm::tir::PrimFunc tvm::runtime::detail::typed_packed_call_dispatcher<tvm::tir::PrimFunc>::run<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::runtime::PackedFunc const&, tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&)
- at ../include/tvm/runtime/packed_func.h:1697
- 3: tvm::runtime::TVMRetValue tvm::runtime::PackedFunc::operator()<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&) const
+ 52: 0xffffffffffffffff
+ 51: _start
+ 50: __libc_start_main
+ 49: _Py_UnixMain
+ 48: 0x0000000000650da0
+ 47: 0x0000000000650afa
+ 46: _PyFunction_FastCallDict
+ 45: _PyEval_EvalCodeWithName
+ 44: _PyEval_EvalFrameDefault
+ 43: _PyFunction_FastCallKeywords
+ 42: _PyEval_EvalCodeWithName
+ 41: _PyEval_EvalFrameDefault
+ 40: _PyMethodDef_RawFastCallKeywords
+ 39: 0x0000000000546369
+ 38: _PyEval_EvalCodeWithName
+ 37: _PyEval_EvalFrameDefault
+ 36: _PyFunction_FastCallKeywords
+ 35: _PyEval_EvalCodeWithName
+ 34: _PyEval_EvalFrameDefault
+ 33: _PyFunction_FastCallDict
+ 32: _PyEval_EvalCodeWithName
+ 31: _PyEval_EvalFrameDefault
+ 30: _PyObject_FastCallDict
+ 29: 0x00000000004c06e1
+ 28: _PyFunction_FastCallDict
+ 27: _PyEval_EvalFrameDefault
+ 26: _PyMethodDescr_FastCallKeywords
+ 25: 0x00000000005dcb58
+ 24: 0x00000000005dc83f
+ 23: 0x00000000004ba127
+ 22: _PyEval_EvalFrameDefault
+ 21: _PyFunction_FastCallKeywords
+ 20: _PyEval_EvalFrameDefault
+ 19: _PyFunction_FastCallKeywords
+ 18: _PyEval_EvalFrameDefault
+ 17: _PyFunction_FastCallKeywords
+ 16: _PyEval_EvalCodeWithName
+ 15: _PyEval_EvalFrameDefault
+ 14: 0x0000000000537c30
+ 13: _PyObject_FastCallKeywords
+ 12: 0x00007f32449a8fa2
+ 11: _ctypes_callproc
+ 10: ffi_call
+ 9: ffi_call_unix64
+ 8: TVMModGetFunction
+ at ../src/runtime/c_runtime_api.cc:408
+ 7: tvm::runtime::ModuleNode::GetFunction(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, bool)
+ at ../src/runtime/module.cc:66
+ 6: tvm::runtime::RPCModuleNode::GetFunction(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, tvm::runtime::ObjectPtr<tvm::runtime::Object> const&)
+ at ../src/runtime/rpc/rpc_module.cc:185
+ 5: tvm::runtime::RPCClientSession::GetFunction(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&)
+ at ../src/runtime/rpc/rpc_endpoint.cc:1007
+ 4: tvm::runtime::TVMRetValue tvm::runtime::RPCEndpoint::SysCallRemote<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&>(tvm::runtime::RPCCode, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&)
+ at ../src/runtime/rpc/rpc_endpoint.h:223
+ 3: tvm::runtime::TVMRetValue tvm::runtime::PackedFunc::operator()<int, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&>(int&&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&) const
at ../include/tvm/runtime/packed_func.h:1621
2: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
at ../include/tvm/runtime/packed_func.h:1217
1: Call
at ../include/tvm/runtime/packed_func.h:1213
0: operator()
- at ../src/runtime/c_runtime_api.cc:534
- File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
- raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel
+ at ../src/runtime/rpc/rpc_endpoint.cc:684
+ File "../src/runtime/rpc/rpc_endpoint.cc", line 684
+ TVMError:
+ ---------------------------------------------------------------
+ An error occurred during the execution of TVM.
+ For more information, please see: https://tvm.apache.org/docs/errors.html
+ ---------------------------------------------------------------
+ Check failed: (code == RPCCode::kReturn) is false: code=1
Traceback (most recent call last):
- 24: TVMFuncCall
- at ../src/runtime/c_runtime_api.cc:477
- 23: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 22: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 21: operator()
- at ../include/tvm/runtime/packed_func.h:1734
- 20: unpack_call<tvm::IRModule, 5, tvm::<lambda(tvm::te::Schedule, const tvm::runtime::Array<tvm::runtime::ObjectRef>&, const tvm::runtime::String&, const tvm::runtime::Map<tvm::te::Tensor, tvm::tir::Buffer>&, bool)> >
- at ../include/tvm/runtime/packed_func.h:1674
- 19: run<>
- at ../include/tvm/runtime/packed_func.h:1634
- 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1649
- 13: operator()
- at ../src/driver/driver_api.cc:402
- 12: tvm::LowerSchedule(tvm::te::Schedule, tvm::runtime::Array<tvm::runtime::ObjectRef, void> const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::unordered_map<tvm::te::Tensor, tvm::tir::Buffer, std::hash<tvm::te::Tensor>, std::equal_to<tvm::te::Tensor>, std::allocator<std::pair<tvm::te::Tensor const, tvm::tir::Buffer> > > const&, tvm::GlobalVarSupply, bool)
- at ../src/driver/driver_api.cc:388
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:283
- 10: tvm::transform::Pass::operator()(tvm::IRModule) const
- at ../src/ir/transform.cc:258
- 9: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 8: tvm::transform::SequentialNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:451
- 7: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 6: tvm::tir::transform::PrimFuncPassNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/tir/ir/transform.cc:100
- 5: tvm::runtime::TypedPackedFunc<tvm::tir::PrimFunc (tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext)>::operator()(tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext) const
- at ../include/tvm/runtime/packed_func.h:1753
- 4: tvm::tir::PrimFunc tvm::runtime::detail::typed_packed_call_dispatcher<tvm::tir::PrimFunc>::run<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::runtime::PackedFunc const&, tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&)
- at ../include/tvm/runtime/packed_func.h:1697
- 3: tvm::runtime::TVMRetValue tvm::runtime::PackedFunc::operator()<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&) const
- at ../include/tvm/runtime/packed_func.h:1621
- 2: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 1: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 0: operator()
- at ../src/runtime/c_runtime_api.cc:534
- File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
- raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 1, 128, 1]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 64, 2]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,5090849
- No: 7 GFLOPS: 0.00/10.92 result: Traceback (most recent call last):
+ 52: 0xffffffffffffffff
+ 51: _start
+ 50: __libc_start_main
+ 49: _Py_UnixMain
+ 48: 0x0000000000650da0
+ 47: 0x0000000000650afa
+ 46: _PyFunction_FastCallDict
+ 45: _PyEval_EvalCodeWithName
+ 44: _PyEval_EvalFrameDefault
+ 43: _PyFunction_FastCallKeywords
+ 42: _PyEval_EvalCodeWithName
+ 41: _PyEval_EvalFrameDefault
+ 40: _PyMethodDef_RawFastCallKeywords
+ 39: 0x0000000000546369
+ 38: _PyEval_EvalCodeWithName
+ 37: _PyEval_EvalFrameDefault
+ 36: _PyFunction_FastCallKeywords
+ 35: _PyEval_EvalCodeWithName
+ 34: _PyEval_EvalFrameDefault
+ 33: _PyFunction_FastCallDict
+ 32: _PyEval_EvalCodeWithName
+ 31: _PyEval_EvalFrameDefault
+ 30: _PyObject_FastCallDict
+ 29: 0x00000000004c06e1
+ 28: _PyFunction_FastCallDict
+ 27: _PyEval_EvalFrameDefault
+ 26: _PyMethodDescr_FastCallKeywords
+ 25: 0x00000000005dcb58
+ 24: 0x00000000005dc83f
+ 23: 0x00000000004ba127
+ 22: _PyEval_EvalFrameDefault
+ 21: _PyFunction_FastCallKeywords
+ 20: _PyEval_EvalFrameDefault
+ 19: _PyFunction_FastCall [('tile_f', [-1, 2, 1, 32]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 1, 4]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,6071966
+ No: 4 GFLOPS: 0.00/42.72 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -1006,9 +667,9 @@ for this template
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 4, 8]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 2, 32]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,9825350
- No: 8 GFLOPS: 37.60/37.60 result: MeasureResult(costs=(0.006157010647058823,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.4753825664520264, timestamp=1679084113.3095083) [('tile_f', [-1, 16, 32, 1]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 2, 8]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,5326464
- No: 9 GFLOPS: 0.00/37.60 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 8, 2, 32]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 2, 8]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,6487993
+ No: 5 GFLOPS: 47.66/47.66 result: MeasureResult(costs=(0.004857183833333334,), error_no=MeasureErrorNo.NO_ERROR, all_cost=7.572648048400879, timestamp=1679085244.793465) [('tile_f', [-1, 8, 2, 4]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 8, 2]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,7405091
+ No: 6 GFLOPS: 0.00/47.66 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -1130,8 +791,9 @@ for this template
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 4, 4, 4]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 4, 16]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,1677617
- No: 10 GFLOPS: 0.00/37.60 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 1, 64, 4]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 256, 2]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,1032813
+ No: 7 GFLOPS: 123.69/123.69 result: MeasureResult(costs=(0.0018716147272727275,), error_no=MeasureErrorNo.NO_ERROR, all_cost=2.975389003753662, timestamp=1679085247.3761587) [('tile_f', [-1, 2, 32, 4]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 2, 16]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,8837311
+ No: 8 GFLOPS: 0.00/123.69 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -1253,8 +915,8 @@ for this template
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 1, 256, 2]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 16, 4]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,1826759
- No: 11 GFLOPS: 0.00/37.60 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 8, 8, 2]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 512, 1]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,1581662
+ No: 9 GFLOPS: 0.00/123.69 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -1376,8 +1038,8 @@ for this template
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 64, 8, 1]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 64, 4]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,2798653
- No: 12 GFLOPS: 0.00/37.60 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 16, 16]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 2, 16]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,3608623
+ No: 10 GFLOPS: 0.00/123.69 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -1499,9 +1161,9 @@ for this template
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 16, 4, 2]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 16, 4]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,4341336
- No: 13 GFLOPS: 248.82/248.82 result: MeasureResult(costs=(0.0009303794031007751,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.3959543704986572, timestamp=1679084116.8684433) [('tile_f', [-1, 2, 8, 4]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 8, 1]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,206482
- No: 14 GFLOPS: 0.00/248.82 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 1, 4, 2]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 128, 4]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,2804192
+ No: 11 GFLOPS: 10.53/123.69 result: MeasureResult(costs=(0.021979039500000002,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.6817421913146973, timestamp=1679085249.2773573) [('tile_f', [-1, 2, 4, 64]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 1, 16]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,5349288
+ No: 12 GFLOPS: 0.00/123.69 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -1623,8 +1285,8 @@ for this template
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 1, 1, 512]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 8, 1]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,3303519
- No: 15 GFLOPS: 0.00/248.82 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 64, 1]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 32, 16]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,8464326
+ No: 13 GFLOPS: 0.00/123.69 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -1746,9 +1408,10 @@ for this template
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 1, 16, 8]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 4, 64]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,5783078
- No: 16 GFLOPS: 1.32/248.82 result: MeasureResult(costs=(0.1759165495,), error_no=MeasureErrorNo.NO_ERROR, all_cost=4.974991321563721, timestamp=1679084119.72121) [('tile_f', [-1, 2, 1, 16]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 16, 1]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,3889105
- No: 17 GFLOPS: 0.00/248.82 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 8, 4]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 8, 32]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,347942
+ No: 14 GFLOPS: 27.41/123.69 result: MeasureResult(costs=(0.008445255833333333,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.5827915668487549, timestamp=1679085251.0693889) [('tile_f', [-1, 32, 1, 1]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 2, 1]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,6008205
+ No: 15 GFLOPS: 151.48/151.48 result: MeasureResult(costs=(0.001528289495238095,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.777534008026123, timestamp=1679085252.1466382) [('tile_f', [-1, 2, 16, 1]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 4, 1]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,1363375
+ No: 16 GFLOPS: 0.00/151.48 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -1870,8 +1533,9 @@ for this template
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 8, 8, 2]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 4, 128]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,2891762
- No: 18 GFLOPS: 0.00/248.82 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 4, 8, 8]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 8, 64]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,6364756
+ No: 17 GFLOPS: 30.41/151.48 result: MeasureResult(costs=(0.007613814714285715,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.7912659645080566, timestamp=1679085254.1267264) [('tile_f', [-1, 16, 1, 1]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 4, 16]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,5936704
+ No: 18 GFLOPS: 0.00/151.48 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -1993,8 +1657,8 @@ for this template
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 1, 2, 256]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 1, 512]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,6194538
- No: 19 GFLOPS: 0.00/248.82 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 1, 8, 8]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 8, 64]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,1139314
+ No: 19 GFLOPS: 0.00/151.48 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -2116,8 +1780,8 @@ for this template
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 16, 2, 2]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 8, 64]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,1718048
- No: 20 GFLOPS: 9.77/248.82 result: MeasureResult(costs=(0.023703824333333335,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.8923025131225586, timestamp=1679084121.846795) [('tile_f', [-1, 8, 1, 8]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 2, 1]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,5814739
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 4, 64]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 16, 8]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,8628608
+ No: 20 GFLOPS: 70.04/151.48 result: MeasureResult(costs=(0.0033050880416666665,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.5203571319580078, timestamp=1679085255.2458742) [('tile_f', [-1, 2, 2, 8]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 16, 1]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,5243404
@@ -2172,9 +1836,9 @@ and measure running time.
Finish loading 20 records
Best config:
- [('tile_f', [-1, 2, 8, 4]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 8, 1]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,206482
+ [('tile_f', [-1, 2, 16, 1]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 4, 1]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,1363375
Finish loading 20 records
- Time cost of this operator: 0.001283
+ Time cost of this operator: 0.001930
diff --git a/docs/_sources/how_to/work_with_microtvm/micro_autotune.rst.txt b/docs/_sources/how_to/work_with_microtvm/micro_autotune.rst.txt
index bc288c2483..6c9cf7026e 100644
--- a/docs/_sources/how_to/work_with_microtvm/micro_autotune.rst.txt
+++ b/docs/_sources/how_to/work_with_microtvm/micro_autotune.rst.txt
@@ -360,10 +360,10 @@ Timing the untuned program
########## Build without Autotuning ##########
Node Name Ops Time(us) Time(%) Shape Inputs Outputs Measurements(us)
--------- --- -------- ------- ----- ------ ------- ----------------
- tvmgen_default_fused_nn_contrib_conv2d_NCHWc tvmgen_default_fused_nn_contrib_conv2d_NCHWc 315.6 98.74 (1, 2, 10, 10, 3) 2 1 [315.6]
- tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 3.06 0.957 (1, 6, 10, 10) 1 1 [3.06]
- tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.967 0.303 (1, 1, 10, 10, 3) 1 1 [0.967]
- Total_time - 319.627 - - - - -
+ tvmgen_default_fused_nn_contrib_conv2d_NCHWc tvmgen_default_fused_nn_contrib_conv2d_NCHWc 316.6 98.736 (1, 2, 10, 10, 3) 2 1 [316.6]
+ tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 3.063 0.955 (1, 6, 10, 10) 1 1 [3.063]
+ tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.99 0.309 (1, 1, 10, 10, 3) 1 1 [0.99]
+ Total_time - 320.653 - - - - -
@@ -428,10 +428,10 @@ Timing the tuned program
########## Build with Autotuning ##########
Node Name Ops Time(us) Time(%) Shape Inputs Outputs Measurements(us)
--------- --- -------- ------- ----- ------ ------- ----------------
- tvmgen_default_fused_nn_contrib_conv2d_NCHWc tvmgen_default_fused_nn_contrib_conv2d_NCHWc 103.0 97.522 (1, 6, 10, 10, 1) 2 1 [103.0]
- tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 1.767 1.673 (1, 6, 10, 10) 1 1 [1.767]
- tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.85 0.805 (1, 3, 10, 10, 1) 1 1 [0.85]
- Total_time - 105.617 - - - - -
+ tvmgen_default_fused_nn_contrib_conv2d_NCHWc tvmgen_default_fused_nn_contrib_conv2d_NCHWc 100.7 97.351 (1, 6, 10, 10, 1) 2 1 [100.7]
+ tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 1.771 1.712 (1, 6, 10, 10) 1 1 [1.771]
+ tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.969 0.937 (1, 1, 10, 10, 3) 1 1 [0.969]
+ Total_time - 103.44 - - - - -
@@ -439,7 +439,7 @@ Timing the tuned program
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 22.329 seconds)
+ **Total running time of the script:** ( 1 minutes 19.674 seconds)
.. _sphx_glr_download_how_to_work_with_microtvm_micro_autotune.py:
diff --git a/docs/_sources/how_to/work_with_microtvm/micro_pytorch.rst.txt b/docs/_sources/how_to/work_with_microtvm/micro_pytorch.rst.txt
index 3acb436998..cd462c4f32 100644
--- a/docs/_sources/how_to/work_with_microtvm/micro_pytorch.rst.txt
+++ b/docs/_sources/how_to/work_with_microtvm/micro_pytorch.rst.txt
@@ -118,7 +118,7 @@ download a cat image and preprocess it to use as the model input.
/venv/apache-tvm-py3.7/lib/python3.7/site-packages/torch/ao/quantization/utils.py:281: UserWarning: must run observer before calling calculate_qparams. Returning default values.
"must run observer before calling calculate_qparams. " +
Downloading: "https://download.pytorch.org/models/quantized/mobilenet_v2_qnnpack_37f702c5.pth" to /workspace/.cache/torch/hub/checkpoints/mobilenet_v2_qnnpack_37f702c5.pth
-
0%| | 0.00/3.42M [00:00<?, ?B/s]
61%|###### | 2.09M/3.42M [00:00<00:00, 10.1MB/s]
100%|##########| 3.42M/3.42M [00:00<00:00, 16.0MB/s]
+
0%| | 0.00/3.42M [00:00<?, ?B/s]
100%|##########| 3.42M/3.42M [00:00<00:00, 160MB/s]
/workspace/python/tvm/relay/frontend/pytorch_utils.py:47: DeprecationWarning: distutils Version classes are deprecated. Use packaging.version instead.
return LooseVersion(torch_ver) > ver
/venv/apache-tvm-py3.7/lib/python3.7/site-packages/setuptools/_distutils/version.py:346: DeprecationWarning: distutils Version classes are deprecated. Use packaging.version instead.
@@ -324,7 +324,7 @@ Look up prediction top 1 index in 1000 class synset.
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 21.807 seconds)
+ **Total running time of the script:** ( 1 minutes 16.632 seconds)
.. _sphx_glr_download_how_to_work_with_microtvm_micro_pytorch.py:
diff --git a/docs/_sources/how_to/work_with_microtvm/micro_train.rst.txt b/docs/_sources/how_to/work_with_microtvm/micro_train.rst.txt
index 5d6f8b3dec..a41c3bd330 100644
--- a/docs/_sources/how_to/work_with_microtvm/micro_train.rst.txt
+++ b/docs/_sources/how_to/work_with_microtvm/micro_train.rst.txt
@@ -218,7 +218,7 @@ take about **2 minutes** to download the Stanford Cars, while COCO 2017 validati
.. code-block:: none
- '/tmp/tmp9sbi5qof/images/random'
+ '/tmp/tmpk8njacwx/images/random'
@@ -309,7 +309,7 @@ objects to other stuff? We can display some examples from our datasets using ``m
.. image-sg:: /how_to/work_with_microtvm/images/sphx_glr_micro_train_001.png
- :alt: [0.0, 1.0], [1.0, 0.0], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0], [1.0, 0.0], [1.0, 0.0], [1.0, 0.0], [1.0, 0.0]
+ :alt: [0.0, 1.0], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0], [1.0, 0.0], [1.0, 0.0], [0.0, 1.0], [1.0, 0.0], [0.0, 1.0], [1.0, 0.0]
:srcset: /how_to/work_with_microtvm/images/sphx_glr_micro_train_001.png
:class: sphx-glr-single-img
@@ -318,8 +318,8 @@ objects to other stuff? We can display some examples from our datasets using ``m
.. code-block:: none
- /tmp/tmp9sbi5qof/images/target contains 8144 images
- /tmp/tmp9sbi5qof/images/random contains 5000 images
+ /tmp/tmpk8njacwx/images/target contains 8144 images
+ /tmp/tmpk8njacwx/images/random contains 5000 images
@@ -494,13 +494,13 @@ the time on our validation set).
.. code-block:: none
Epoch 1/3
- 328/328 - 48s - loss: 0.2123 - accuracy: 0.9259 - val_loss: 0.1314 - val_accuracy: 0.9607 - 48s/epoch - 146ms/step
+ 328/328 - 47s - loss: 0.2252 - accuracy: 0.9217 - val_loss: 0.2626 - val_accuracy: 0.8992 - 47s/epoch - 143ms/step
Epoch 2/3
- 328/328 - 43s - loss: 0.0985 - accuracy: 0.9648 - val_loss: 0.1363 - val_accuracy: 0.9562 - 43s/epoch - 132ms/step
+ 328/328 - 43s - loss: 0.0975 - accuracy: 0.9618 - val_loss: 0.0954 - val_accuracy: 0.9622 - 43s/epoch - 132ms/step
Epoch 3/3
- 328/328 - 42s - loss: 0.0669 - accuracy: 0.9745 - val_loss: 0.0953 - val_accuracy: 0.9675 - 42s/epoch - 128ms/step
+ 328/328 - 43s - loss: 0.0641 - accuracy: 0.9760 - val_loss: 0.1629 - val_accuracy: 0.9366 - 43s/epoch - 131ms/step
- <keras.callbacks.History object at 0x7fb002be9450>
+ <keras.callbacks.History object at 0x7f3be8b52f90>
@@ -861,7 +861,7 @@ Arduino tutorial for how to do that `on GitHub <https://github.com/guberti/tvm-a
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 4 minutes 42.728 seconds)
+ **Total running time of the script:** ( 4 minutes 35.152 seconds)
.. _sphx_glr_download_how_to_work_with_microtvm_micro_train.py:
diff --git a/docs/_sources/how_to/work_with_microtvm/sg_execution_times.rst.txt b/docs/_sources/how_to/work_with_microtvm/sg_execution_times.rst.txt
index e6e28aeee6..bed04d4a23 100644
--- a/docs/_sources/how_to/work_with_microtvm/sg_execution_times.rst.txt
+++ b/docs/_sources/how_to/work_with_microtvm/sg_execution_times.rst.txt
@@ -5,24 +5,24 @@
Computation times
=================
-**07:54.442** total execution time for **how_to_work_with_microtvm** files:
+**07:36.586** total execution time for **how_to_work_with_microtvm** files:
+-----------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_microtvm_micro_train.py` (``micro_train.py``) | 04:42.728 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_microtvm_micro_train.py` (``micro_train.py``) | 04:35.152 | 0.0 MB |
+-----------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_microtvm_micro_autotune.py` (``micro_autotune.py``) | 01:22.329 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_microtvm_micro_autotune.py` (``micro_autotune.py``) | 01:19.674 | 0.0 MB |
+-----------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_microtvm_micro_pytorch.py` (``micro_pytorch.py``) | 01:21.807 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_microtvm_micro_pytorch.py` (``micro_pytorch.py``) | 01:16.632 | 0.0 MB |
+-----------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_microtvm_micro_aot.py` (``micro_aot.py``) | 00:10.649 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_microtvm_micro_aot.py` (``micro_aot.py``) | 00:09.934 | 0.0 MB |
+-----------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_microtvm_micro_custom_ide.py` (``micro_custom_ide.py``) | 00:09.153 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_microtvm_micro_custom_ide.py` (``micro_custom_ide.py``) | 00:08.059 | 0.0 MB |
+-----------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_microtvm_micro_tflite.py` (``micro_tflite.py``) | 00:07.776 | 0.0 MB |
-+-----------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_microtvm_micro_tvmc.py` (``micro_tvmc.py``) | 00:00.000 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_microtvm_micro_tflite.py` (``micro_tflite.py``) | 00:07.135 | 0.0 MB |
+-----------------------------------------------------------------------------------------+-----------+--------+
| :ref:`sphx_glr_how_to_work_with_microtvm_micro_ethosu.py` (``micro_ethosu.py``) | 00:00.000 | 0.0 MB |
+-----------------------------------------------------------------------------------------+-----------+--------+
+| :ref:`sphx_glr_how_to_work_with_microtvm_micro_tvmc.py` (``micro_tvmc.py``) | 00:00.000 | 0.0 MB |
++-----------------------------------------------------------------------------------------+-----------+--------+
| :ref:`sphx_glr_how_to_work_with_microtvm_micro_mlperftiny.py` (``micro_mlperftiny.py``) | 00:00.000 | 0.0 MB |
+-----------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/how_to/work_with_relay/sg_execution_times.rst.txt b/docs/_sources/how_to/work_with_relay/sg_execution_times.rst.txt
index d7951f1cf6..cefb9f21d0 100644
--- a/docs/_sources/how_to/work_with_relay/sg_execution_times.rst.txt
+++ b/docs/_sources/how_to/work_with_relay/sg_execution_times.rst.txt
@@ -5,14 +5,14 @@
Computation times
=================
-**00:48.787** total execution time for **how_to_work_with_relay** files:
+**00:45.382** total execution time for **how_to_work_with_relay** files:
+----------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_relay_using_pipeline_executor.py` (``using_pipeline_executor.py``) | 00:36.436 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_relay_using_pipeline_executor.py` (``using_pipeline_executor.py``) | 00:33.542 | 0.0 MB |
+----------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_relay_using_external_lib.py` (``using_external_lib.py``) | 00:10.647 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_relay_using_external_lib.py` (``using_external_lib.py``) | 00:10.234 | 0.0 MB |
+----------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_relay_build_gcn.py` (``build_gcn.py``) | 00:01.697 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_relay_build_gcn.py` (``build_gcn.py``) | 00:01.600 | 0.0 MB |
+----------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_relay_using_relay_viz.py` (``using_relay_viz.py``) | 00:00.007 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_relay_using_relay_viz.py` (``using_relay_viz.py``) | 00:00.006 | 0.0 MB |
+----------------------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/how_to/work_with_schedules/intrin_math.rst.txt b/docs/_sources/how_to/work_with_schedules/intrin_math.rst.txt
index 923e78e5e1..c3a441b413 100644
--- a/docs/_sources/how_to/work_with_schedules/intrin_math.rst.txt
+++ b/docs/_sources/how_to/work_with_schedules/intrin_math.rst.txt
@@ -264,7 +264,7 @@ The following example customizes CUDA lowering rule for :code:`exp`.
.. code-block:: none
- <function my_cuda_math_rule at 0x7faea0657cb0>
+ <function my_cuda_math_rule at 0x7f3a9e102cb0>
diff --git a/docs/_sources/how_to/work_with_schedules/sg_execution_times.rst.txt b/docs/_sources/how_to/work_with_schedules/sg_execution_times.rst.txt
index 109859e4d3..c2ed4a0c1c 100644
--- a/docs/_sources/how_to/work_with_schedules/sg_execution_times.rst.txt
+++ b/docs/_sources/how_to/work_with_schedules/sg_execution_times.rst.txt
@@ -5,22 +5,22 @@
Computation times
=================
-**00:08.875** total execution time for **how_to_work_with_schedules** files:
+**00:08.704** total execution time for **how_to_work_with_schedules** files:
+------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_intrin_math.py` (``intrin_math.py``) | 00:06.132 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_intrin_math.py` (``intrin_math.py``) | 00:06.180 | 0.0 MB |
+------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_tensorize.py` (``tensorize.py``) | 00:01.211 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_tensorize.py` (``tensorize.py``) | 00:01.138 | 0.0 MB |
+------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_reduction.py` (``reduction.py``) | 00:00.626 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_reduction.py` (``reduction.py``) | 00:00.577 | 0.0 MB |
+------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_scan.py` (``scan.py``) | 00:00.616 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_scan.py` (``scan.py``) | 00:00.556 | 0.0 MB |
+------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_extern_op.py` (``extern_op.py``) | 00:00.120 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_extern_op.py` (``extern_op.py``) | 00:00.116 | 0.0 MB |
+------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_tedd.py` (``tedd.py``) | 00:00.078 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_tedd.py` (``tedd.py``) | 00:00.060 | 0.0 MB |
+------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_schedule_primitives.py` (``schedule_primitives.py``) | 00:00.061 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_schedule_primitives.py` (``schedule_primitives.py``) | 00:00.051 | 0.0 MB |
+------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_tuple_inputs.py` (``tuple_inputs.py``) | 00:00.031 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_tuple_inputs.py` (``tuple_inputs.py``) | 00:00.026 | 0.0 MB |
+------------------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/topic/vta/tutorials/autotvm/sg_execution_times.rst.txt b/docs/_sources/topic/vta/tutorials/autotvm/sg_execution_times.rst.txt
index f1ec71f19c..f2b4983038 100644
--- a/docs/_sources/topic/vta/tutorials/autotvm/sg_execution_times.rst.txt
+++ b/docs/_sources/topic/vta/tutorials/autotvm/sg_execution_times.rst.txt
@@ -5,10 +5,10 @@
Computation times
=================
-**00:32.919** total execution time for **topic_vta_tutorials_autotvm** files:
+**00:30.632** total execution time for **topic_vta_tutorials_autotvm** files:
+---------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_autotvm_tune_relay_vta.py` (``tune_relay_vta.py``) | 00:32.912 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_autotvm_tune_relay_vta.py` (``tune_relay_vta.py``) | 00:30.626 | 0.0 MB |
+---------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_autotvm_tune_alu_vta.py` (``tune_alu_vta.py``) | 00:00.007 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_autotvm_tune_alu_vta.py` (``tune_alu_vta.py``) | 00:00.006 | 0.0 MB |
+---------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/topic/vta/tutorials/frontend/deploy_classification.rst.txt b/docs/_sources/topic/vta/tutorials/frontend/deploy_classification.rst.txt
index 9d10b2be4e..9c054f0c6d 100644
--- a/docs/_sources/topic/vta/tutorials/frontend/deploy_classification.rst.txt
+++ b/docs/_sources/topic/vta/tutorials/frontend/deploy_classification.rst.txt
@@ -293,7 +293,7 @@ The compilation steps are:
DeprecationWarning,
/workspace/vta/tutorials/frontend/deploy_classification.py:213: DeprecationWarning: legacy graph executor behavior of producing json / lib / params will be removed in the next release. Please see documents of tvm.contrib.graph_executor.GraphModule for the new recommended usage.
relay_prog, target=tvm.target.Target(target, host=env.target_host), params=params
- resnet18_v1 inference graph built in 37.02s!
+ resnet18_v1 inference graph built in 32.43s!
diff --git a/docs/_sources/topic/vta/tutorials/frontend/deploy_detection.rst.txt b/docs/_sources/topic/vta/tutorials/frontend/deploy_detection.rst.txt
index c23ba9e1c2..7451ae647e 100644
--- a/docs/_sources/topic/vta/tutorials/frontend/deploy_detection.rst.txt
+++ b/docs/_sources/topic/vta/tutorials/frontend/deploy_detection.rst.txt
@@ -337,7 +337,7 @@ The compilation steps are:
/workspace/python/tvm/relay/build_module.py:348: DeprecationWarning: Please use input parameter mod (tvm.IRModule) instead of deprecated parameter mod (tvm.relay.function.Function)
DeprecationWarning,
- yolov3-tiny inference graph built in 22.67s!
+ yolov3-tiny inference graph built in 22.22s!
diff --git a/docs/_sources/topic/vta/tutorials/frontend/sg_execution_times.rst.txt b/docs/_sources/topic/vta/tutorials/frontend/sg_execution_times.rst.txt
index bcfb592653..133cb75fd2 100644
--- a/docs/_sources/topic/vta/tutorials/frontend/sg_execution_times.rst.txt
+++ b/docs/_sources/topic/vta/tutorials/frontend/sg_execution_times.rst.txt
@@ -5,10 +5,10 @@
Computation times
=================
-**01:43.496** total execution time for **topic_vta_tutorials_frontend** files:
+**01:37.955** total execution time for **topic_vta_tutorials_frontend** files:
+------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_frontend_deploy_classification.py` (``deploy_classification.py``) | 00:53.842 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_frontend_deploy_classification.py` (``deploy_classification.py``) | 00:49.103 | 0.0 MB |
+------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_frontend_deploy_detection.py` (``deploy_detection.py``) | 00:49.655 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_frontend_deploy_detection.py` (``deploy_detection.py``) | 00:48.852 | 0.0 MB |
+------------------------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/topic/vta/tutorials/optimize/sg_execution_times.rst.txt b/docs/_sources/topic/vta/tutorials/optimize/sg_execution_times.rst.txt
index a1d843b1f0..d0d503846f 100644
--- a/docs/_sources/topic/vta/tutorials/optimize/sg_execution_times.rst.txt
+++ b/docs/_sources/topic/vta/tutorials/optimize/sg_execution_times.rst.txt
@@ -5,10 +5,10 @@
Computation times
=================
-**00:03.568** total execution time for **topic_vta_tutorials_optimize** files:
+**00:03.153** total execution time for **topic_vta_tutorials_optimize** files:
+--------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_optimize_convolution_opt.py` (``convolution_opt.py``) | 00:03.068 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_optimize_convolution_opt.py` (``convolution_opt.py``) | 00:02.700 | 0.0 MB |
+--------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_optimize_matrix_multiply_opt.py` (``matrix_multiply_opt.py``) | 00:00.500 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_optimize_matrix_multiply_opt.py` (``matrix_multiply_opt.py``) | 00:00.453 | 0.0 MB |
+--------------------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/topic/vta/tutorials/sg_execution_times.rst.txt b/docs/_sources/topic/vta/tutorials/sg_execution_times.rst.txt
index ea37505614..bf09786bf3 100644
--- a/docs/_sources/topic/vta/tutorials/sg_execution_times.rst.txt
+++ b/docs/_sources/topic/vta/tutorials/sg_execution_times.rst.txt
@@ -5,10 +5,10 @@
Computation times
=================
-**00:00.792** total execution time for **topic_vta_tutorials** files:
+**00:00.777** total execution time for **topic_vta_tutorials** files:
+---------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_matrix_multiply.py` (``matrix_multiply.py``) | 00:00.417 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_matrix_multiply.py` (``matrix_multiply.py``) | 00:00.402 | 0.0 MB |
+---------------------------------------------------------------------------------+-----------+--------+
| :ref:`sphx_glr_topic_vta_tutorials_vta_get_started.py` (``vta_get_started.py``) | 00:00.375 | 0.0 MB |
+---------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/tutorial/auto_scheduler_matmul_x86.rst.txt b/docs/_sources/tutorial/auto_scheduler_matmul_x86.rst.txt
index e46195422e..6d7a7205ee 100644
--- a/docs/_sources/tutorial/auto_scheduler_matmul_x86.rst.txt
+++ b/docs/_sources/tutorial/auto_scheduler_matmul_x86.rst.txt
@@ -318,7 +318,7 @@ We build the binary and check its correctness and performance.
.. code-block:: none
- Execution time of this operator: 96.930 ms
+ Execution time of this operator: 97.029 ms
@@ -434,7 +434,7 @@ operations.
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 48.356 seconds)
+ **Total running time of the script:** ( 1 minutes 41.541 seconds)
.. _sphx_glr_download_tutorial_auto_scheduler_matmul_x86.py:
diff --git a/docs/_sources/tutorial/autotvm_matmul_x86.rst.txt b/docs/_sources/tutorial/autotvm_matmul_x86.rst.txt
index 4d4bf7f06b..49da074f5c 100644
--- a/docs/_sources/tutorial/autotvm_matmul_x86.rst.txt
+++ b/docs/_sources/tutorial/autotvm_matmul_x86.rst.txt
@@ -454,16 +454,16 @@ reduce variance, we take 5 measurements and average them.
waiting for device...
device available
Get devices for measurement successfully!
- No: 1 GFLOPS: 4.13/4.13 result: MeasureResult(costs=(0.0650117556,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.3118185997009277, timestamp=1679082375.4901636) [('tile_y', [-1, 4]), ('tile_x', [-1, 16])],None,42
- No: 2 GFLOPS: 2.91/4.13 result: MeasureResult(costs=(0.0921846846,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.737344741821289, timestamp=1679082377.2266726) [('tile_y', [-1, 8]), ('tile_x', [-1, 4])],None,23
- No: 3 GFLOPS: 11.48/11.48 result: MeasureResult(costs=(0.023378502399999997,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.6373653411865234, timestamp=1679082379.0981555) [('tile_y', [-1, 8]), ('tile_x', [-1, 512])],None,93
- No: 4 GFLOPS: 1.16/11.48 result: MeasureResult(costs=(0.23051691700000002,), error_no=MeasureErrorNo.NO_ERROR, all_cost=3.924243927001953, timestamp=1679082383.050405) [('tile_y', [-1, 1]), ('tile_x', [-1, 1])],None,0
- No: 5 GFLOPS: 11.12/11.48 result: MeasureResult(costs=(0.0241488672,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.6910970211029053, timestamp=1679082383.8645668) [('tile_y', [-1, 16]), ('tile_x', [-1, 256])],None,84
- No: 6 GFLOPS: 9.22/11.48 result: MeasureResult(costs=(0.0291021168,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.7192752361297607, timestamp=1679082385.8878448) [('tile_y', [-1, 2]), ('tile_x', [-1, 64])],None,61
- No: 7 GFLOPS: 8.99/11.48 result: MeasureResult(costs=(0.029863789800000002,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.7994594573974609, timestamp=1679082387.8821983) [('tile_y', [-1, 512]), ('tile_x', [-1, 64])],None,69
- No: 8 GFLOPS: 3.00/11.48 result: MeasureResult(costs=(0.0893364948,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.6753859519958496, timestamp=1679082389.5679548) [('tile_y', [-1, 4]), ('tile_x', [-1, 4])],None,22
- No: 9 GFLOPS: 8.97/11.48 result: MeasureResult(costs=(0.029928205800000003,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.697211503982544, timestamp=1679082390.3920758) [('tile_y', [-1, 1]), ('tile_x', [-1, 32])],None,50
- No: 10 GFLOPS: 2.62/11.48 result: MeasureResult(costs=(0.10263271139999999,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.8616254329681396, timestamp=1679082392.2895982) [('tile_y', [-1, 512]), ('tile_x', [-1, 16])],None,49
+ No: 1 GFLOPS: 14.82/14.82 result: MeasureResult(costs=(0.0181141654,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.8080611228942871, timestamp=1679083606.7423198) [('tile_y', [-1, 64]), ('tile_x', [-1, 64])],None,66
+ No: 2 GFLOPS: 4.53/14.82 result: MeasureResult(costs=(0.0592012732,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.2131140232086182, timestamp=1679083609.1746023) [('tile_y', [-1, 8]), ('tile_x', [-1, 16])],None,43
+ No: 3 GFLOPS: 3.06/14.82 result: MeasureResult(costs=(0.0877241708,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.6729583740234375, timestamp=1679083610.8397174) [('tile_y', [-1, 2]), ('tile_x', [-1, 16])],None,41
+ No: 4 GFLOPS: 1.26/14.82 result: MeasureResult(costs=(0.212710966,), error_no=MeasureErrorNo.NO_ERROR, all_cost=3.6374735832214355, timestamp=1679083615.721666) [('tile_y', [-1, 2]), ('tile_x', [-1, 1])],None,1
+ No: 5 GFLOPS: 3.34/14.82 result: MeasureResult(costs=(0.0803117228,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.5404706001281738, timestamp=1679083617.386324) [('tile_y', [-1, 64]), ('tile_x', [-1, 8])],None,36
+ No: 6 GFLOPS: 9.47/14.82 result: MeasureResult(costs=(0.0283350406,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.6895656585693359, timestamp=1679083619.3318534) [('tile_y', [-1, 1]), ('tile_x', [-1, 512])],None,90
+ No: 7 GFLOPS: 0.51/14.82 result: MeasureResult(costs=(0.527877173,), error_no=MeasureErrorNo.NO_ERROR, all_cost=8.685568809509277, timestamp=1679083628.0369303) [('tile_y', [-1, 128]), ('tile_x', [-1, 1])],None,7
+ No: 8 GFLOPS: 2.72/14.82 result: MeasureResult(costs=(0.0985765196,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.8126258850097656, timestamp=1679083629.86775) [('tile_y', [-1, 1]), ('tile_x', [-1, 16])],None,40
+ No: 9 GFLOPS: 11.40/14.82 result: MeasureResult(costs=(0.0235393256,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.5920979976654053, timestamp=1679083630.5899296) [('tile_y', [-1, 4]), ('tile_x', [-1, 512])],None,92
+ No: 10 GFLOPS: 10.67/14.82 result: MeasureResult(costs=(0.0251462134,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.6306076049804688, timestamp=1679083631.2539246) [('tile_y', [-1, 2]), ('tile_x', [-1, 256])],None,81
diff --git a/docs/_sources/tutorial/autotvm_relay_x86.rst.txt b/docs/_sources/tutorial/autotvm_relay_x86.rst.txt
index 8fa2619504..12b4248a91 100644
--- a/docs/_sources/tutorial/autotvm_relay_x86.rst.txt
+++ b/docs/_sources/tutorial/autotvm_relay_x86.rst.txt
@@ -311,7 +311,7 @@ standard deviation.
.. code-block:: none
- {'mean': 514.94100076, 'median': 514.3353281999964, 'std': 2.3532936330814875}
+ {'mean': 514.6666772300001, 'median': 515.163708800003, 'std': 2.3164496047830663}
@@ -545,31 +545,31 @@ the tuning data to.
.. code-block:: none
-
[Task 1/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 1/25] Current/Best: 9.32/ 19.08 GFLOPS | Progress: (4/20) | 11.60 s
[Task 1/25] Current/Best: 12.79/ 19.08 GFLOPS | Progress: (8/20) | 15.40 s
[Task 1/25] Current/Best: 6.83/ 19.08 GFLOPS | Progress: (12/20) | 18.36 s
[Task 1/25] Current/Best: 6.07/ 23.16 GFLOPS | Progress: (16/20) | 21.06 s
[Task 1/25] Current/Best: 9.77/ 23.16 GFLOPS | Progress: (20/20) | 23.90 s Done.
-
[Task 2/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 2/25] Current/Best: 11.91/ 19.43 GFLOPS | Progress: (4/20) | 4.21 s
[Task 2/25] Current/Best: 6.80/ 19.43 GFLOPS | Progress: (8/20) | 6.33 s
[Task 2/25] Current/Best: 18.51/ 21.37 GFLOPS | Progress: (12/20) | 7.98 s
[Task 2/25] Current/Best: 12.94/ 21.37 GFLOPS | Progress: (16/20) | 9.35 s
[Task 2/25] Current/Best: 15.27/ 21.37 GFLOPS | Progress: (20/20) | 12.43 s Done.
-
[Task 3/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 3/25] Current/Best: 12.48/ 18.20 GFLOPS | Progress: (4/20) | 5.46 s
[Task 3/25] Current/Best: 20.27/ 20.27 GFLOPS | Progress: (8/20) | 7.62 s
[Task 3/25] Current/Best: 18.49/ 20.27 GFLOPS | Progress: (12/20) | 10.51 s
[Task 3/25] Current/Best: 19.43/ 20.27 GFLOPS | Progress: (16/20) | 12.82 s
[Task 3/25] Current/Best: 17.64/ 20.27 GFLOPS | Progress: (20/20) | 15.24 s Done.
-
[Task 4/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 4/25] Current/Best: 4.76/ 14.02 GFLOPS | Progress: (4/20) | 10.21 s
[Task 4/25] Current/Best: 11.55/ 14.02 GFLOPS | Progress: (8/20) | 13.33 s
[Task 4/25] Current/Best: 18.17/ 18.17 GFLOPS | Progress: (12/20) | 15.62 s
[Task 4/25] Current/Best: 16.23/ 18.17 GFLOPS | Progress: (16/20) | 17.59 s
[Task 4/25] Current/Best: 14.90/ 18.47 GFLOPS | Progress: (20/20) | 19.66 s Done.
-
[Task 5/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 5/25] Current/Best: 4.45/ 15.66 GFLOPS | Progress: (4/20) | 4.81 s
[Task 5/25] Current/Best: 2.79/ 18.19 GFLOPS | Progress: (8/20) | 7.43 s
[Task 5/25] Current/Best: 9.39/ 18.59 GFLOPS | Progress: (12/20) | 9.18 s
[Task 5/25] Current/Best: 12.15/ 18.59 GFLOPS | Progress: (16/20) | 11.28 s
[Task 5/25] Current/Best: 10.96/ 18.59 GFLOPS | Progress: (20/20) | 13.98 s Done.
-
[Task 6/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 6/25] Current/Best: 5.73/ 16.36 GFLOPS | Progress: (4/20) | 6.10 s
[Task 6/25] Current/Best: 16.54/ 16.54 GFLOPS | Progress: (8/20) | 9.90 s
[Task 6/25] Current/Best: 11.06/ 17.24 GFLOPS | Progress: (12/20) | 16.48 s
[Task 6/25] Current/Best: 11.45/ 20.41 GFLOPS | Progress: (16/20) | 19.14 s
[Task 6/25] Current/Best: 13.28/ 20.41 GFLOPS | Progress: (20/20) | 21.94 s Done.
-
[Task 7/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 7/25] Current/Best: 12.46/ 17.18 GFLOPS | Progress: (4/20) | 6.35 s
[Task 7/25] Current/Best: 11.85/ 17.72 GFLOPS | Progress: (8/20) | 10.66 s
[Task 7/25] Current/Best: 12.87/ 17.72 GFLOPS | Progress: (12/20) | 12.97 s
[Task 7/25] Current/Best: 19.59/ 20.88 GFLOPS | Progress: (16/20) | 15.26 s
[Task 7/25] Current/Best: 12.98/ 20.88 GFLOPS | Progress: (20/20) | 17.81 s Done.
-
[Task 8/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 8/25] Current/Best: 7.07/ 20.30 GFLOPS | Progress: (4/20) | 4.90 s
[Task 8/25] Current/Best: 3.27/ 20.30 GFLOPS | Progress: (8/20) | 14.18 s
[Task 8/25] Current/Best: 12.21/ 20.30 GFLOPS | Progress: (12/20) | 25.70 s
[Task 8/25] Current/Best: 3.87/ 20.30 GFLOPS | Progress: (16/20) | 31.60 s
[Task 8/25] Current/Best: 3.66/ 20.30 GFLOPS | Progress: (20/20) | 35.13 s
[Task 9/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 9/25] Current/Best: 12.61/ 17.58 GFLOPS | Progress: (4/20) | 5.55 s
[Task 9/25] Current/Best: 15.70/ 17.58 GFLOPS | Progress: (8/20) | 7.74 s Done.
-
[Task 9/25] Current/Best: 4.78/ 19.18 GFLOPS | Progress: (12/20) | 18.87 s
[Task 9/25] Current/Best: 6.55/ 19.18 GFLOPS | Progress: (16/20) | 30.05 s
[Task 9/25] Current/Best: 16.06/ 19.18 GFLOPS | Progress: (20/20) | 33.09 s
[Task 10/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 10/25] Current/Best: 8.01/ 18.14 GFLOPS | Progress: (4/20) | 6.12 s
[Task 10/25] Current/Best: 5.60/ 18.14 GFLOPS | Progress: (8/20) | 9.21 s
[Task 10/25] Current/Best: 11.92/ 18.14 GFLOPS | Progress: (12/20) | 10.93 s
[Task 10/25] Current/Best: 14.38/ 19.10 GFLOPS | Progress: (16/20) | 12.60 s
[Task 10/25] Current/Best: 16.07/ 19.10 GFLOPS | Progress: (20/20) | 14.81 s Done.
-
[Task 11/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 11/25] Current/Best: 8.34/ 8.51 GFLOPS | Progress: (4/20) | 6.41 s
[Task 11/25] Current/Best: 10.87/ 18.46 GFLOPS | Progress: (8/20) | 10.00 s
[Task 11/25] Current/Best: 14.86/ 19.44 GFLOPS | Progress: (12/20) | 12.32 s
[Task 11/25] Current/Best: 11.97/ 19.44 GFLOPS | Progress: (16/20) | 14.98 s
[Task 11/25] Current/Best: 12.59/ 23.21 GFLOPS | Progress: (20/20) | 17.37 s Done.
-
[Task 12/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 12/25] Current/Best: 7.97/ 15.58 GFLOPS | Progress: (4/20) | 8.35 s
[Task 12/25] Current/Best: 5.28/ 15.91 GFLOPS | Progress: (8/20) | 10.64 s
[Task 12/25] Current/Best: 13.76/ 15.91 GFLOPS | Progress: (12/20) | 13.70 s
[Task 12/25] Current/Best: 13.54/ 15.91 GFLOPS | Progress: (16/20) | 16.74 s
[Task 12/25] Current/Best: 11.51/ 20.87 GFLOPS | Progress: (20/20) | 22.27 s Done.
-
[Task 13/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 13/25] Current/Best: 11.70/ 17.90 GFLOPS | Progress: (4/20) | 5.44 s
[Task 13/25] Current/Best: 19.07/ 19.07 GFLOPS | Progress: (8/20) | 7.67 s
[Task 13/25] Current/Best: 5.22/ 19.07 GFLOPS | Progress: (12/20) | 10.19 s
[Task 13/25] Current/Best: 16.73/ 19.85 GFLOPS | Progress: (16/20) | 13.53 s
[Task 13/25] Current/Best: 21.25/ 21.25 GFLOPS | Progress: (20/20) | 17.19 s Done.
-
[Task 14/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 14/25] Current/Best: 3.92/ 15.03 GFLOPS | Progress: (4/20) | 6.07 s
[Task 14/25] Current/Best: 7.11/ 15.03 GFLOPS | Progress: (8/20) | 8.65 s
[Task 14/25] Current/Best: 16.53/ 18.32 GFLOPS | Progress: (12/20) | 12.28 s
[Task 14/25] Current/Best: 19.92/ 19.92 GFLOPS | Progress: (16/20) | 15.28 s
[Task 14/25] Current/Best: 10.85/ 19.92 GFLOPS | Progress: (20/20) | 18.34 s
[Task 15/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 15/25] Current/Best: 13.32/ 13.32 GFLOPS | Progress: (4/20) | 9.12 s
[Task 15/25] Current/Best: 12.73/ 19.34 GFLOPS | Progress: (8/20) | 10.71 s
[Task 15/25] Current/Best: 11.56/ 19.34 GFLOPS | Progress: (12/20) | 13.91 s
[Task 15/25] Current/Best: 12.41/ 19.34 GFLOPS | Progress: (16/20) | 16.58 s
[Task 15/25] Current/Best: 19.31/ 21.60 GFLOPS | Progress: (20/2
0) | 18.48 s Done.
-
[Task 16/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 16/25] Current/Best: 10.11/ 13.91 GFLOPS | Progress: (4/20) | 5.35 s
[Task 16/25] Current/Best: 19.16/ 19.16 GFLOPS | Progress: (8/20) | 8.41 s
[Task 16/25] Current/Best: 5.83/ 19.16 GFLOPS | Progress: (12/20) | 10.91 s
[Task 16/25] Current/Best: 9.49/ 19.16 GFLOPS | Progress: (16/20) | 13.37 s
[Task 16/25] Current/Best: 19.90/ 20.78 GFLOPS | Progress: (20/20) | 14.96 s Done.
-
[Task 17/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 17/25] Current/Best: 7.10/ 22.79 GFLOPS | Progress: (4/20) | 5.36 s Done.
+
[Task 1/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 1/25] Current/Best: 19.22/ 22.90 GFLOPS | Progress: (4/20) | 10.66 s
[Task 1/25] Current/Best: 17.58/ 23.22 GFLOPS | Progress: (8/20) | 14.07 s
[Task 1/25] Current/Best: 16.73/ 23.22 GFLOPS | Progress: (12/20) | 16.64 s
[Task 1/25] Current/Best: 14.65/ 23.22 GFLOPS | Progress: (16/20) | 18.88 s
[Task 1/25] Current/Best: 5.75/ 23.22 GFLOPS | Progress: (20/20) | 23.11 s Done.
+
[Task 2/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 2/25] Current/Best: 10.20/ 17.71 GFLOPS | Progress: (4/20) | 5.77 s
[Task 2/25] Current/Best: 8.65/ 17.71 GFLOPS | Progress: (8/20) | 7.75 s
[Task 2/25] Current/Best: 6.60/ 19.14 GFLOPS | Progress: (12/20) | 9.71 s
[Task 2/25] Current/Best: 10.70/ 19.14 GFLOPS | Progress: (16/20) | 12.62 s
[Task 2/25] Current/Best: 18.75/ 19.14 GFLOPS | Progress: (20/20) | 14.77 s Done.
+
[Task 3/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 3/25] Current/Best: 14.49/ 15.80 GFLOPS | Progress: (4/20) | 5.09 s
[Task 3/25] Current/Best: 18.18/ 18.25 GFLOPS | Progress: (8/20) | 6.99 s
[Task 3/25] Current/Best: 10.30/ 18.25 GFLOPS | Progress: (12/20) | 9.84 s
[Task 3/25] Current/Best: 10.47/ 18.25 GFLOPS | Progress: (16/20) | 12.00 s
[Task 3/25] Current/Best: 13.97/ 18.25 GFLOPS | Progress: (20/20) | 14.30 s Done.
+
[Task 4/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 4/25] Current/Best: 13.07/ 22.29 GFLOPS | Progress: (4/20) | 4.68 s
[Task 4/25] Current/Best: 17.79/ 22.29 GFLOPS | Progress: (8/20) | 6.61 s
[Task 4/25] Current/Best: 19.40/ 22.29 GFLOPS | Progress: (12/20) | 10.97 s
[Task 4/25] Current/Best: 7.34/ 22.29 GFLOPS | Progress: (16/20) | 13.51 s
[Task 4/25] Current/Best: 12.40/ 22.29 GFLOPS | Progress: (20/20) | 15.56 s Done.
+
[Task 5/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 5/25] Current/Best: 13.10/ 20.01 GFLOPS | Progress: (4/20) | 5.93 s
[Task 5/25] Current/Best: 16.87/ 20.01 GFLOPS | Progress: (8/20) | 7.81 s
[Task 5/25] Current/Best: 16.04/ 20.01 GFLOPS | Progress: (12/20) | 10.07 s
[Task 5/25] Current/Best: 6.73/ 21.87 GFLOPS | Progress: (16/20) | 12.23 s
[Task 5/25] Current/Best: 13.02/ 21.87 GFLOPS | Progress: (20/20) | 14.67 s Done.
+
[Task 6/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 6/25] Current/Best: 6.00/ 19.48 GFLOPS | Progress: (4/20) | 5.26 s
[Task 6/25] Current/Best: 8.47/ 19.48 GFLOPS | Progress: (8/20) | 7.81 s
[Task 6/25] Current/Best: 5.23/ 19.48 GFLOPS | Progress: (12/20) | 10.96 s
[Task 6/25] Current/Best: 13.80/ 19.48 GFLOPS | Progress: (16/20) | 13.60 s
[Task 6/25] Current/Best: 2.63/ 19.48 GFLOPS | Progress: (20/20) | 17.47 s Done.
+
[Task 7/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 7/25] Current/Best: 11.08/ 17.55 GFLOPS | Progress: (4/20) | 5.16 s
[Task 7/25] Current/Best: 8.49/ 17.55 GFLOPS | Progress: (8/20) | 7.76 s
[Task 7/25] Current/Best: 13.00/ 17.55 GFLOPS | Progress: (12/20) | 11.21 s
[Task 7/25] Current/Best: 8.26/ 17.55 GFLOPS | Progress: (16/20) | 13.75 s
[Task 7/25] Current/Best: 18.38/ 18.81 GFLOPS | Progress: (20/20) | 16.06 s Done.
+
[Task 8/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 8/25] Current/Best: 11.69/ 15.56 GFLOPS | Progress: (4/20) | 14.63 s
[Task 8/25] Current/Best: 13.21/ 18.78 GFLOPS | Progress: (8/20) | 18.72 s
[Task 8/25] Current/Best: 2.80/ 18.78 GFLOPS | Progress: (12/20) | 21.83 s
[Task 8/25] Current/Best: 3.34/ 18.78 GFLOPS | Progress: (16/20) | 33.98 s
[Task 8/25] Current/Best: 14.08/ 18.78 GFLOPS | Progress: (20/20) | 37.50 s
[Task 9/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 9/25] Current/Best: 11.99/ 18.01 GFLOPS | Progress: (4/20) | 8.44 s
[Task 9/25] Current/Best: 9.95/ 18.01 GFLOPS | Progress: (8/20) | 12.07 s Done.
+
[Task 9/25] Current/Best: 9.07/ 18.01 GFLOPS | Progress: (12/20) | 19.84 s
[Task 9/25] Current/Best: 20.27/ 20.27 GFLOPS | Progress: (16/20) | 26.59 s
[Task 9/25] Current/Best: 15.97/ 22.87 GFLOPS | Progress: (20/20) | 29.35 s Done.
+
[Task 10/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 10/25] Current/Best: 15.63/ 15.63 GFLOPS | Progress: (4/20) | 5.38 s
[Task 10/25] Current/Best: 10.13/ 16.57 GFLOPS | Progress: (8/20) | 7.25 s
[Task 10/25] Current/Best: 3.52/ 17.50 GFLOPS | Progress: (12/20) | 9.67 s
[Task 10/25] Current/Best: 8.48/ 17.50 GFLOPS | Progress: (16/20) | 11.61 s
[Task 10/25] Current/Best: 13.86/ 17.62 GFLOPS | Progress: (20/20) | 13.58 s Done.
+
[Task 11/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 11/25] Current/Best: 5.53/ 11.98 GFLOPS | Progress: (4/20) | 6.13 s
[Task 11/25] Current/Best: 14.44/ 18.47 GFLOPS | Progress: (8/20) | 8.68 s
[Task 11/25] Current/Best: 13.35/ 18.47 GFLOPS | Progress: (12/20) | 12.00 s
[Task 11/25] Current/Best: 11.46/ 18.47 GFLOPS | Progress: (16/20) | 15.56 s
[Task 11/25] Current/Best: 17.57/ 21.78 GFLOPS | Progress: (20/20) | 18.62 s Done.
+
[Task 12/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 12/25] Current/Best: 8.48/ 14.67 GFLOPS | Progress: (4/20) | 6.51 s
[Task 12/25] Current/Best: 13.21/ 14.67 GFLOPS | Progress: (8/20) | 8.88 s
[Task 12/25] Current/Best: 4.07/ 21.93 GFLOPS | Progress: (12/20) | 11.77 s
[Task 12/25] Current/Best: 2.57/ 21.93 GFLOPS | Progress: (16/20) | 15.10 s
[Task 12/25] Current/Best: 12.99/ 21.93 GFLOPS | Progress: (20/20) | 17.73 s Done.
+
[Task 13/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 13/25] Current/Best: 18.82/ 19.92 GFLOPS | Progress: (4/20) | 5.19 s
[Task 13/25] Current/Best: 8.35/ 19.92 GFLOPS | Progress: (8/20) | 8.25 s
[Task 13/25] Current/Best: 11.94/ 19.92 GFLOPS | Progress: (12/20) | 11.07 s
[Task 13/25] Current/Best: 19.28/ 19.92 GFLOPS | Progress: (16/20) | 13.54 s
[Task 13/25] Current/Best: 12.26/ 19.92 GFLOPS | Progress: (20/20) | 17.33 s Done.
+
[Task 14/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 14/25] Current/Best: 13.85/ 13.85 GFLOPS | Progress: (4/20) | 5.32 s
[Task 14/25] Current/Best: 5.75/ 15.17 GFLOPS | Progress: (8/20) | 8.23 s
[Task 14/25] Current/Best: 6.29/ 15.43 GFLOPS | Progress: (12/20) | 10.45 s
[Task 14/25] Current/Best: 9.00/ 17.94 GFLOPS | Progress: (16/20) | 13.69 s
[Task 14/25] Current/Best: 4.94/ 17.94 GFLOPS | Progress: (20/20) | 17.61 s Done.
+
[Task 15/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 15/25] Current/Best: 17.87/ 20.91 GFLOPS | Progress: (4/20) | 4.55 s
[Task 15/25] Current/Best: 6.28/ 20.91 GFLOPS | Progress: (8/20) | 6.21 s
[Task 15/25] Current/Best: 4.94/ 20.91 GFLOPS | Progress: (12/20) | 9.44 s
[Task 15/25] Current/Best: 11.45/ 20.91 GFLOPS | Progress: (16/20) | 12.95 s
[Task 15/25] Current/Best: 18.65/ 20.91 GFLOPS | Progress: (20/20) | 14.77 s
[Task 16/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 16/25] Current/Best: 10.99/ 20.86 GFLOPS | Progress: (4/20) | 5.59 s
[Task 16/25] Current/Best: 13.44/ 20.86 GFLOPS | Progress: (8/20) | 9.13 s
[Task 16/25] Current/Best: 12.78/ 20.86 GFLOPS | Progress: (12/20) | 12.72 s
[Task 16/25] Current/Best: 17.47/ 20.86 GFLOPS | Progress: (16/20) | 16.37 s
[Task 16/25] Current/Best: 6.41/ 20.86 GFLOPS | Progress: (20/20)
| 17.99 s Done.
+
[Task 17/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 17/25] Current/Best: 8.16/ 17.00 GFLOPS | Progress: (4/20) | 5.34 s
[Task 17/25] Current/Best: 22.39/ 22.39 GFLOPS | Progress: (8/20) | 8.13 s
[Task 17/25] Current/Best: 10.49/ 22.39 GFLOPS | Progress: (12/20) | 11.01 s
[Task 17/25] Current/Best: 1.56/ 22.39 GFLOPS | Progress: (16/20) | 15.52 s
[Task 17/25] Current/Best: 12.35/ 22.39 GFLOPS | Progress: (20/20) | 18.97 s Done.
+
[Task 18/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 18/25] Current/Best: 10.80/ 10.80 GFLOPS | Progress: (4/20) | 7.08 s
[Task 18/25] Current/Best: 18.90/ 19.82 GFLOPS | Progress: (8/20) | 9.69 s
[Task 18/25] Current/Best: 18.67/ 19.82 GFLOPS | Progress: (12/20) | 11.79 s
[Task 18/25] Current/Best: 14.44/ 19.82 GFLOPS | Progress: (16/20) | 15.31 s
[Task 18/25] Current/Best: 14.54/ 19.82 GFLOPS | Progress: (20/20) | 18.86 s Done.
+
[Task 19/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 19/25] Current/Best: 9.01/ 21.54 GFLOPS | Progress: (4/20) | 6.12 s
[Task 19/25] Current/Best: 9.13/ 21.54 GFLOPS | Progress: (8/20) | 11.66 s
[Task 19/25] Current/Best: 16.80/ 21.54 GFLOPS | Progress: (12/20) | 15.10 s
[Task 19/25] Current/Best: 12.46/ 21.54 GFLOPS | Progress: (16/20) | 18.77 s
[Task 19/25] Current/Best: 6.18/ 21.54 GFLOPS | Progress: (20/20) | 22.57 s Done.
+
[Task 20/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 20/25] Current/Best: 9.12/ 11.79 GFLOPS | Progress: (4/20) | 6.08 s
[Task 20/25] Current/Best: 10.79/ 17.56 GFLOPS | Progress: (8/20) | 8.53 s
[Task 20/25] Current/Best: 8.72/ 17.56 GFLOPS | Progress: (12/20) | 11.37 s
[Task 20/25] Current/Best: 9.83/ 17.56 GFLOPS | Progress: (16/20) | 14.67 s
[Task 20/25] Current/Best: 2.51/ 17.56 GFLOPS | Progress: (20/20) | 18.10 s
[Task 21/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s Done.
Done.
-
[Task 17/25] Current/Best: 14.55/ 22.79 GFLOPS | Progress: (8/20) | 8.37 s
[Task 17/25] Current/Best: 10.27/ 22.79 GFLOPS | Progress: (12/20) | 11.02 s
[Task 17/25] Current/Best: 21.23/ 22.79 GFLOPS | Progress: (16/20) | 14.00 s
[Task 17/25] Current/Best: 9.58/ 22.79 GFLOPS | Progress: (20/20) | 16.65 s Done.
-
[Task 18/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 18/25] Current/Best: 14.60/ 14.62 GFLOPS | Progress: (4/20) | 5.56 s
[Task 18/25] Current/Best: 9.53/ 18.89 GFLOPS | Progress: (8/20) | 9.03 s
[Task 18/25] Current/Best: 9.70/ 18.89 GFLOPS | Progress: (12/20) | 11.63 s
[Task 18/25] Current/Best: 8.67/ 18.89 GFLOPS | Progress: (16/20) | 18.97 s
[Task 18/25] Current/Best: 6.38/ 18.89 GFLOPS | Progress: (20/20) | 23.59 s Done.
-
[Task 19/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 19/25] Current/Best: 7.89/ 18.51 GFLOPS | Progress: (4/20) | 5.39 s
[Task 19/25] Current/Best: 9.30/ 18.51 GFLOPS | Progress: (8/20) | 11.95 s
[Task 19/25] Current/Best: 12.08/ 19.95 GFLOPS | Progress: (12/20) | 14.63 s
[Task 19/25] Current/Best: 10.74/ 19.95 GFLOPS | Progress: (16/20) | 17.94 s
[Task 19/25] Current/Best: 3.08/ 22.61 GFLOPS | Progress: (20/20) | 23.03 s Done.
-
[Task 20/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 20/25] Current/Best: 6.52/ 16.18 GFLOPS | Progress: (4/20) | 6.72 s
[Task 20/25] Current/Best: 14.47/ 16.46 GFLOPS | Progress: (8/20) | 9.56 s
[Task 20/25] Current/Best: 18.36/ 18.36 GFLOPS | Progress: (12/20) | 12.52 s
[Task 20/25] Current/Best: 8.67/ 19.83 GFLOPS | Progress: (16/20) | 16.44 s
[Task 20/25] Current/Best: 19.79/ 19.83 GFLOPS | Progress: (20/20) | 19.19 s
[Task 21/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 21/25] Current/Best: 7.40/ 14.37 GFLOPS | Progress: (4/20) | 5.85 s
[Task 21/25] Current/Best: 17.53/ 18.47 GFLOPS | Progress: (8/20) | 7.10 s
[Task 21/25] Current/Best: 6.43/ 18.47 GFLOPS | Progress: (12/20) | 9.94 s Done.
-
[Task 21/25] Current/Best: 12.30/ 18.47 GFLOPS | Progress: (16/20) | 12.90 s
[Task 21/25] Current/Best: 18.77/ 20.08 GFLOPS | Progress: (20/20) | 17.07 s
[Task 22/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 22/25] Current/Best: 21.48/ 21.48 GFLOPS | Progress: (4/20) | 4.58 s
[Task 22/25] Current/Best: 9.01/ 21.48 GFLOPS | Progress: (8/20) | 7.68 s
[Task 22/25] Current/Best: 9.78/ 21.48 GFLOPS | Progress: (12/20) | 9.99 s
[Task 22/25] Current/Best: 15.46/ 21.48 GFLOPS | Progress: (16/20) | 11.59 s
[Task 22/25] Current/Best: 17.22/ 21.48 GFLOPS | Progress: (20/20) | 14.80 s Done.
-
[Task 23/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 23/25] Current/Best: 6.16/ 23.16 GFLOPS | Progress: (4/20) | 5.28 s
[Task 23/25] Current/Best: 11.89/ 23.16 GFLOPS | Progress: (8/20) | 9.45 s
[Task 23/25] Current/Best: 20.66/ 23.16 GFLOPS | Progress: (12/20) | 12.57 s
[Task 23/25] Current/Best: 1.55/ 23.26 GFLOPS | Progress: (16/20) | 16.94 s
[Task 23/25] Current/Best: 11.20/ 23.26 GFLOPS | Progress: (20/20) | 20.06 s Done.
-
[Task 24/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 24/25] Current/Best: 4.00/ 4.00 GFLOPS | Progress: (4/20) | 13.48 s
[Task 24/25] Current/Best: 9.69/ 10.26 GFLOPS | Progress: (8/20) | 16.17 s
[Task 24/25] Current/Best: 3.48/ 10.26 GFLOPS | Progress: (12/20) | 25.04 s
[Task 24/25] Current/Best: 1.87/ 10.26 GFLOPS | Progress: (16/20) | 37.32 s
[Task 24/25] Current/Best: 1.85/ 10.26 GFLOPS | Progress: (20/20) | 49.65 s
[Task 25/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s Done.
+
[Task 21/25] Current/Best: 10.80/ 19.41 GFLOPS | Progress: (4/20) | 4.53 s
[Task 21/25] Current/Best: 16.30/ 19.41 GFLOPS | Progress: (8/20) | 6.83 s
[Task 21/25] Current/Best: 12.27/ 19.41 GFLOPS | Progress: (12/20) | 8.83 s
[Task 21/25] Current/Best: 18.09/ 20.26 GFLOPS | Progress: (16/20) | 10.43 s
[Task 21/25] Current/Best: 8.01/ 20.26 GFLOPS | Progress: (20/20) | 12.05 s
[Task 22/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 22/25] Current/Best: 10.59/ 20.61 GFLOPS | Progress: (4/20) | 5.69 s
[Task 22/25] Current/Best: 8.46/ 20.61 GFLOPS | Progress: (8/20) | 7.97 s
[Task 22/25] Current/Best: 15.10/ 20.61 GFLOPS | Progress: (12/20) | 9.98 s
[Task 22/25] Current/Best: 9.66/ 20.61 GFLOPS | Progress: (16/20) | 12.10 s
[Task 22/25] Current/Best: 3.08/ 20.61 GFLOPS | Progress: (20/20) | 15.93 s Done.
+
[Task 23/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 23/25] Current/Best: 10.11/ 12.24 GFLOPS | Progress: (4/20) | 7.28 s
[Task 23/25] Current/Best: 14.56/ 18.48 GFLOPS | Progress: (8/20) | 10.46 s
[Task 23/25] Current/Best: 11.51/ 18.48 GFLOPS | Progress: (12/20) | 15.89 s
[Task 23/25] Current/Best: 10.22/ 18.63 GFLOPS | Progress: (16/20) | 18.64 s
[Task 23/25] Current/Best: 23.89/ 23.89 GFLOPS | Progress: (20/20) | 21.17 s Done.
+
[Task 24/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 24/25] Current/Best: 6.11/ 6.95 GFLOPS | Progress: (4/20) | 13.70 s
[Task 24/25] Current/Best: 10.04/ 10.04 GFLOPS | Progress: (8/20) | 24.67 s
[Task 24/25] Current/Best: 0.76/ 10.04 GFLOPS | Progress: (12/20) | 28.14 s
[Task 24/25] Current/Best: 4.67/ 10.77 GFLOPS | Progress: (16/20) | 39.08 s
[Task 24/25] Current/Best: 3.98/ 10.77 GFLOPS | Progress: (20/20) | 49.76 s
[Task 25/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s Done.
Done.
-
[Task 25/25] Current/Best: 6.11/ 7.42 GFLOPS | Progress: (4/20) | 8.13 s
[Task 25/25] Current/Best: 4.66/ 8.93 GFLOPS | Progress: (8/20) | 16.27 s
[Task 25/25] Current/Best: 5.71/ 8.93 GFLOPS | Progress: (12/20) | 17.62 s
[Task 25/25] Current/Best: 9.06/ 9.06 GFLOPS | Progress: (16/20) | 28.56 s
[Task 25/25] Current/Best: 5.51/ 9.06 GFLOPS | Progress: (20/20) | 31.66 s
+
[Task 25/25] Current/Best: 4.17/ 4.62 GFLOPS | Progress: (4/20) | 15.05 s
[Task 25/25] Current/Best: 9.86/ 9.86 GFLOPS | Progress: (8/20) | 21.80 s
[Task 25/25] Current/Best: 1.54/ 9.86 GFLOPS | Progress: (12/20) | 23.28 s
[Task 25/25] Current/Best: 3.03/ 9.86 GFLOPS | Progress: (16/20) | 31.93 s
[Task 25/25] Current/Best: 9.19/ 9.86 GFLOPS | Progress: (20/20) | 42.90 s
@@ -665,8 +665,8 @@ Verify that the optimized model runs and produces the same results:
.. code-block:: none
- class='n02123045 tabby, tabby cat' with probability=0.621104
- class='n02123159 tiger cat' with probability=0.356378
+ class='n02123045 tabby, tabby cat' with probability=0.621102
+ class='n02123159 tiger cat' with probability=0.356380
class='n02124075 Egyptian cat' with probability=0.019712
class='n02129604 tiger, Panthera tigris' with probability=0.001215
class='n04040759 radiator' with probability=0.000262
@@ -723,8 +723,8 @@ improvement in comparing the optimized model to the unoptimized model.
.. code-block:: none
- optimized: {'mean': 423.0942322499982, 'median': 424.0394484000035, 'std': 5.271453740563966}
- unoptimized: {'mean': 514.94100076, 'median': 514.3353281999964, 'std': 2.3532936330814875}
+ optimized: {'mean': 420.7347483700005, 'median': 421.86745820000624, 'std': 3.534493201096029}
+ unoptimized: {'mean': 514.6666772300001, 'median': 515.163708800003, 'std': 2.3164496047830663}
@@ -747,7 +747,7 @@ profiling/benchmarking.
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 12 minutes 54.544 seconds)
+ **Total running time of the script:** ( 12 minutes 29.178 seconds)
.. _sphx_glr_download_tutorial_autotvm_relay_x86.py:
diff --git a/docs/_sources/tutorial/cross_compilation_and_rpc.rst.txt b/docs/_sources/tutorial/cross_compilation_and_rpc.rst.txt
index 84f0ccebe4..d60e6741b4 100644
--- a/docs/_sources/tutorial/cross_compilation_and_rpc.rst.txt
+++ b/docs/_sources/tutorial/cross_compilation_and_rpc.rst.txt
@@ -274,7 +274,7 @@ device and returns the measured cost. Network overhead is excluded.
.. code-block:: none
- 1.289e-07 secs/op
+ 1.335e-07 secs/op
diff --git a/docs/_sources/tutorial/intro_topi.rst.txt b/docs/_sources/tutorial/intro_topi.rst.txt
index 5170824701..12569422f1 100644
--- a/docs/_sources/tutorial/intro_topi.rst.txt
+++ b/docs/_sources/tutorial/intro_topi.rst.txt
@@ -270,7 +270,7 @@ As you can see, scheduled stages of computation have been accumulated and we can
.. code-block:: none
- [stage(a, placeholder(a, 0x22c70ab0)), stage(b, placeholder(b, 0x753a130)), stage(T_add, compute(T_add, body=[a[ax0, ax1, ax2] + b[ax1, ax2]], axis=[T.iter_var(ax0, T.Range(0, 100), "DataPar", ""), T.iter_var(ax1, T.Range(0, 10), "DataPar", ""), T.iter_var(ax2, T.Range(0, 10), "DataPar", "")], reduce_axis=[], tag=broadcast, attrs={})), stage(T_multiply, compute(T_multiply, body=[a[ax0, ax1, ax2] * b[ax1, ax2]], axis=[T.iter_var(ax0, T.Range(0, 100), "DataPar", ""), T.iter_var(ax1, T. [...]
+ [stage(a, placeholder(a, 0x2a1653c0)), stage(b, placeholder(b, 0x7cff960)), stage(T_add, compute(T_add, body=[a[ax0, ax1, ax2] + b[ax1, ax2]], axis=[T.iter_var(ax0, T.Range(0, 100), "DataPar", ""), T.iter_var(ax1, T.Range(0, 10), "DataPar", ""), T.iter_var(ax2, T.Range(0, 10), "DataPar", "")], reduce_axis=[], tag=broadcast, attrs={})), stage(T_multiply, compute(T_multiply, body=[a[ax0, ax1, ax2] * b[ax1, ax2]], axis=[T.iter_var(ax0, T.Range(0, 100), "DataPar", ""), T.iter_var(ax1, T. [...]
diff --git a/docs/_sources/tutorial/sg_execution_times.rst.txt b/docs/_sources/tutorial/sg_execution_times.rst.txt
index 06d5aec49e..a92ccac50d 100644
--- a/docs/_sources/tutorial/sg_execution_times.rst.txt
+++ b/docs/_sources/tutorial/sg_execution_times.rst.txt
@@ -5,32 +5,32 @@
Computation times
=================
-**16:49.309** total execution time for **tutorial** files:
+**16:20.567** total execution time for **tutorial** files:
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_autotvm_relay_x86.py` (``autotvm_relay_x86.py``) | 12:54.544 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_autotvm_relay_x86.py` (``autotvm_relay_x86.py``) | 12:29.178 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_auto_scheduler_matmul_x86.py` (``auto_scheduler_matmul_x86.py``) | 01:48.356 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_auto_scheduler_matmul_x86.py` (``auto_scheduler_matmul_x86.py``) | 01:41.541 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_tensor_expr_get_started.py` (``tensor_expr_get_started.py``) | 01:00.412 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_tensor_expr_get_started.py` (``tensor_expr_get_started.py``) | 00:59.162 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_relay_quick_start.py` (``relay_quick_start.py``) | 00:38.756 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_relay_quick_start.py` (``relay_quick_start.py``) | 00:36.213 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_autotvm_matmul_x86.py` (``autotvm_matmul_x86.py``) | 00:24.591 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_autotvm_matmul_x86.py` (``autotvm_matmul_x86.py``) | 00:31.908 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_tensor_ir_blitz_course.py` (``tensor_ir_blitz_course.py``) | 00:01.553 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_tensor_ir_blitz_course.py` (``tensor_ir_blitz_course.py``) | 00:01.538 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_intro_topi.py` (``intro_topi.py``) | 00:00.914 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_intro_topi.py` (``intro_topi.py``) | 00:00.847 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_cross_compilation_and_rpc.py` (``cross_compilation_and_rpc.py``) | 00:00.181 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_cross_compilation_and_rpc.py` (``cross_compilation_and_rpc.py``) | 00:00.180 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
| :ref:`sphx_glr_tutorial_uma.py` (``uma.py``) | 00:00.000 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_introduction.py` (``introduction.py``) | 00:00.000 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_tvmc_python.py` (``tvmc_python.py``) | 00:00.000 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
| :ref:`sphx_glr_tutorial_tvmc_command_line_driver.py` (``tvmc_command_line_driver.py``) | 00:00.000 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_tvmc_python.py` (``tvmc_python.py``) | 00:00.000 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_introduction.py` (``introduction.py``) | 00:00.000 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
| :ref:`sphx_glr_tutorial_install.py` (``install.py``) | 00:00.000 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/tutorial/tensor_expr_get_started.rst.txt b/docs/_sources/tutorial/tensor_expr_get_started.rst.txt
index b0a17adbfc..40e84986a1 100644
--- a/docs/_sources/tutorial/tensor_expr_get_started.rst.txt
+++ b/docs/_sources/tutorial/tensor_expr_get_started.rst.txt
@@ -285,7 +285,7 @@ helper function to run a profile of the TVM generated code.
.. code-block:: none
- Numpy running time: 0.000008
+ Numpy running time: 0.000007
naive: 0.000007
@@ -389,7 +389,7 @@ compile and run this new schedule with the parallel operation applied:
.. code-block:: none
- parallel: 0.000007
+ parallel: 0.000006
@@ -444,7 +444,7 @@ factor to be the number of threads on your CPU.
.. code-block:: none
- vector: 0.000026
+ vector: 0.000025
# from tvm.script import ir as I
# from tvm.script import tir as T
@@ -498,10 +498,10 @@ We can now compare the different schedules
.. code-block:: none
Operator Timing Performance
- numpy 7.821710000825988e-06 1.0
- naive 6.690300000000001e-06 0.8553500448486957
- parallel 6.973e-06 0.8914930365947648
- vector 2.57299e-05 3.2895492158725994
+ numpy 7.400439999400987e-06 1.0
+ naive 6.7052e-06 0.9060542346864157
+ parallel 6.0795e-06 0.8215052078649502
+ vector 2.4561500000000002e-05 3.318924280446579
@@ -922,7 +922,7 @@ matrix multiplication.
.. code-block:: none
- Numpy running time: 0.017789
+ Numpy running time: 0.017987
@@ -980,7 +980,7 @@ optimizations.
.. code-block:: none
- none: 3.374645
+ none: 3.286562
@@ -1080,7 +1080,7 @@ schedule.
.. code-block:: none
- blocking: 0.294012
+ blocking: 0.295853
@@ -1164,7 +1164,7 @@ already cache friendly from our previous optimizations.
.. code-block:: none
- vectorization: 0.339040
+ vectorization: 0.333159
# from tvm.script import ir as I
# from tvm.script import tir as T
@@ -1230,7 +1230,7 @@ more cache friendly.
.. code-block:: none
- loop permutation: 0.118001
+ loop permutation: 0.114596
# from tvm.script import ir as I
# from tvm.script import tir as T
@@ -1321,7 +1321,7 @@ optimized schedule.
.. code-block:: none
- array packing: 0.108492
+ array packing: 0.108702
# from tvm.script import ir as I
# from tvm.script import tir as T
@@ -1404,7 +1404,7 @@ to `C` when all the block results are ready.
.. code-block:: none
- block caching: 0.109656
+ block caching: 0.110778
# from tvm.script import ir as I
# from tvm.script import tir as T
@@ -1478,7 +1478,7 @@ of thread-level parallelization.
.. code-block:: none
- parallelization: 0.146285
+ parallelization: 0.145846
# from tvm.script import ir as I
# from tvm.script import tir as T
@@ -1548,13 +1548,13 @@ working, we can compare the results.
.. code-block:: none
Operator Timing Performance
- none 3.374644795699999 1.0
- blocking 0.29401200330000005 0.08712383705527545
- vectorization 0.3390397117 0.10046678457300373
- loop permutation 0.11800106399999999 0.03496695834487764
- array packing 0.1084916743 0.03214906482550134
- block caching 0.109655824 0.03249403437651405
- parallelization 0.1462847955 0.04334820532412695
+ none 3.2865615013000005 1.0
+ blocking 0.2958530499 0.09001902133368726
+ vectorization 0.33315863290000003 0.10136996759933414
+ loop permutation 0.11459570240000001 0.03486796226228283
+ array packing 0.1087022213 0.03307475647633638
+ block caching 0.11077763330000001 0.033706240779666496
+ parallelization 0.1458455159 0.04437632335263185
@@ -1594,11 +1594,6 @@ operations with tunable parameters that allows you to automatically optimize
the computation for specific platforms.
-.. rst-class:: sphx-glr-timing
-
- **Total running time of the script:** ( 1 minutes 0.412 seconds)
-
-
.. _sphx_glr_download_tutorial_tensor_expr_get_started.py:
.. only:: html
diff --git a/docs/commit_hash b/docs/commit_hash
index f6e7af89a2..a8fa033473 100644
--- a/docs/commit_hash
+++ b/docs/commit_hash
@@ -1 +1 @@
-84e8f869607905be56b325ba8e2acd3dfd6a0394
+4ae08d87105afa663ac1c0392a30058912093eb4
diff --git a/docs/how_to/compile_models/from_darknet.html b/docs/how_to/compile_models/from_darknet.html
index e3bd09d48a..dd6fef8d6b 100644
--- a/docs/how_to/compile_models/from_darknet.html
+++ b/docs/how_to/compile_models/from_darknet.html
@@ -590,7 +590,7 @@ class:['truck 0.9266'] left:471 top:83 right:689 bottom:169
class:['bicycle 0.9984'] left:111 top:113 right:577 bottom:447
</pre></div>
</div>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 25.999 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 17.208 seconds)</p>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-compile-models-from-darknet-py">
<div class="sphx-glr-download sphx-glr-download-python docutils container">
<p><a class="reference download internal" download="" href="../../_downloads/7716f96385bd5abb6e822041e285be54/from_darknet.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">from_darknet.py</span></code></a></p>
diff --git a/docs/how_to/compile_models/from_keras.html b/docs/how_to/compile_models/from_keras.html
index 28d3f4b274..b4eb9ad70d 100644
--- a/docs/how_to/compile_models/from_keras.html
+++ b/docs/how_to/compile_models/from_keras.html
@@ -511,7 +511,7 @@ Tensorflow is also required since it’s used as the default backend of keras.</
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Relay top-1 id: 285, class name: Egyptian cat
1/1 [==============================] - ETA: 0s
-1/1 [==============================] - 1s 1s/step
+1/1 [==============================] - 1s 945ms/step
Keras top-1 id: 285, class name: Egyptian cat
</pre></div>
</div>
diff --git a/docs/how_to/compile_models/from_mxnet.html b/docs/how_to/compile_models/from_mxnet.html
index 9551867f00..7d7ebbc71e 100644
--- a/docs/how_to/compile_models/from_mxnet.html
+++ b/docs/how_to/compile_models/from_mxnet.html
@@ -444,7 +444,7 @@
<span class="nb">print</span><span class="p">(</span><span class="s2">"x"</span><span class="p">,</span> <a href="https://docs.python.org/3/library/stdtypes.html#tuple" title="builtins.tuple" class="sphx-glr-backref-module-builtins sphx-glr-backref-type-py-class sphx-glr-backref-instance"><span class="n">x</span><span class="o">.</span><span class="n">shape</span></a><span class="p">)</span>
</pre></div>
</div>
-<img src="../../_images/sphx_glr_from_mxnet_001.png" srcset="../../_images/sphx_glr_from_mxnet_001.png" alt="from mxnet" class = "sphx-glr-single-img"/><div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Downloading /workspace/.mxnet/models/resnet18_v1-a0666292.zipb9f2e74b-857a-4d43-8ef5-b9cf3ae31ea9 from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/resnet18_v1-a0666292.zip...
+<img src="../../_images/sphx_glr_from_mxnet_001.png" srcset="../../_images/sphx_glr_from_mxnet_001.png" alt="from mxnet" class = "sphx-glr-single-img"/><div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Downloading /workspace/.mxnet/models/resnet18_v1-a0666292.zipa4a1b008-0ef6-41a2-8421-d5c49a2c614a from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/resnet18_v1-a0666292.zip...
x (1, 3, 224, 224)
</pre></div>
</div>
diff --git a/docs/how_to/compile_models/from_oneflow.html b/docs/how_to/compile_models/from_oneflow.html
index 51476de7e1..00ef57777b 100644
--- a/docs/how_to/compile_models/from_oneflow.html
+++ b/docs/how_to/compile_models/from_oneflow.html
@@ -454,12 +454,11 @@ Deprecated in NumPy 1.20; for more details and guidance: https://numpy.org/devdo
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Downloading: "https://oneflow-public.oss-cn-beijing.aliyuncs.com/model_zoo/flowvision/classification/ResNet/resnet18.zip" to /workspace/.oneflow/flowvision_cache/resnet18.zip
0%| | 0.00/41.5M [00:00<?, ?B/s]
- 19%|#9 | 7.99M/41.5M [00:00<00:00, 57.2MB/s]
- 35%|###4 | 14.3M/41.5M [00:00<00:00, 60.5MB/s]
- 54%|#####3 | 22.3M/41.5M [00:00<00:00, 54.8MB/s]
- 67%|######6 | 27.7M/41.5M [00:00<00:00, 53.9MB/s]
- 92%|#########2| 38.3M/41.5M [00:00<00:00, 63.9MB/s]
-100%|##########| 41.5M/41.5M [00:00<00:00, 58.2MB/s]
+ 19%|#9 | 7.99M/41.5M [00:00<00:00, 69.3MB/s]
+ 39%|###8 | 16.0M/41.5M [00:00<00:00, 58.5MB/s]
+ 58%|#####7 | 24.0M/41.5M [00:00<00:00, 61.8MB/s]
+ 87%|########7 | 36.2M/41.5M [00:00<00:00, 81.0MB/s]
+100%|##########| 41.5M/41.5M [00:00<00:00, 74.6MB/s]
</pre></div>
</div>
</div>
diff --git a/docs/how_to/compile_models/from_pytorch.html b/docs/how_to/compile_models/from_pytorch.html
index fa661c7929..4c0cc3b0dc 100644
--- a/docs/how_to/compile_models/from_pytorch.html
+++ b/docs/how_to/compile_models/from_pytorch.html
@@ -437,11 +437,10 @@ be unstable.</p>
Downloading: "https://download.pytorch.org/models/resnet18-f37072fd.pth" to /workspace/.cache/torch/hub/checkpoints/resnet18-f37072fd.pth
0%| | 0.00/44.7M [00:00<?, ?B/s]
- 18%|#8 | 8.12M/44.7M [00:00<00:00, 81.0MB/s]
- 40%|#### | 18.1M/44.7M [00:00<00:00, 93.8MB/s]
- 61%|###### | 27.0M/44.7M [00:00<00:00, 67.9MB/s]
- 90%|########9 | 40.0M/44.7M [00:00<00:00, 84.1MB/s]
-100%|##########| 44.7M/44.7M [00:00<00:00, 76.1MB/s]
+ 30%|##9 | 13.4M/44.7M [00:00<00:00, 140MB/s]
+ 60%|#####9 | 26.7M/44.7M [00:00<00:00, 111MB/s]
+ 84%|########4 | 37.6M/44.7M [00:00<00:00, 108MB/s]
+100%|##########| 44.7M/44.7M [00:00<00:00, 105MB/s]
</pre></div>
</div>
</div>
diff --git a/docs/how_to/compile_models/from_tensorflow.html b/docs/how_to/compile_models/from_tensorflow.html
index 31475fd94d..6f24008b10 100644
--- a/docs/how_to/compile_models/from_tensorflow.html
+++ b/docs/how_to/compile_models/from_tensorflow.html
@@ -654,7 +654,7 @@ banana (score = 0.00022)
desk (score = 0.00019)
</pre></div>
</div>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 30.735 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 22.350 seconds)</p>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-compile-models-from-tensorflow-py">
<div class="sphx-glr-download sphx-glr-download-python docutils container">
<p><a class="reference download internal" download="" href="../../_downloads/7f1d3d1b878694c201c614c807cdebc8/from_tensorflow.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">from_tensorflow.py</span></code></a></p>
diff --git a/docs/how_to/compile_models/sg_execution_times.html b/docs/how_to/compile_models/sg_execution_times.html
index 2a9cb13cfb..05b57338cd 100644
--- a/docs/how_to/compile_models/sg_execution_times.html
+++ b/docs/how_to/compile_models/sg_execution_times.html
@@ -345,7 +345,7 @@
<div class="section" id="computation-times">
<span id="sphx-glr-how-to-compile-models-sg-execution-times"></span><h1>Computation times<a class="headerlink" href="#computation-times" title="Permalink to this headline">¶</a></h1>
-<p><strong>06:58.560</strong> total execution time for <strong>how_to_compile_models</strong> files:</p>
+<p><strong>06:32.674</strong> total execution time for <strong>how_to_compile_models</strong> files:</p>
<table class="docutils align-default">
<colgroup>
<col style="width: 81%" />
@@ -354,43 +354,43 @@
</colgroup>
<tbody>
<tr class="row-odd"><td><p><a class="reference internal" href="from_tensorflow.html#sphx-glr-how-to-compile-models-from-tensorflow-py"><span class="std std-ref">Compile Tensorflow Models</span></a> (<code class="docutils literal notranslate"><span class="pre">from_tensorflow.py</span></code>)</p></td>
-<td><p>01:30.735</p></td>
+<td><p>01:22.350</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="from_darknet.html#sphx-glr-how-to-compile-models-from-darknet-py"><span class="std std-ref">Compile YOLO-V2 and YOLO-V3 in DarkNet Models</span></a> (<code class="docutils literal notranslate"><span class="pre">from_darknet.py</span></code>)</p></td>
-<td><p>01:25.999</p></td>
+<td><p>01:17.208</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="from_paddle.html#sphx-glr-how-to-compile-models-from-paddle-py"><span class="std std-ref">Compile PaddlePaddle Models</span></a> (<code class="docutils literal notranslate"><span class="pre">from_paddle.py</span></code>)</p></td>
-<td><p>00:55.487</p></td>
+<td><p>00:55.122</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="from_oneflow.html#sphx-glr-how-to-compile-models-from-oneflow-py"><span class="std std-ref">Compile OneFlow Models</span></a> (<code class="docutils literal notranslate"><span class="pre">from_oneflow.py</span></code>)</p></td>
-<td><p>00:37.646</p></td>
+<td><p>00:37.119</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="from_coreml.html#sphx-glr-how-to-compile-models-from-coreml-py"><span class="std std-ref">Compile CoreML Models</span></a> (<code class="docutils literal notranslate"><span class="pre">from_coreml.py</span></code>)</p></td>
-<td><p>00:35.740</p></td>
+<td><p>00:31.687</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="from_mxnet.html#sphx-glr-how-to-compile-models-from-mxnet-py"><span class="std std-ref">Compile MXNet Models</span></a> (<code class="docutils literal notranslate"><span class="pre">from_mxnet.py</span></code>)</p></td>
-<td><p>00:33.060</p></td>
+<td><p>00:31.006</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="from_tflite.html#sphx-glr-how-to-compile-models-from-tflite-py"><span class="std std-ref">Compile TFLite Models</span></a> (<code class="docutils literal notranslate"><span class="pre">from_tflite.py</span></code>)</p></td>
-<td><p>00:29.017</p></td>
+<td><p>00:27.332</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="from_pytorch.html#sphx-glr-how-to-compile-models-from-pytorch-py"><span class="std std-ref">Compile PyTorch Models</span></a> (<code class="docutils literal notranslate"><span class="pre">from_pytorch.py</span></code>)</p></td>
-<td><p>00:26.231</p></td>
+<td><p>00:25.712</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="from_keras.html#sphx-glr-how-to-compile-models-from-keras-py"><span class="std std-ref">Compile Keras Models</span></a> (<code class="docutils literal notranslate"><span class="pre">from_keras.py</span></code>)</p></td>
-<td><p>00:21.903</p></td>
+<td><p>00:22.387</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="from_onnx.html#sphx-glr-how-to-compile-models-from-onnx-py"><span class="std std-ref">Compile ONNX Models</span></a> (<code class="docutils literal notranslate"><span class="pre">from_onnx.py</span></code>)</p></td>
-<td><p>00:02.741</p></td>
+<td><p>00:02.750</p></td>
<td><p>0.0 MB</p></td>
</tr>
</tbody>
diff --git a/docs/how_to/deploy_models/deploy_model_on_adreno.html b/docs/how_to/deploy_models/deploy_model_on_adreno.html
index d35be5c993..80d205d4d0 100644
--- a/docs/how_to/deploy_models/deploy_model_on_adreno.html
+++ b/docs/how_to/deploy_models/deploy_model_on_adreno.html
@@ -799,7 +799,7 @@ Top5 predictions:
Evaluate inference time cost...
Execution time summary:
mean (ms) median (ms) max (ms) min (ms) std (ms)
- 2539.6140 2539.3624 2542.1952 2537.9811 1.3877
+ 2540.1147 2539.6340 2544.0332 2538.5511 1.6081
</pre></div>
</div>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-deploy-models-deploy-model-on-adreno-py">
diff --git a/docs/how_to/deploy_models/deploy_model_on_adreno_tvmc.html b/docs/how_to/deploy_models/deploy_model_on_adreno_tvmc.html
index 439fbba150..79c2b28d0f 100644
--- a/docs/how_to/deploy_models/deploy_model_on_adreno_tvmc.html
+++ b/docs/how_to/deploy_models/deploy_model_on_adreno_tvmc.html
@@ -443,24 +443,25 @@ to run this tutorial with a real device over rpc.</p>
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Downloading data from https://storage.googleapis.com/tensorflow/keras-applications/resnet/resnet50_weights_tf_dim_ordering_tf_kernels.h5
8192/102967424 [..............................] - ETA: 0s
- 7725056/102967424 [=>............................] - ETA: 0s
- 12697600/102967424 [==>...........................] - ETA: 0s
- 16908288/102967424 [===>..........................] - ETA: 0s
- 22945792/102967424 [=====>........................] - ETA: 0s
- 27918336/102967424 [=======>......................] - ETA: 0s
- 32964608/102967424 [========>.....................] - ETA: 0s
- 37969920/102967424 [==========>...................] - ETA: 0s
- 42328064/102967424 [===========>..................] - ETA: 0s
- 48013312/102967424 [============>.................] - ETA: 0s
- 53018624/102967424 [==============>...............] - ETA: 0s
- 58023936/102967424 [===============>..............] - ETA: 0s
- 63004672/102967424 [=================>............] - ETA: 0s
- 68001792/102967424 [==================>...........] - ETA: 0s
- 70123520/102967424 [===================>..........] - ETA: 0s
- 78610432/102967424 [=====================>........] - ETA: 0s
- 80936960/102967424 [======================>.......] - ETA: 0s
- 91070464/102967424 [=========================>....] - ETA: 0s
- 99901440/102967424 [============================>.] - ETA: 0s
+ 131072/102967424 [..............................] - ETA: 39s
+ 7012352/102967424 [=>............................] - ETA: 1s
+ 14532608/102967424 [===>..........................] - ETA: 0s
+ 16769024/102967424 [===>..........................] - ETA: 1s
+ 25157632/102967424 [======>.......................] - ETA: 1s
+ 33546240/102967424 [========>.....................] - ETA: 0s
+ 39198720/102967424 [==========>...................] - ETA: 0s
+ 44130304/102967424 [===========>..................] - ETA: 0s
+ 49438720/102967424 [=============>................] - ETA: 0s
+ 51462144/102967424 [=============>................] - ETA: 0s
+ 58712064/102967424 [================>.............] - ETA: 0s
+ 67305472/102967424 [==================>...........] - ETA: 0s
+ 69296128/102967424 [===================>..........] - ETA: 0s
+ 72540160/102967424 [====================>.........] - ETA: 0s
+ 75489280/102967424 [====================>.........] - ETA: 0s
+ 83877888/102967424 [=======================>......] - ETA: 0s
+ 88711168/102967424 [========================>.....] - ETA: 0s
+ 92266496/102967424 [=========================>....] - ETA: 0s
+100794368/102967424 [============================>.] - ETA: 0s
102967424/102967424 [==============================] - 1s 0us/step
</pre></div>
</div>
diff --git a/docs/how_to/deploy_models/deploy_model_on_android.html b/docs/how_to/deploy_models/deploy_model_on_android.html
index e3242e5788..6a166af85e 100644
--- a/docs/how_to/deploy_models/deploy_model_on_android.html
+++ b/docs/how_to/deploy_models/deploy_model_on_android.html
@@ -667,7 +667,7 @@ to the remote android device.</p>
Evaluate inference time cost...
Execution time summary:
mean (ms) median (ms) max (ms) min (ms) std (ms)
- 16.3628 16.2917 17.5737 15.7696 0.5427
+ 15.4342 15.4281 15.5401 15.3692 0.0423
</pre></div>
</div>
</div>
diff --git a/docs/how_to/deploy_models/deploy_object_detection_pytorch.html b/docs/how_to/deploy_models/deploy_object_detection_pytorch.html
index 87427d4e1a..f567946c2a 100644
--- a/docs/how_to/deploy_models/deploy_object_detection_pytorch.html
+++ b/docs/how_to/deploy_models/deploy_object_detection_pytorch.html
@@ -459,30 +459,21 @@ be unstable.</p>
Downloading: "https://download.pytorch.org/models/maskrcnn_resnet50_fpn_coco-bf2d0c1e.pth" to /workspace/.cache/torch/hub/checkpoints/maskrcnn_resnet50_fpn_coco-bf2d0c1e.pth
0%| | 0.00/170M [00:00<?, ?B/s]
- 4%|3 | 6.30M/170M [00:00<00:06, 26.9MB/s]
- 5%|5 | 8.88M/170M [00:00<00:06, 26.8MB/s]
- 9%|9 | 16.0M/170M [00:00<00:04, 38.7MB/s]
- 14%|#4 | 24.0M/170M [00:00<00:02, 51.2MB/s]
- 19%|#9 | 32.6M/170M [00:00<00:02, 63.0MB/s]
- 26%|##5 | 43.4M/170M [00:00<00:01, 77.8MB/s]
- 30%|### | 51.3M/170M [00:00<00:01, 74.0MB/s]
- 35%|###4 | 58.7M/170M [00:01<00:02, 52.3MB/s]
- 38%|###8 | 64.7M/170M [00:01<00:02, 46.7MB/s]
- 42%|####2 | 72.0M/170M [00:01<00:02, 47.4MB/s]
- 47%|####7 | 80.0M/170M [00:01<00:01, 53.4MB/s]
- 52%|#####1 | 88.0M/170M [00:01<00:01, 59.5MB/s]
- 57%|#####6 | 96.0M/170M [00:01<00:01, 56.8MB/s]
- 61%|######1 | 104M/170M [00:02<00:01, 53.3MB/s]
- 66%|######5 | 112M/170M [00:02<00:01, 56.2MB/s]
- 71%|####### | 120M/170M [00:02<00:00, 55.4MB/s]
- 75%|#######5 | 128M/170M [00:02<00:00, 57.5MB/s]
- 80%|######## | 136M/170M [00:02<00:00, 52.7MB/s]
- 85%|########4 | 144M/170M [00:02<00:00, 54.6MB/s]
- 89%|########9 | 152M/170M [00:02<00:00, 56.6MB/s]
- 93%|#########3| 158M/170M [00:03<00:00, 54.4MB/s]
- 96%|#########6| 164M/170M [00:03<00:00, 48.9MB/s]
- 99%|#########9| 168M/170M [00:03<00:00, 44.3MB/s]
-100%|##########| 170M/170M [00:03<00:00, 52.5MB/s]
+ 8%|7 | 12.8M/170M [00:00<00:01, 134MB/s]
+ 15%|#5 | 25.5M/170M [00:00<00:01, 111MB/s]
+ 22%|##1 | 36.6M/170M [00:00<00:01, 113MB/s]
+ 28%|##8 | 48.0M/170M [00:00<00:01, 99.2MB/s]
+ 34%|###3 | 57.7M/170M [00:00<00:01, 99.7MB/s]
+ 42%|####2 | 71.7M/170M [00:00<00:00, 107MB/s]
+ 49%|####8 | 82.6M/170M [00:00<00:01, 91.4MB/s]
+ 54%|#####3 | 91.7M/170M [00:00<00:00, 89.9MB/s]
+ 59%|#####9 | 100M/170M [00:01<00:00, 85.4MB/s]
+ 66%|######5 | 112M/170M [00:01<00:00, 90.9MB/s]
+ 75%|#######4 | 127M/170M [00:01<00:00, 109MB/s]
+ 81%|########1 | 138M/170M [00:01<00:00, 82.0MB/s]
+ 89%|########9 | 152M/170M [00:01<00:00, 94.7MB/s]
+ 95%|#########5| 162M/170M [00:01<00:00, 85.6MB/s]
+100%|##########| 170M/170M [00:01<00:00, 94.8MB/s]
/venv/apache-tvm-py3.7/lib/python3.7/site-packages/torch/nn/functional.py:3897: UserWarning: To copy construct from a tensor, it is recommended to use sourceTensor.clone().detach() or sourceTensor.clone().detach().requires_grad_(True), rather than torch.tensor(sourceTensor).
for i in range(dim)
/venv/apache-tvm-py3.7/lib/python3.7/site-packages/torchvision/models/detection/anchor_utils.py:124: UserWarning: __floordiv__ is deprecated, and its behavior will change in a future version of pytorch. It currently rounds toward 0 (like the 'trunc' function NOT 'floor'). This results in incorrect rounding for negative values. To keep the current behavior, use torch.div(a, b, rounding_mode='trunc'), or for actual floor division, use torch.div(a, b, rounding_mode=& [...]
@@ -580,7 +571,7 @@ torchvision rcnn models.</p>
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Get 9 valid boxes
</pre></div>
</div>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 3 minutes 48.306 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 3 minutes 28.701 seconds)</p>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-deploy-models-deploy-object-detection-pytorch-py">
<div class="sphx-glr-download sphx-glr-download-python docutils container">
<p><a class="reference download internal" download="" href="../../_downloads/7795da4b258c8feff986668b95ef57ad/deploy_object_detection_pytorch.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">deploy_object_detection_pytorch.py</span></code></a></p>
diff --git a/docs/how_to/deploy_models/deploy_prequantized.html b/docs/how_to/deploy_models/deploy_prequantized.html
index 88806e237f..10419c7c81 100644
--- a/docs/how_to/deploy_models/deploy_prequantized.html
+++ b/docs/how_to/deploy_models/deploy_prequantized.html
@@ -500,8 +500,8 @@ training. Other models require a full post training calibration.</p>
Downloading: "https://download.pytorch.org/models/mobilenet_v2-b0353104.pth" to /workspace/.cache/torch/hub/checkpoints/mobilenet_v2-b0353104.pth
0%| | 0.00/13.6M [00:00<?, ?B/s]
- 59%|#####8 | 7.99M/13.6M [00:00<00:00, 37.8MB/s]
-100%|##########| 13.6M/13.6M [00:00<00:00, 49.3MB/s]
+ 59%|#####8 | 7.99M/13.6M [00:00<00:00, 78.7MB/s]
+100%|##########| 13.6M/13.6M [00:00<00:00, 85.4MB/s]
</pre></div>
</div>
</div>
@@ -592,7 +592,7 @@ output values are identical out of 1000 outputs from mobilenet v2.</p>
</div>
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Execution time summary:
mean (ms) median (ms) max (ms) min (ms) std (ms)
- 90.0792 90.0454 90.8264 89.8506 0.1604
+ 90.3108 90.1330 102.9750 89.7820 1.3601
</pre></div>
</div>
<div class="admonition note">
@@ -631,7 +631,7 @@ This includes support for the VNNI 8 bit dot product instruction (CascadeLake or
<div class="section" id="deploy-a-quantized-tflite-model">
<h2>Deploy a quantized TFLite Model<a class="headerlink" href="#deploy-a-quantized-tflite-model" title="Permalink to this headline">¶</a></h2>
<p>TODO</p>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 23.441 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 15.641 seconds)</p>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-deploy-models-deploy-prequantized-py">
<div class="sphx-glr-download sphx-glr-download-python docutils container">
<p><a class="reference download internal" download="" href="../../_downloads/fb8217c13f4351224c6cf3aacf1a87fc/deploy_prequantized.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">deploy_prequantized.py</span></code></a></p>
diff --git a/docs/how_to/deploy_models/deploy_prequantized_tflite.html b/docs/how_to/deploy_models/deploy_prequantized_tflite.html
index 81b0f93866..712bb5a717 100644
--- a/docs/how_to/deploy_models/deploy_prequantized_tflite.html
+++ b/docs/how_to/deploy_models/deploy_prequantized_tflite.html
@@ -585,7 +585,7 @@ TFLite Top-5 labels: [387 102 386 341 349]
</div>
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Execution time summary:
mean (ms) median (ms) max (ms) min (ms) std (ms)
- 117.3747 117.1252 123.4665 115.3360 1.2965
+ 118.6768 118.4045 121.8308 116.8414 1.0319
</pre></div>
</div>
<div class="admonition note">
@@ -613,7 +613,7 @@ network for ARM CPU</span></a>.</p></li>
</ul>
</div></blockquote>
</div>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 2 minutes 35.759 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 2 minutes 29.659 seconds)</p>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-deploy-models-deploy-prequantized-tflite-py">
<div class="sphx-glr-download sphx-glr-download-python docutils container">
<p><a class="reference download internal" download="" href="../../_downloads/56691c7a27d45da61d112276334640d3/deploy_prequantized_tflite.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">deploy_prequantized_tflite.py</span></code></a></p>
diff --git a/docs/how_to/deploy_models/deploy_quantized.html b/docs/how_to/deploy_models/deploy_quantized.html
index 68ddb24bad..9ba25e7d3f 100644
--- a/docs/how_to/deploy_models/deploy_quantized.html
+++ b/docs/how_to/deploy_models/deploy_quantized.html
@@ -526,7 +526,7 @@ for calibration. But the accuracy might be impacted.</p>
DeprecationWarning,
</pre></div>
</div>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 43.626 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 34.221 seconds)</p>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-deploy-models-deploy-quantized-py">
<div class="sphx-glr-download sphx-glr-download-python docutils container">
<p><a class="reference download internal" download="" href="../../_downloads/7810ecf51bfc05f7d5e8a400ac3e815d/deploy_quantized.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">deploy_quantized.py</span></code></a></p>
diff --git a/docs/how_to/deploy_models/deploy_ssd_gluoncv.html b/docs/how_to/deploy_models/deploy_ssd_gluoncv.html
index 37db47aaa7..2307c5f95d 100644
--- a/docs/how_to/deploy_models/deploy_ssd_gluoncv.html
+++ b/docs/how_to/deploy_models/deploy_ssd_gluoncv.html
@@ -468,23 +468,24 @@ to your device.</p>
Downloading /workspace/.mxnet/models/ssd_512_resnet50_v1_voc-9c8b225a.zip from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/ssd_512_resnet50_v1_voc-9c8b225a.zip...
0%| | 0/132723 [00:00<?, ?KB/s]
- 4%|3 | 5069/132723 [00:00<00:02, 50681.95KB/s]
- 10%|9 | 13250/132723 [00:00<00:01, 68989.48KB/s]
- 15%|#5 | 20150/132723 [00:00<00:01, 63403.97KB/s]
- 21%|##1 | 28398/132723 [00:00<00:01, 70549.80KB/s]
- 27%|##6 | 35521/132723 [00:00<00:02, 45403.86KB/s]
- 33%|###2 | 43717/132723 [00:00<00:01, 54254.78KB/s]
- 39%|###9 | 51912/132723 [00:00<00:01, 61372.09KB/s]
- 45%|####5 | 60122/132723 [00:00<00:01, 66936.75KB/s]
- 52%|#####1 | 68397/132723 [00:01<00:00, 71320.27KB/s]
- 58%|#####7 | 76670/132723 [00:01<00:00, 74555.92KB/s]
- 64%|######4 | 84979/132723 [00:01<00:00, 77015.80KB/s]
- 70%|####### | 93264/132723 [00:01<00:00, 78715.69KB/s]
- 76%|#######6 | 101347/132723 [00:01<00:00, 75423.52KB/s]
- 83%|########2 | 109619/132723 [00:01<00:00, 77504.71KB/s]
- 89%|########8 | 117503/132723 [00:01<00:00, 65667.40KB/s]
- 95%|#########4| 125700/132723 [00:01<00:00, 69882.50KB/s]
-100%|##########| 132723/132723 [00:01<00:00, 68132.15KB/s]
+ 5%|4 | 6208/132723 [00:00<00:02, 62072.22KB/s]
+ 10%|# | 13908/132723 [00:00<00:01, 70849.27KB/s]
+ 16%|#6 | 21588/132723 [00:00<00:01, 73564.38KB/s]
+ 22%|##2 | 29313/132723 [00:00<00:01, 75014.22KB/s]
+ 28%|##7 | 36979/132723 [00:00<00:01, 75605.60KB/s]
+ 34%|###3 | 44633/132723 [00:00<00:01, 75922.07KB/s]
+ 39%|###9 | 52295/132723 [00:00<00:01, 76148.00KB/s]
+ 45%|####5 | 60042/132723 [00:00<00:00, 76566.82KB/s]
+ 51%|#####1 | 67705/132723 [00:00<00:00, 76560.29KB/s]
+ 57%|#####6 | 75362/132723 [00:01<00:00, 76424.57KB/s]
+ 63%|######2 | 83005/132723 [00:01<00:00, 68789.41KB/s]
+ 68%|######8 | 90464/132723 [00:01<00:00, 70424.65KB/s]
+ 74%|#######3 | 98172/132723 [00:01<00:00, 72333.93KB/s]
+ 80%|#######9 | 105588/132723 [00:01<00:00, 72864.33KB/s]
+ 85%|########5 | 113207/132723 [00:01<00:00, 73838.60KB/s]
+ 91%|######### | 120638/132723 [00:01<00:00, 65562.52KB/s]
+ 97%|#########6| 128183/132723 [00:01<00:00, 68252.09KB/s]
+100%|##########| 132723/132723 [00:01<00:00, 72121.45KB/s]
</pre></div>
</div>
<p>Create TVM runtime and do inference
@@ -523,7 +524,7 @@ Downloading /workspace/.mxnet/models/ssd_512_resnet50_v1_voc-9c8b225a.zip from h
<span class="n">plt</span><span class="o">.</span><span class="n">show</span><span class="p">()</span>
</pre></div>
</div>
-<img src="../../_images/sphx_glr_deploy_ssd_gluoncv_001.png" srcset="../../_images/sphx_glr_deploy_ssd_gluoncv_001.png" alt="deploy ssd gluoncv" class = "sphx-glr-single-img"/><p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 4 minutes 8.321 seconds)</p>
+<img src="../../_images/sphx_glr_deploy_ssd_gluoncv_001.png" srcset="../../_images/sphx_glr_deploy_ssd_gluoncv_001.png" alt="deploy ssd gluoncv" class = "sphx-glr-single-img"/><p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 3 minutes 43.650 seconds)</p>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-deploy-models-deploy-ssd-gluoncv-py">
<div class="sphx-glr-download sphx-glr-download-python docutils container">
<p><a class="reference download internal" download="" href="../../_downloads/cccb17d28e5e8b2e94ea8cd5ec59f6ed/deploy_ssd_gluoncv.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">deploy_ssd_gluoncv.py</span></code></a></p>
diff --git a/docs/how_to/deploy_models/sg_execution_times.html b/docs/how_to/deploy_models/sg_execution_times.html
index 2b225f84de..3495232707 100644
--- a/docs/how_to/deploy_models/sg_execution_times.html
+++ b/docs/how_to/deploy_models/sg_execution_times.html
@@ -345,7 +345,7 @@
<div class="section" id="computation-times">
<span id="sphx-glr-how-to-deploy-models-sg-execution-times"></span><h1>Computation times<a class="headerlink" href="#computation-times" title="Permalink to this headline">¶</a></h1>
-<p><strong>17:09.664</strong> total execution time for <strong>how_to_deploy_models</strong> files:</p>
+<p><strong>15:54.268</strong> total execution time for <strong>how_to_deploy_models</strong> files:</p>
<table class="docutils align-default">
<colgroup>
<col style="width: 86%" />
@@ -354,43 +354,43 @@
</colgroup>
<tbody>
<tr class="row-odd"><td><p><a class="reference internal" href="deploy_ssd_gluoncv.html#sphx-glr-how-to-deploy-models-deploy-ssd-gluoncv-py"><span class="std std-ref">Deploy Single Shot Multibox Detector(SSD) model</span></a> (<code class="docutils literal notranslate"><span class="pre">deploy_ssd_gluoncv.py</span></code>)</p></td>
-<td><p>04:08.321</p></td>
+<td><p>03:43.650</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="deploy_object_detection_pytorch.html#sphx-glr-how-to-deploy-models-deploy-object-detection-pytorch-py"><span class="std std-ref">Compile PyTorch Object Detection Models</span></a> (<code class="docutils literal notranslate"><span class="pre">deploy_object_detection_pytorch.py</span></code>)</p></td>
-<td><p>03:48.306</p></td>
+<td><p>03:28.701</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="deploy_prequantized_tflite.html#sphx-glr-how-to-deploy-models-deploy-prequantized-tflite-py"><span class="std std-ref">Deploy a Framework-prequantized Model with TVM - Part 3 (TFLite)</span></a> (<code class="docutils literal notranslate"><span class="pre">deploy_prequantized_tflite.py</span></code>)</p></td>
-<td><p>02:35.759</p></td>
+<td><p>02:29.659</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="deploy_quantized.html#sphx-glr-how-to-deploy-models-deploy-quantized-py"><span class="std std-ref">Deploy a Quantized Model on Cuda</span></a> (<code class="docutils literal notranslate"><span class="pre">deploy_quantized.py</span></code>)</p></td>
-<td><p>01:43.626</p></td>
+<td><p>01:34.221</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="deploy_prequantized.html#sphx-glr-how-to-deploy-models-deploy-prequantized-py"><span class="std std-ref">Deploy a Framework-prequantized Model with TVM</span></a> (<code class="docutils literal notranslate"><span class="pre">deploy_prequantized.py</span></code>)</p></td>
-<td><p>01:23.441</p></td>
+<td><p>01:15.641</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="deploy_model_on_adreno.html#sphx-glr-how-to-deploy-models-deploy-model-on-adreno-py"><span class="std std-ref">Deploy the Pretrained Model on Adreno™</span></a> (<code class="docutils literal notranslate"><span class="pre">deploy_model_on_adreno.py</span></code>)</p></td>
-<td><p>00:54.517</p></td>
+<td><p>00:54.542</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="deploy_model_on_adreno_tvmc.html#sphx-glr-how-to-deploy-models-deploy-model-on-adreno-tvmc-py"><span class="std std-ref">Deploy the Pretrained Model on Adreno™ with tvmc Interface</span></a> (<code class="docutils literal notranslate"><span class="pre">deploy_model_on_adreno_tvmc.py</span></code>)</p></td>
-<td><p>00:50.879</p></td>
+<td><p>00:50.612</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="deploy_model_on_android.html#sphx-glr-how-to-deploy-models-deploy-model-on-android-py"><span class="std std-ref">Deploy the Pretrained Model on Android</span></a> (<code class="docutils literal notranslate"><span class="pre">deploy_model_on_android.py</span></code>)</p></td>
-<td><p>00:47.062</p></td>
+<td><p>00:41.939</p></td>
<td><p>0.0 MB</p></td>
</tr>
-<tr class="row-odd"><td><p><a class="reference internal" href="deploy_model_on_rasp.html#sphx-glr-how-to-deploy-models-deploy-model-on-rasp-py"><span class="std std-ref">Deploy the Pretrained Model on Raspberry Pi</span></a> (<code class="docutils literal notranslate"><span class="pre">deploy_model_on_rasp.py</span></code>)</p></td>
-<td><p>00:29.830</p></td>
+<tr class="row-odd"><td><p><a class="reference internal" href="deploy_model_on_nano.html#sphx-glr-how-to-deploy-models-deploy-model-on-nano-py"><span class="std std-ref">Deploy the Pretrained Model on Jetson Nano</span></a> (<code class="docutils literal notranslate"><span class="pre">deploy_model_on_nano.py</span></code>)</p></td>
+<td><p>00:27.808</p></td>
<td><p>0.0 MB</p></td>
</tr>
-<tr class="row-even"><td><p><a class="reference internal" href="deploy_model_on_nano.html#sphx-glr-how-to-deploy-models-deploy-model-on-nano-py"><span class="std std-ref">Deploy the Pretrained Model on Jetson Nano</span></a> (<code class="docutils literal notranslate"><span class="pre">deploy_model_on_nano.py</span></code>)</p></td>
-<td><p>00:27.917</p></td>
+<tr class="row-even"><td><p><a class="reference internal" href="deploy_model_on_rasp.html#sphx-glr-how-to-deploy-models-deploy-model-on-rasp-py"><span class="std std-ref">Deploy the Pretrained Model on Raspberry Pi</span></a> (<code class="docutils literal notranslate"><span class="pre">deploy_model_on_rasp.py</span></code>)</p></td>
+<td><p>00:27.491</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="deploy_sparse.html#sphx-glr-how-to-deploy-models-deploy-sparse-py"><span class="std std-ref">Deploy a Hugging Face Pruned Model on CPU</span></a> (<code class="docutils literal notranslate"><span class="pre">deploy_sparse.py</span></code>)</p></td>
diff --git a/docs/how_to/extend_tvm/bring_your_own_datatypes.html b/docs/how_to/extend_tvm/bring_your_own_datatypes.html
index abfa445e16..0826575feb 100644
--- a/docs/how_to/extend_tvm/bring_your_own_datatypes.html
+++ b/docs/how_to/extend_tvm/bring_your_own_datatypes.html
@@ -624,7 +624,7 @@ In this alpha state of the Bring Your Own Datatypes framework, we have not imple
<span class="n">module</span><span class="p">,</span> <a href="https://docs.python.org/3/library/stdtypes.html#dict" title="builtins.dict" class="sphx-glr-backref-module-builtins sphx-glr-backref-type-py-class sphx-glr-backref-instance"><span class="n">params</span></a> <span class="o">=</span> <span class="n">get_mobilenet</span><span class="p">()</span>
</pre></div>
</div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Downloading /workspace/.mxnet/models/mobilenet0.25-9f83e440.zipf00e4e20-651a-4595-b06a-fae9a812d3ab from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/mobilenet0.25-9f83e440.zip...
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Downloading /workspace/.mxnet/models/mobilenet0.25-9f83e440.zip156d0c7d-0023-4dbc-9bc2-6c10836a16c5 from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/mobilenet0.25-9f83e440.zip...
</pre></div>
</div>
<p>It’s easy to execute MobileNet with native TVM:</p>
diff --git a/docs/how_to/extend_tvm/sg_execution_times.html b/docs/how_to/extend_tvm/sg_execution_times.html
index 5a3f5c0c8c..3a780e73ed 100644
--- a/docs/how_to/extend_tvm/sg_execution_times.html
+++ b/docs/how_to/extend_tvm/sg_execution_times.html
@@ -345,7 +345,7 @@
<div class="section" id="computation-times">
<span id="sphx-glr-how-to-extend-tvm-sg-execution-times"></span><h1>Computation times<a class="headerlink" href="#computation-times" title="Permalink to this headline">¶</a></h1>
-<p><strong>00:54.808</strong> total execution time for <strong>how_to_extend_tvm</strong> files:</p>
+<p><strong>00:53.388</strong> total execution time for <strong>how_to_extend_tvm</strong> files:</p>
<table class="docutils align-default">
<colgroup>
<col style="width: 84%" />
@@ -354,19 +354,19 @@
</colgroup>
<tbody>
<tr class="row-odd"><td><p><a class="reference internal" href="bring_your_own_datatypes.html#sphx-glr-how-to-extend-tvm-bring-your-own-datatypes-py"><span class="std std-ref">Bring Your Own Datatypes to TVM</span></a> (<code class="docutils literal notranslate"><span class="pre">bring_your_own_datatypes.py</span></code>)</p></td>
-<td><p>00:50.656</p></td>
+<td><p>00:49.641</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="use_pass_instrument.html#sphx-glr-how-to-extend-tvm-use-pass-instrument-py"><span class="std std-ref">How to Use TVM Pass Instrument</span></a> (<code class="docutils literal notranslate"><span class="pre">use_pass_instrument.py</span></code>)</p></td>
-<td><p>00:03.027</p></td>
+<td><p>00:02.694</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="use_pass_infra.html#sphx-glr-how-to-extend-tvm-use-pass-infra-py"><span class="std std-ref">How to Use TVM Pass Infra</span></a> (<code class="docutils literal notranslate"><span class="pre">use_pass_infra.py</span></code>)</p></td>
-<td><p>00:01.118</p></td>
+<td><p>00:01.046</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="low_level_custom_pass.html#sphx-glr-how-to-extend-tvm-low-level-custom-pass-py"><span class="std std-ref">Writing a Customized Pass</span></a> (<code class="docutils literal notranslate"><span class="pre">low_level_custom_pass.py</span></code>)</p></td>
-<td><p>00:00.008</p></td>
+<td><p>00:00.007</p></td>
<td><p>0.0 MB</p></td>
</tr>
</tbody>
diff --git a/docs/how_to/extend_tvm/use_pass_instrument.html b/docs/how_to/extend_tvm/use_pass_instrument.html
index e91a15f008..65cc3c69c7 100644
--- a/docs/how_to/extend_tvm/use_pass_instrument.html
+++ b/docs/how_to/extend_tvm/use_pass_instrument.html
@@ -531,10 +531,10 @@ profile the execution time of each passes.</p>
</pre></div>
</div>
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Printing results of timing profile...
-InferType: 27356us [27356us] (48.95%; 48.95%)
-FoldScaleAxis: 28531us [7us] (51.05%; 51.05%)
- FoldConstant: 28524us [1702us] (51.04%; 99.97%)
- InferType: 26821us [26821us] (47.99%; 94.03%)
+InferType: 21921us [21921us] (48.72%; 48.72%)
+FoldScaleAxis: 23073us [7us] (51.28%; 51.28%)
+ FoldConstant: 23066us [1735us] (51.27%; 99.97%)
+ InferType: 21331us [21331us] (47.41%; 92.48%)
</pre></div>
</div>
</div>
@@ -556,10 +556,10 @@ Refer to following sections and <a class="reference internal" href="../../refere
</pre></div>
</div>
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Printing results of timing profile...
-InferType: 26704us [26704us] (48.31%; 48.31%)
-FoldScaleAxis: 28571us [6us] (51.69%; 51.69%)
- FoldConstant: 28564us [1746us] (51.68%; 99.98%)
- InferType: 26818us [26818us] (48.52%; 93.89%)
+InferType: 21382us [21382us] (47.49%; 47.49%)
+FoldScaleAxis: 23639us [5us] (52.51%; 52.51%)
+ FoldConstant: 23635us [1672us] (52.50%; 99.98%)
+ InferType: 21962us [21962us] (48.78%; 92.92%)
</pre></div>
</div>
<p>Register empty list to clear existing instruments.</p>
diff --git a/docs/how_to/optimize_operators/opt_conv_cuda.html b/docs/how_to/optimize_operators/opt_conv_cuda.html
index 54ca4c6bac..5e8398ebf6 100644
--- a/docs/how_to/optimize_operators/opt_conv_cuda.html
+++ b/docs/how_to/optimize_operators/opt_conv_cuda.html
@@ -580,7 +580,7 @@ latency of convolution.</p>
<span class="nb">print</span><span class="p">(</span><span class="s2">"Convolution: </span><span class="si">%f</span><span class="s2"> ms"</span> <span class="o">%</span> <span class="p">(</span><span class="n">evaluator</span><span class="p">(</span><span class="n">a</span><span class="p">,</span> <span class="n">w</span><span class="p">,</span> <span class="n">b</span><span class="p">)</span><span class="o">.</span><span class="n">mean</span> <span class="o">*</span> <span cl [...]
</pre></div>
</div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Convolution: 52.545440 ms
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Convolution: 54.152862 ms
</pre></div>
</div>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-optimize-operators-opt-conv-cuda-py">
diff --git a/docs/how_to/optimize_operators/opt_conv_tensorcore.html b/docs/how_to/optimize_operators/opt_conv_tensorcore.html
index 647c3dd538..0bb9388c84 100644
--- a/docs/how_to/optimize_operators/opt_conv_tensorcore.html
+++ b/docs/how_to/optimize_operators/opt_conv_tensorcore.html
@@ -862,7 +862,7 @@ be able to run on our build server</p>
<span class="nb">print</span><span class="p">(</span><span class="s2">"conv2d with tensor core: </span><span class="si">%f</span><span class="s2"> ms"</span> <span class="o">%</span> <span class="p">(</span><span class="n">evaluator</span><span class="p">(</span><span class="n">a</span><span class="p">,</span> <span class="n">w</span><span class="p">,</span> <span class="n">c</span><span class="p">)</span><span class="o">.</span><span class="n">mean</span> <span class="o">* [...]
</pre></div>
</div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>conv2d with tensor core: 13.357066 ms
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>conv2d with tensor core: 6.998237 ms
</pre></div>
</div>
</div>
diff --git a/docs/how_to/optimize_operators/opt_gemm.html b/docs/how_to/optimize_operators/opt_gemm.html
index e6ecd4567b..2206deccc1 100644
--- a/docs/how_to/optimize_operators/opt_gemm.html
+++ b/docs/how_to/optimize_operators/opt_gemm.html
@@ -477,8 +477,8 @@ Then we write a baseline implementation, the simplest way to write a matrix mult
<span class="nb">print</span><span class="p">(</span><span class="s2">"Baseline: </span><span class="si">%f</span><span class="s2">"</span> <span class="o">%</span> <span class="n">evaluator</span><span class="p">(</span><span class="n">a</span><span class="p">,</span> <span class="n">b</span><span class="p">,</span> <span class="n">c</span><span class="p">)</span><span class="o">.</span><span class="n">mean</span><span class="p">)</span>
</pre></div>
</div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Numpy running time: 0.020222
-Baseline: 3.602864
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Numpy running time: 0.017828
+Baseline: 3.281359
</pre></div>
</div>
<p>In TVM, we can always inspect lower level IR to debug or optimize our schedule.
@@ -537,7 +537,7 @@ fill 32 * 32 * sizeof(float) which is 4KB in the cache whose total size is 32KB
<span class="nb">print</span><span class="p">(</span><span class="s2">"Opt1: </span><span class="si">%f</span><span class="s2">"</span> <span class="o">%</span> <span class="n">evaluator</span><span class="p">(</span><span class="n">a</span><span class="p">,</span> <span class="n">b</span><span class="p">,</span> <span class="n">c</span><span class="p">)</span><span class="o">.</span><span class="n">mean</span><span class="p">)</span>
</pre></div>
</div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt1: 0.332328
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt1: 0.307110
</pre></div>
</div>
<p>Here is the generated IR after blocking.</p>
@@ -594,7 +594,7 @@ vastly.</p>
<span class="nb">print</span><span class="p">(</span><span class="s2">"Opt2: </span><span class="si">%f</span><span class="s2">"</span> <span class="o">%</span> <span class="n">evaluator</span><span class="p">(</span><span class="n">a</span><span class="p">,</span> <span class="n">b</span><span class="p">,</span> <span class="n">c</span><span class="p">)</span><span class="o">.</span><span class="n">mean</span><span class="p">)</span>
</pre></div>
</div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt2: 0.358482
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt2: 0.330430
</pre></div>
</div>
<p>Here is the generated IR after vectorization.</p>
@@ -649,7 +649,7 @@ the access pattern for A matrix is more cache friendly.</p>
<span class="nb">print</span><span class="p">(</span><span class="s2">"Opt3: </span><span class="si">%f</span><span class="s2">"</span> <span class="o">%</span> <span class="n">evaluator</span><span class="p">(</span><span class="n">a</span><span class="p">,</span> <span class="n">b</span><span class="p">,</span> <span class="n">c</span><span class="p">)</span><span class="o">.</span><span class="n">mean</span><span class="p">)</span>
</pre></div>
</div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt3: 0.136659
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt3: 0.113476
</pre></div>
</div>
<p>Here is the generated IR after loop permutation.</p>
@@ -726,7 +726,7 @@ flattening.</p>
<span class="nb">print</span><span class="p">(</span><span class="s2">"Opt4: </span><span class="si">%f</span><span class="s2">"</span> <span class="o">%</span> <span class="n">evaluator</span><span class="p">(</span><span class="n">a</span><span class="p">,</span> <span class="n">b</span><span class="p">,</span> <span class="n">c</span><span class="p">)</span><span class="o">.</span><span class="n">mean</span><span class="p">)</span>
</pre></div>
</div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt4: 0.112330
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt4: 0.109412
</pre></div>
</div>
<p>Here is the generated IR after array packing.</p>
@@ -804,7 +804,7 @@ write to C when all the block results are ready.</p>
<span class="nb">print</span><span class="p">(</span><span class="s2">"Opt5: </span><span class="si">%f</span><span class="s2">"</span> <span class="o">%</span> <span class="n">evaluator</span><span class="p">(</span><span class="n">a</span><span class="p">,</span> <span class="n">b</span><span class="p">,</span> <span class="n">c</span><span class="p">)</span><span class="o">.</span><span class="n">mean</span><span class="p">)</span>
</pre></div>
</div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt5: 0.113374
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt5: 0.111865
</pre></div>
</div>
<p>Here is the generated IR after blocking.</p>
@@ -884,7 +884,7 @@ class Module:
<span class="nb">print</span><span class="p">(</span><span class="s2">"Opt6: </span><span class="si">%f</span><span class="s2">"</span> <span class="o">%</span> <span class="n">opt6_time</span><span class="p">)</span>
</pre></div>
</div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt6: 0.148179
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt6: 0.146553
</pre></div>
</div>
<p>Here is the generated IR after parallelization.</p>
diff --git a/docs/how_to/optimize_operators/sg_execution_times.html b/docs/how_to/optimize_operators/sg_execution_times.html
index 57160db87a..b2f03f5cbb 100644
--- a/docs/how_to/optimize_operators/sg_execution_times.html
+++ b/docs/how_to/optimize_operators/sg_execution_times.html
@@ -345,7 +345,7 @@
<div class="section" id="computation-times">
<span id="sphx-glr-how-to-optimize-operators-sg-execution-times"></span><h1>Computation times<a class="headerlink" href="#computation-times" title="Permalink to this headline">¶</a></h1>
-<p><strong>00:36.875</strong> total execution time for <strong>how_to_optimize_operators</strong> files:</p>
+<p><strong>00:34.451</strong> total execution time for <strong>how_to_optimize_operators</strong> files:</p>
<table class="docutils align-default">
<colgroup>
<col style="width: 83%" />
@@ -354,15 +354,15 @@
</colgroup>
<tbody>
<tr class="row-odd"><td><p><a class="reference internal" href="opt_gemm.html#sphx-glr-how-to-optimize-operators-opt-gemm-py"><span class="std std-ref">How to optimize GEMM on CPU</span></a> (<code class="docutils literal notranslate"><span class="pre">opt_gemm.py</span></code>)</p></td>
-<td><p>00:34.125</p></td>
+<td><p>00:31.797</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="opt_conv_tensorcore.html#sphx-glr-how-to-optimize-operators-opt-conv-tensorcore-py"><span class="std std-ref">How to optimize convolution using TensorCores</span></a> (<code class="docutils literal notranslate"><span class="pre">opt_conv_tensorcore.py</span></code>)</p></td>
-<td><p>00:01.630</p></td>
+<td><p>00:01.529</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="opt_conv_cuda.html#sphx-glr-how-to-optimize-operators-opt-conv-cuda-py"><span class="std std-ref">How to optimize convolution on GPU</span></a> (<code class="docutils literal notranslate"><span class="pre">opt_conv_cuda.py</span></code>)</p></td>
-<td><p>00:01.120</p></td>
+<td><p>00:01.126</p></td>
<td><p>0.0 MB</p></td>
</tr>
</tbody>
diff --git a/docs/how_to/tune_with_autoscheduler/sg_execution_times.html b/docs/how_to/tune_with_autoscheduler/sg_execution_times.html
index d74ea222c9..433237f4ce 100644
--- a/docs/how_to/tune_with_autoscheduler/sg_execution_times.html
+++ b/docs/how_to/tune_with_autoscheduler/sg_execution_times.html
@@ -345,7 +345,7 @@
<div class="section" id="computation-times">
<span id="sphx-glr-how-to-tune-with-autoscheduler-sg-execution-times"></span><h1>Computation times<a class="headerlink" href="#computation-times" title="Permalink to this headline">¶</a></h1>
-<p><strong>10:53.147</strong> total execution time for <strong>how_to_tune_with_autoscheduler</strong> files:</p>
+<p><strong>10:00.162</strong> total execution time for <strong>how_to_tune_with_autoscheduler</strong> files:</p>
<table class="docutils align-default">
<colgroup>
<col style="width: 85%" />
@@ -354,27 +354,27 @@
</colgroup>
<tbody>
<tr class="row-odd"><td><p><a class="reference internal" href="tune_conv2d_layer_cuda.html#sphx-glr-how-to-tune-with-autoscheduler-tune-conv2d-layer-cuda-py"><span class="std std-ref">Auto-scheduling a Convolution Layer for GPU</span></a> (<code class="docutils literal notranslate"><span class="pre">tune_conv2d_layer_cuda.py</span></code>)</p></td>
-<td><p>06:47.248</p></td>
+<td><p>06:12.870</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="tune_network_x86.html#sphx-glr-how-to-tune-with-autoscheduler-tune-network-x86-py"><span class="std std-ref">Auto-scheduling a Neural Network for x86 CPU</span></a> (<code class="docutils literal notranslate"><span class="pre">tune_network_x86.py</span></code>)</p></td>
-<td><p>01:48.279</p></td>
+<td><p>01:40.793</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="tune_network_cuda.html#sphx-glr-how-to-tune-with-autoscheduler-tune-network-cuda-py"><span class="std std-ref">Auto-scheduling a Neural Network for NVIDIA GPU</span></a> (<code class="docutils literal notranslate"><span class="pre">tune_network_cuda.py</span></code>)</p></td>
-<td><p>01:11.065</p></td>
+<td><p>01:06.524</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="tune_sparse_x86.html#sphx-glr-how-to-tune-with-autoscheduler-tune-sparse-x86-py"><span class="std std-ref">Auto-scheduling Sparse Matrix Multiplication on CPU with Custom Sketch Rule</span></a> (<code class="docutils literal notranslate"><span class="pre">tune_sparse_x86.py</span></code>)</p></td>
-<td><p>00:36.619</p></td>
+<td><p>00:32.697</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="tune_network_arm.html#sphx-glr-how-to-tune-with-autoscheduler-tune-network-arm-py"><span class="std std-ref">Auto-scheduling a Neural Network for ARM CPU</span></a> (<code class="docutils literal notranslate"><span class="pre">tune_network_arm.py</span></code>)</p></td>
-<td><p>00:15.794</p></td>
+<td><p>00:13.965</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="tune_network_mali.html#sphx-glr-how-to-tune-with-autoscheduler-tune-network-mali-py"><span class="std std-ref">Auto-scheduling a Neural Network for mali GPU</span></a> (<code class="docutils literal notranslate"><span class="pre">tune_network_mali.py</span></code>)</p></td>
-<td><p>00:14.143</p></td>
+<td><p>00:13.313</p></td>
<td><p>0.0 MB</p></td>
</tr>
</tbody>
diff --git a/docs/how_to/tune_with_autoscheduler/tune_conv2d_layer_cuda.html b/docs/how_to/tune_with_autoscheduler/tune_conv2d_layer_cuda.html
index eb1814e6b2..b2a0585579 100644
--- a/docs/how_to/tune_with_autoscheduler/tune_conv2d_layer_cuda.html
+++ b/docs/how_to/tune_with_autoscheduler/tune_conv2d_layer_cuda.html
@@ -510,32 +510,1316 @@ class Module:
@T.prim_func
def main(data: T.Buffer((1, 512, 7, 7), "float32"), kernel: T.Buffer((512, 512, 3, 3), "float32"), bias: T.Buffer((1, 512, 1, 1), "float32"), compute: T.Buffer((1, 512, 7, 7), "float32")):
T.func_attr({"from_legacy_te_schedule": True, "global_symbol": "main", "tir.noalias": True})
- blockIdx_x = T.launch_thread("blockIdx.x", 16)
- conv2d_nchw = T.allocate([2], "float32", "local")
- pad_temp_shared = T.allocate([784], "float32", "shared")
- kernel_shared = T.allocate([512], "float32", "shared")
- threadIdx_x = T.launch_thread("threadIdx.x", 784)
- conv2d_nchw_1 = T.Buffer((2,), data=conv2d_nchw, scope="local", align=8)
- for ff_inner_init in range(2):
- conv2d_nchw_1[ff_inner_init] = T.float32(0)
- for rc_outer_outer, ry_outer_outer, rx_outer_outer in T.grid(32, 3, 3):
- pad_temp_shared_1 = T.Buffer((784,), data=pad_temp_shared, scope="shared")
- with T.launch_thread("threadIdx.x", 784) as threadIdx_x_1:
- for ax0_ax1_fused_ax2_fused_ax3_fused_inner_s in range(4):
- if T.likely(threadIdx_x_1 < 196):
- data_1 = T.Buffer((25088,), data=data.data)
- pad_temp_shared_1[threadIdx_x_1 * 4 + ax0_ax1_fused_ax2_fused_ax3_fused_inner_s] = T.if_then_else(1 <= (threadIdx_x_1 * 4 + ax0_ax1_fused_ax2_fused_ax3_fused_inner_s) % 49 // 7 + ry_outer_outer and (threadIdx_x_1 * 4 + ax0_ax1_fused_ax2_fused_ax3_fused_inner_s) % 49 // 7 + ry_outer_outer < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 4 + ax0_ax1_fused_ax2_fused_ax3_fused_inner_s) % 7 and rx_outer_outer + (threadIdx_x_1 * 4 + ax0_ax1_fused_ax2_fused_ax3_fu [...]
- kernel_shared_1 = T.Buffer((512,), data=kernel_shared, scope="shared")
- with T.launch_thread("threadIdx.x", 784) as threadIdx_x_1:
- if T.likely(threadIdx_x_1 < 512):
- kernel_1 = T.Buffer((2359296,), data=kernel.data)
- kernel_shared_1[threadIdx_x_1] = kernel_1[blockIdx_x * 147456 + threadIdx_x_1 // 16 * 4608 + rc_outer_outer * 144 + threadIdx_x_1 % 16 * 9 + ry_outer_outer * 3 + rx_outer_outer]
- for rc_outer_inner, ff_inner in T.grid(16, 2):
- conv2d_nchw_1[ff_inner] = conv2d_nchw_1[ff_inner] + pad_temp_shared_1[rc_outer_inner * 49 + threadIdx_x % 49] * kernel_shared_1[threadIdx_x // 49 * 32 + ff_inner * 16 + rc_outer_inner]
+ blockIdx_x = T.launch_thread("blockIdx.x", 32)
+ conv2d_nchw = T.allocate([14], "float32", "local")
+ pad_temp_shared = T.allocate([2592], "float32", "shared")
+ kernel_shared = T.allocate([4608], "float32", "shared")
+ threadIdx_x = T.launch_thread("threadIdx.x", 56)
+ conv2d_nchw_1 = T.Buffer((4,), data=conv2d_nchw, scope="local", align=8)
+ conv2d_nchw_1[0] = T.float32(0)
+ conv2d_nchw_1[2] = T.float32(0)
+ conv2d_nchw_1[4] = T.float32(0)
+ conv2d_nchw_1[6] = T.float32(0)
+ conv2d_nchw_1[8] = T.float32(0)
+ conv2d_nchw_1[10] = T.float32(0)
+ conv2d_nchw_1[12] = T.float32(0)
+ conv2d_nchw_1[1] = T.float32(0)
+ conv2d_nchw_1[3] = T.float32(0)
+ conv2d_nchw_1[5] = T.float32(0)
+ conv2d_nchw_1[7] = T.float32(0)
+ conv2d_nchw_1[9] = T.float32(0)
+ conv2d_nchw_1[11] = T.float32(0)
+ conv2d_nchw_1[13] = T.float32(0)
+ for rc_outer_outer in range(16):
+ cse_var_2: T.int32 = rc_outer_outer * 1568
+ cse_var_1: T.int32 = rc_outer_outer * 288
+ threadIdx_x_1 = T.env_thread("threadIdx.x")
+ pad_temp_shared_1 = T.Buffer((2592,), data=pad_temp_shared, scope="shared")
+ data_1 = T.Buffer((25088,), data=data.data)
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1] = T.if_then_else(9 <= threadIdx_x_1 and 1 <= threadIdx_x_1 % 9 and threadIdx_x_1 % 9 < 8, data_1[cse_var_2 + threadIdx_x_1 // 9 * 7 + threadIdx_x_1 % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 56] = T.if_then_else(9 <= (threadIdx_x_1 + 56) % 81 and (threadIdx_x_1 + 56) % 81 < 72 and 1 <= (threadIdx_x_1 + 2) % 9 and (threadIdx_x_1 + 2) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 56) // 81 * 49 + (threadIdx_x_1 + 56) % 81 // 9 * 7 + (threadIdx_x_1 + 2) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 112] = T.if_then_else(9 <= (threadIdx_x_1 + 31) % 81 and (threadIdx_x_1 + 31) % 81 < 72 and 1 <= (threadIdx_x_1 + 4) % 9 and (threadIdx_x_1 + 4) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 112) // 81 * 49 + (threadIdx_x_1 + 31) % 81 // 9 * 7 + (threadIdx_x_1 + 4) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 168] = T.if_then_else(3 <= threadIdx_x_1 and 1 <= (threadIdx_x_1 + 6) % 9 and (threadIdx_x_1 + 6) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 168) // 81 * 49 + (threadIdx_x_1 + 6) // 9 * 7 + (threadIdx_x_1 + 6) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 224] = T.if_then_else(9 <= (threadIdx_x_1 + 62) % 81 and (threadIdx_x_1 + 62) % 81 < 72 and 1 <= (threadIdx_x_1 + 8) % 9 and (threadIdx_x_1 + 8) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 224) // 81 * 49 + (threadIdx_x_1 + 62) % 81 // 9 * 7 + (threadIdx_x_1 + 8) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 280] = T.if_then_else(9 <= (threadIdx_x_1 + 37) % 81 and (threadIdx_x_1 + 37) % 81 < 72 and 1 <= (threadIdx_x_1 + 1) % 9 and (threadIdx_x_1 + 1) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 280) // 81 * 49 + (threadIdx_x_1 + 37) % 81 // 9 * 7 + (threadIdx_x_1 + 1) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 336] = T.if_then_else(1 <= (threadIdx_x_1 + 3) % 9 and (threadIdx_x_1 + 3) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 336) // 81 * 49 + (threadIdx_x_1 + 12) // 9 * 7 + (threadIdx_x_1 + 3) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 392] = T.if_then_else(9 <= (threadIdx_x_1 + 68) % 81 and (threadIdx_x_1 + 68) % 81 < 72 and 1 <= (threadIdx_x_1 + 5) % 9 and (threadIdx_x_1 + 5) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 392) // 81 * 49 + (threadIdx_x_1 + 68) % 81 // 9 * 7 + (threadIdx_x_1 + 5) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 448] = T.if_then_else(9 <= (threadIdx_x_1 + 43) % 81 and (threadIdx_x_1 + 43) % 81 < 72 and 1 <= (threadIdx_x_1 + 7) % 9 and (threadIdx_x_1 + 7) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 448) // 81 * 49 + (threadIdx_x_1 + 43) % 81 // 9 * 7 + (threadIdx_x_1 + 7) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 504] = T.if_then_else(threadIdx_x_1 < 54 and 1 <= threadIdx_x_1 % 9 and threadIdx_x_1 % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 504) // 81 * 49 + threadIdx_x_1 // 9 * 7 + threadIdx_x_1 % 9 + 6], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 560] = T.if_then_else(9 <= (threadIdx_x_1 + 74) % 81 and (threadIdx_x_1 + 74) % 81 < 72 and 1 <= (threadIdx_x_1 + 2) % 9 and (threadIdx_x_1 + 2) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 560) // 81 * 49 + (threadIdx_x_1 + 74) % 81 // 9 * 7 + (threadIdx_x_1 + 2) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 616] = T.if_then_else(9 <= (threadIdx_x_1 + 49) % 81 and (threadIdx_x_1 + 49) % 81 < 72 and 1 <= (threadIdx_x_1 + 4) % 9 and (threadIdx_x_1 + 4) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 616) // 81 * 49 + (threadIdx_x_1 + 49) % 81 // 9 * 7 + (threadIdx_x_1 + 4) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 672] = T.if_then_else(threadIdx_x_1 < 48 and 1 <= (threadIdx_x_1 + 6) % 9 and (threadIdx_x_1 + 6) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 672) // 81 * 49 + (threadIdx_x_1 + 24) // 9 * 7 + (threadIdx_x_1 + 6) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 728] = T.if_then_else(9 <= (threadIdx_x_1 + 80) % 81 and (threadIdx_x_1 + 80) % 81 < 72 and 1 <= (threadIdx_x_1 + 8) % 9 and (threadIdx_x_1 + 8) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 728) // 81 * 49 + (threadIdx_x_1 + 80) % 81 // 9 * 7 + (threadIdx_x_1 + 8) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 784] = T.if_then_else(9 <= (threadIdx_x_1 + 55) % 81 and (threadIdx_x_1 + 55) % 81 < 72 and 1 <= (threadIdx_x_1 + 1) % 9 and (threadIdx_x_1 + 1) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 784) // 81 * 49 + (threadIdx_x_1 + 55) % 81 // 9 * 7 + (threadIdx_x_1 + 1) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 840] = T.if_then_else(9 <= (threadIdx_x_1 + 30) % 81 and (threadIdx_x_1 + 30) % 81 < 72 and 1 <= (threadIdx_x_1 + 3) % 9 and (threadIdx_x_1 + 3) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 840) // 81 * 49 + (threadIdx_x_1 + 30) % 81 // 9 * 7 + (threadIdx_x_1 + 3) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 896] = T.if_then_else(4 <= threadIdx_x_1 and 1 <= (threadIdx_x_1 + 5) % 9 and (threadIdx_x_1 + 5) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 896) // 81 * 49 + (threadIdx_x_1 + 5) // 9 * 7 + (threadIdx_x_1 + 5) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 952] = T.if_then_else(9 <= (threadIdx_x_1 + 61) % 81 and (threadIdx_x_1 + 61) % 81 < 72 and 1 <= (threadIdx_x_1 + 7) % 9 and (threadIdx_x_1 + 7) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 952) // 81 * 49 + (threadIdx_x_1 + 61) % 81 // 9 * 7 + (threadIdx_x_1 + 7) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 1008] = T.if_then_else(1 <= (threadIdx_x_1 // 9 + 4) % 9 and (threadIdx_x_1 + 36) % 81 < 72 and 1 <= threadIdx_x_1 % 9 and threadIdx_x_1 % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 1008) // 81 * 49 + (threadIdx_x_1 // 9 + 4) % 9 * 7 + threadIdx_x_1 % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 1064] = T.if_then_else(1 <= (threadIdx_x_1 + 2) % 9 and (threadIdx_x_1 + 2) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 1064) // 81 * 49 + (threadIdx_x_1 + 11) // 9 * 7 + (threadIdx_x_1 + 2) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 1120] = T.if_then_else(9 <= (threadIdx_x_1 + 67) % 81 and (threadIdx_x_1 + 67) % 81 < 72 and 1 <= (threadIdx_x_1 + 4) % 9 and (threadIdx_x_1 + 4) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 1120) // 81 * 49 + (threadIdx_x_1 + 67) % 81 // 9 * 7 + (threadIdx_x_1 + 4) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 1176] = T.if_then_else(9 <= (threadIdx_x_1 + 42) % 81 and (threadIdx_x_1 + 42) % 81 < 72 and 1 <= (threadIdx_x_1 + 6) % 9 and (threadIdx_x_1 + 6) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 1176) // 81 * 49 + (threadIdx_x_1 + 42) % 81 // 9 * 7 + (threadIdx_x_1 + 6) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 1232] = T.if_then_else(threadIdx_x_1 < 55 and 1 <= (threadIdx_x_1 + 8) % 9 and (threadIdx_x_1 + 8) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 1232) // 81 * 49 + (threadIdx_x_1 + 17) // 9 * 7 + (threadIdx_x_1 + 8) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 1288] = T.if_then_else(9 <= (threadIdx_x_1 + 73) % 81 and (threadIdx_x_1 + 73) % 81 < 72 and 1 <= (threadIdx_x_1 + 1) % 9 and (threadIdx_x_1 + 1) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 1288) // 81 * 49 + (threadIdx_x_1 + 73) % 81 // 9 * 7 + (threadIdx_x_1 + 1) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 1344] = T.if_then_else(9 <= (threadIdx_x_1 + 48) % 81 and (threadIdx_x_1 + 48) % 81 < 72 and 1 <= (threadIdx_x_1 + 3) % 9 and (threadIdx_x_1 + 3) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 1344) // 81 * 49 + (threadIdx_x_1 + 48) % 81 // 9 * 7 + (threadIdx_x_1 + 3) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 1400] = T.if_then_else(threadIdx_x_1 < 49 and 1 <= (threadIdx_x_1 + 5) % 9 and (threadIdx_x_1 + 5) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 1400) // 81 * 49 + (threadIdx_x_1 + 23) // 9 * 7 + (threadIdx_x_1 + 5) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 1456] = T.if_then_else(9 <= (threadIdx_x_1 + 79) % 81 and (threadIdx_x_1 + 79) % 81 < 72 and 1 <= (threadIdx_x_1 + 7) % 9 and (threadIdx_x_1 + 7) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 1456) // 81 * 49 + (threadIdx_x_1 + 79) % 81 // 9 * 7 + (threadIdx_x_1 + 7) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 1512] = T.if_then_else(1 <= (threadIdx_x_1 // 9 + 6) % 9 and (threadIdx_x_1 + 54) % 81 < 72 and 1 <= threadIdx_x_1 % 9 and threadIdx_x_1 % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 1512) // 81 * 49 + (threadIdx_x_1 // 9 + 6) % 9 * 7 + threadIdx_x_1 % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 1568] = T.if_then_else(9 <= (threadIdx_x_1 + 29) % 81 and (threadIdx_x_1 + 29) % 81 < 72 and 1 <= (threadIdx_x_1 + 2) % 9 and (threadIdx_x_1 + 2) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 1568) // 81 * 49 + (threadIdx_x_1 + 29) % 81 // 9 * 7 + (threadIdx_x_1 + 2) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 1624] = T.if_then_else(5 <= threadIdx_x_1 and 1 <= (threadIdx_x_1 + 4) % 9 and (threadIdx_x_1 + 4) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 1624) // 81 * 49 + (threadIdx_x_1 + 4) // 9 * 7 + (threadIdx_x_1 + 4) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 1680] = T.if_then_else(9 <= (threadIdx_x_1 + 60) % 81 and (threadIdx_x_1 + 60) % 81 < 72 and 1 <= (threadIdx_x_1 + 6) % 9 and (threadIdx_x_1 + 6) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 1680) // 81 * 49 + (threadIdx_x_1 + 60) % 81 // 9 * 7 + (threadIdx_x_1 + 6) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 1736] = T.if_then_else(9 <= (threadIdx_x_1 + 35) % 81 and (threadIdx_x_1 + 35) % 81 < 72 and 1 <= (threadIdx_x_1 + 8) % 9 and (threadIdx_x_1 + 8) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 1736) // 81 * 49 + (threadIdx_x_1 + 35) % 81 // 9 * 7 + (threadIdx_x_1 + 8) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 1792] = T.if_then_else(1 <= (threadIdx_x_1 + 1) % 9 and (threadIdx_x_1 + 1) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 1792) // 81 * 49 + (threadIdx_x_1 + 10) // 9 * 7 + (threadIdx_x_1 + 1) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 1848] = T.if_then_else(9 <= (threadIdx_x_1 + 66) % 81 and (threadIdx_x_1 + 66) % 81 < 72 and 1 <= (threadIdx_x_1 + 3) % 9 and (threadIdx_x_1 + 3) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 1848) // 81 * 49 + (threadIdx_x_1 + 66) % 81 // 9 * 7 + (threadIdx_x_1 + 3) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 1904] = T.if_then_else(9 <= (threadIdx_x_1 + 41) % 81 and (threadIdx_x_1 + 41) % 81 < 72 and 1 <= (threadIdx_x_1 + 5) % 9 and (threadIdx_x_1 + 5) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 1904) // 81 * 49 + (threadIdx_x_1 + 41) % 81 // 9 * 7 + (threadIdx_x_1 + 5) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 1960] = T.if_then_else(1 <= (threadIdx_x_1 + 7) % 9 and (threadIdx_x_1 + 7) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 1960) // 81 * 49 + (threadIdx_x_1 + 16) // 9 * 7 + (threadIdx_x_1 + 7) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 2016] = T.if_then_else(1 <= (threadIdx_x_1 // 9 + 8) % 9 and (threadIdx_x_1 + 72) % 81 < 72 and 1 <= threadIdx_x_1 % 9 and threadIdx_x_1 % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 2016) // 81 * 49 + (threadIdx_x_1 // 9 + 8) % 9 * 7 + threadIdx_x_1 % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 2072] = T.if_then_else(9 <= (threadIdx_x_1 + 47) % 81 and (threadIdx_x_1 + 47) % 81 < 72 and 1 <= (threadIdx_x_1 + 2) % 9 and (threadIdx_x_1 + 2) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 2072) // 81 * 49 + (threadIdx_x_1 + 47) % 81 // 9 * 7 + (threadIdx_x_1 + 2) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 2128] = T.if_then_else(threadIdx_x_1 < 50 and 1 <= (threadIdx_x_1 + 4) % 9 and (threadIdx_x_1 + 4) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 2128) // 81 * 49 + (threadIdx_x_1 + 22) // 9 * 7 + (threadIdx_x_1 + 4) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 2184] = T.if_then_else(9 <= (threadIdx_x_1 + 78) % 81 and (threadIdx_x_1 + 78) % 81 < 72 and 1 <= (threadIdx_x_1 + 6) % 9 and (threadIdx_x_1 + 6) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 2184) // 81 * 49 + (threadIdx_x_1 + 78) % 81 // 9 * 7 + (threadIdx_x_1 + 6) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 2240] = T.if_then_else(9 <= (threadIdx_x_1 + 53) % 81 and (threadIdx_x_1 + 53) % 81 < 72 and 1 <= (threadIdx_x_1 + 8) % 9 and (threadIdx_x_1 + 8) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 2240) // 81 * 49 + (threadIdx_x_1 + 53) % 81 // 9 * 7 + (threadIdx_x_1 + 8) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 2296] = T.if_then_else(9 <= (threadIdx_x_1 + 28) % 81 and (threadIdx_x_1 + 28) % 81 < 72 and 1 <= (threadIdx_x_1 + 1) % 9 and (threadIdx_x_1 + 1) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 2296) // 81 * 49 + (threadIdx_x_1 + 28) % 81 // 9 * 7 + (threadIdx_x_1 + 1) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 2352] = T.if_then_else(6 <= threadIdx_x_1 and 1 <= (threadIdx_x_1 + 3) % 9 and (threadIdx_x_1 + 3) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 2352) // 81 * 49 + (threadIdx_x_1 + 3) // 9 * 7 + (threadIdx_x_1 + 3) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 2408] = T.if_then_else(9 <= (threadIdx_x_1 + 59) % 81 and (threadIdx_x_1 + 59) % 81 < 72 and 1 <= (threadIdx_x_1 + 5) % 9 and (threadIdx_x_1 + 5) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 2408) // 81 * 49 + (threadIdx_x_1 + 59) % 81 // 9 * 7 + (threadIdx_x_1 + 5) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 2464] = T.if_then_else(9 <= (threadIdx_x_1 + 34) % 81 and (threadIdx_x_1 + 34) % 81 < 72 and 1 <= (threadIdx_x_1 + 7) % 9 and (threadIdx_x_1 + 7) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 2464) // 81 * 49 + (threadIdx_x_1 + 34) % 81 // 9 * 7 + (threadIdx_x_1 + 7) % 9 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 2520] = T.if_then_else(1 <= threadIdx_x_1 % 9 and threadIdx_x_1 % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 2520) // 81 * 49 + threadIdx_x_1 // 9 * 7 + threadIdx_x_1 % 9 - 1], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ if T.likely(threadIdx_x_1 < 16):
+ pad_temp_shared_1[threadIdx_x_1 + 2576] = T.if_then_else(threadIdx_x_1 < 7 and 1 <= (threadIdx_x_1 + 2) % 9 and (threadIdx_x_1 + 2) % 9 < 8, data_1[cse_var_2 + (threadIdx_x_1 + 2576) // 81 * 49 + (threadIdx_x_1 + 65) // 9 * 7 + threadIdx_x_1 - 6], T.float32(0))
+ threadIdx_x_2 = T.env_thread("threadIdx.x")
+ kernel_shared_1 = T.Buffer((4608,), data=kernel_shared, scope="shared")
+ kernel_1 = T.Buffer((2359296,), data=kernel.data)
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2] = kernel_1[blockIdx_x * 73728 + cse_var_1 + threadIdx_x_2]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 56] = kernel_1[blockIdx_x * 73728 + cse_var_1 + threadIdx_x_2 + 56]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 112] = kernel_1[blockIdx_x * 73728 + cse_var_1 + threadIdx_x_2 + 112]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 168] = kernel_1[blockIdx_x * 73728 + cse_var_1 + threadIdx_x_2 + 168]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 224] = kernel_1[blockIdx_x * 73728 + cse_var_1 + threadIdx_x_2 + 224]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 280] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 280) // 288 * 4608 + cse_var_1 + (threadIdx_x_2 + 280) % 288]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 336] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 336) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 48]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 392] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 392) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 104]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 448] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 448) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 160]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 504] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 504) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 216]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 560] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 560) // 288 * 4608 + cse_var_1 + (threadIdx_x_2 + 272) % 288]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 616] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 616) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 40]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 672] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 672) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 96]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 728] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 728) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 152]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 784] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 784) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 208]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 840] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 840) // 288 * 4608 + cse_var_1 + (threadIdx_x_2 + 264) % 288]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 896] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 896) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 32]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 952] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 952) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 88]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1008] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1008) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 144]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1064] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1064) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 200]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1120] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1120) // 288 * 4608 + cse_var_1 + (threadIdx_x_2 + 256) % 288]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1176] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1176) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 24]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1232] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1232) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 80]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1288] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1288) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 136]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1344] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1344) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 192]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1400] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1400) // 288 * 4608 + cse_var_1 + (threadIdx_x_2 + 248) % 288]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1456] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1456) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 16]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1512] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1512) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 72]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1568] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1568) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 128]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1624] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1624) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 184]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1680] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1680) // 288 * 4608 + cse_var_1 + (threadIdx_x_2 + 240) % 288]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1736] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1736) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 8]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1792] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1792) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 64]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1848] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1848) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 120]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1904] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1904) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 176]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1960] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1960) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 232]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 2016] = kernel_1[blockIdx_x * 73728 + cse_var_1 + threadIdx_x_2 + 32256]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 2072] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2072) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 56]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 2128] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2128) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 112]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 2184] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2184) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 168]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 2240] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2240) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 224]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 2296] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2296) // 288 * 4608 + cse_var_1 + (threadIdx_x_2 + 280) % 288]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 2352] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2352) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 48]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 2408] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2408) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 104]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 2464] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2464) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 160]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 2520] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2520) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 216]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 2576] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2576) // 288 * 4608 + cse_var_1 + (threadIdx_x_2 + 272) % 288]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 2632] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2632) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 40]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 2688] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2688) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 96]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 2744] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2744) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 152]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 2800] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2800) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 208]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 2856] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2856) // 288 * 4608 + cse_var_1 + (threadIdx_x_2 + 264) % 288]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 2912] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2912) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 32]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 2968] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2968) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 88]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 3024] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3024) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 144]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 3080] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3080) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 200]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 3136] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3136) // 288 * 4608 + cse_var_1 + (threadIdx_x_2 + 256) % 288]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 3192] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3192) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 24]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 3248] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3248) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 80]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 3304] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3304) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 136]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 3360] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3360) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 192]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 3416] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3416) // 288 * 4608 + cse_var_1 + (threadIdx_x_2 + 248) % 288]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 3472] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3472) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 16]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 3528] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3528) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 72]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 3584] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3584) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 128]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 3640] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3640) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 184]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 3696] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3696) // 288 * 4608 + cse_var_1 + (threadIdx_x_2 + 240) % 288]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 3752] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3752) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 8]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 3808] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3808) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 64]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 3864] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3864) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 120]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 3920] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3920) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 176]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 3976] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3976) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 232]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 4032] = kernel_1[blockIdx_x * 73728 + cse_var_1 + threadIdx_x_2 + 64512]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 4088] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 4088) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 56]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 4144] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 4144) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 112]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 4200] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 4200) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 168]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 4256] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 4256) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 224]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 4312] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 4312) // 288 * 4608 + cse_var_1 + (threadIdx_x_2 + 280) % 288]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 4368] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 4368) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 48]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 4424] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 4424) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 104]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 4480] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 4480) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 160]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 4536] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 4536) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 216]
+ with T.launch_thread(threadIdx_x_2, 56):
+ if T.likely(threadIdx_x_2 < 16):
+ kernel_shared_1[threadIdx_x_2 + 4592] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 4592) // 288 * 4608 + cse_var_1 + threadIdx_x_2 + 272]
+ for rc_outer_inner in range(4):
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 9] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 18] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 27] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 36] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 45] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 54] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 288]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 9] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 288]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 18] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 288]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 27] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 288]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 36] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 288]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 45] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 288]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 54] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 288]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 1] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 1]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 10] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 1]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 19] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 1]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 28] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 1]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 37] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 1]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 46] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 1]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 55] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 1]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 1] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 289]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 10] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 289]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 19] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 289]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 28] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 289]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 37] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 289]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 46] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 289]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 55] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 289]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 2] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 2]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 11] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 2]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 20] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 2]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 29] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 2]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 38] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 2]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 47] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 2]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 56] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 2]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 2] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 290]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 11] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 290]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 20] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 290]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 29] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 290]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 38] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 290]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 47] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 290]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 56] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 290]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 81] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 9]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 90] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 9]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 99] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 9]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 108] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 9]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 117] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 9]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 126] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 9]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 135] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 9]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 81] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 297]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 90] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 297]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 99] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 297]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 108] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 297]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 117] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 297]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 126] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 297]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 135] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 297]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 82] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 10]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 91] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 10]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 100] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 10]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 109] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 10]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 118] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 10]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 127] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 10]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 136] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 10]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 82] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 298]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 91] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 298]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 100] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 298]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 109] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 298]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 118] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 298]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 127] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 298]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 136] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 298]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 83] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 11]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 92] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 11]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 101] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 11]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 110] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 11]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 119] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 11]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 128] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 11]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 137] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 11]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 83] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 299]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 92] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 299]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 101] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 299]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 110] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 299]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 119] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 299]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 128] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 299]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 137] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 299]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 162] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 18]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 171] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 18]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 180] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 18]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 189] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 18]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 198] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 18]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 207] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 18]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 216] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 18]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 162] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 306]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 171] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 306]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 180] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 306]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 189] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 306]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 198] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 306]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 207] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 306]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 216] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 306]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 163] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 19]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 172] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 19]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 181] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 19]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 190] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 19]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 199] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 19]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 208] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 19]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 217] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 19]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 163] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 307]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 172] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 307]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 181] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 307]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 190] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 307]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 199] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 307]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 208] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 307]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 217] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 307]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 164] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 20]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 173] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 20]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 182] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 20]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 191] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 20]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 200] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 20]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 209] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 20]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 218] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 20]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 164] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 308]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 173] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 308]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 182] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 308]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 191] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 308]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 200] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 308]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 209] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 308]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 218] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 308]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 243] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 27]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 252] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 27]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 261] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 27]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 270] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 27]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 279] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 27]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 288] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 27]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 297] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 27]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 243] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 315]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 252] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 315]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 261] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 315]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 270] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 315]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 279] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 315]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 288] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 315]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 297] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 315]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 244] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 28]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 253] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 28]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 262] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 28]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 271] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 28]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 280] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 28]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 289] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 28]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 298] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 28]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 244] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 316]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 253] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 316]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 262] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 316]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 271] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 316]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 280] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 316]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 289] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 316]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 298] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 316]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 245] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 29]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 254] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 29]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 263] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 29]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 272] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 29]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 281] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 29]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 290] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 29]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 299] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 29]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 245] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 317]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 254] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 317]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 263] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 317]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 272] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 317]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 281] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 317]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 290] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 317]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 299] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 317]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 324] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 36]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 333] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 36]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 342] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 36]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 351] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 36]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 360] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 36]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 369] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 36]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 378] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 36]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 324] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 324]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 333] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 324]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 342] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 324]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 351] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 324]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 360] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 324]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 369] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 324]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 378] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 324]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 325] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 37]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 334] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 37]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 343] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 37]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 352] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 37]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 361] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 37]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 370] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 37]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 379] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 37]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 325] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 325]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 334] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 325]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 343] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 325]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 352] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 325]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 361] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 325]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 370] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 325]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 379] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 325]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 326] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 38]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 335] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 38]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 344] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 38]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 353] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 38]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 362] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 38]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 371] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 38]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 380] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 38]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 326] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 326]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 335] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 326]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 344] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 326]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 353] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 326]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 362] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 326]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 371] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 326]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 380] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 326]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 405] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 45]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 414] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 45]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 423] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 45]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 432] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 45]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 441] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 45]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 450] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 45]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 459] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 45]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 405] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 333]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 414] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 333]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 423] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 333]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 432] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 333]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 441] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 333]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 450] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 333]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 459] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 333]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 406] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 46]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 415] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 46]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 424] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 46]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 433] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 46]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 442] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 46]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 451] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 46]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 460] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 46]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 406] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 334]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 415] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 334]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 424] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 334]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 433] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 334]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 442] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 334]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 451] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 334]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 460] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 334]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 407] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 47]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 416] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 47]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 425] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 47]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 434] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 47]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 443] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 47]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 452] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 47]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 461] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 47]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 407] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 335]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 416] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 335]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 425] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 335]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 434] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 335]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 443] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 335]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 452] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 335]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 461] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 335]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 486] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 54]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 495] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 54]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 504] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 54]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 513] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 54]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 522] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 54]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 531] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 54]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 540] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 54]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 486] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 342]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 495] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 342]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 504] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 342]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 513] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 342]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 522] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 342]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 531] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 342]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 540] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 342]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 487] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 55]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 496] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 55]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 505] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 55]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 514] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 55]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 523] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 55]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 532] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 55]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 541] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 55]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 487] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 343]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 496] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 343]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 505] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 343]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 514] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 343]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 523] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 343]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 532] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 343]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 541] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 343]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 488] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 56]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 497] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 56]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 506] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 56]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 515] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 56]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 524] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 56]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 533] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 56]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 542] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 56]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 488] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 344]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 497] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 344]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 506] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 344]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 515] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 344]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 524] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 344]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 533] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 344]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 542] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 344]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 567] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 63]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 576] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 63]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 585] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 63]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 594] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 63]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 603] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 63]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 612] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 63]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 621] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 63]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 567] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 351]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 576] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 351]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 585] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 351]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 594] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 351]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 603] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 351]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 612] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 351]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 621] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 351]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 568] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 64]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 577] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 64]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 586] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 64]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 595] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 64]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 604] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 64]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 613] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 64]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 622] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 64]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 568] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 352]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 577] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 352]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 586] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 352]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 595] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 352]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 604] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 352]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 613] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 352]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 622] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 352]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 569] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 65]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 578] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 65]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 587] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 65]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 596] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 65]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 605] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 65]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 614] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 65]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 623] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 65]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 569] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 353]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 578] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 353]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 587] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 353]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 596] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 353]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 605] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 353]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 614] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 353]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 623] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 353]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 9] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 3]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 18] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 3]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 27] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 3]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 36] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 3]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 45] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 3]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 54] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 3]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 63] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 3]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 9] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 291]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 18] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 291]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 27] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 291]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 36] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 291]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 45] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 291]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 54] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 291]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 63] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 291]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 10] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 4]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 19] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 4]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 28] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 4]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 37] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 4]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 46] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 4]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 55] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 4]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 64] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 4]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 10] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 292]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 19] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 292]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 28] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 292]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 37] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 292]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 46] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 292]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 55] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 292]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 64] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 292]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 11] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 5]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 20] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 5]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 29] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 5]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 38] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 5]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 47] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 5]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 56] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 5]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 65] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 5]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 11] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 293]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 20] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 293]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 29] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 293]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 38] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 293]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 47] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 293]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 56] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 293]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 65] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 293]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 90] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 12]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 99] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 12]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 108] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 12]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 117] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 12]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 126] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 12]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 135] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 12]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 144] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 12]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 90] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 300]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 99] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 300]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 108] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 300]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 117] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 300]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 126] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 300]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 135] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 300]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 144] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 300]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 91] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 13]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 100] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 13]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 109] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 13]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 118] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 13]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 127] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 13]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 136] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 13]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 145] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 13]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 91] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 301]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 100] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 301]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 109] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 301]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 118] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 301]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 127] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 301]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 136] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 301]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 145] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 301]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 92] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 14]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 101] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 14]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 110] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 14]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 119] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 14]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 128] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 14]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 137] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 14]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 146] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 14]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 92] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 302]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 101] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 302]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 110] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 302]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 119] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 302]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 128] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 302]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 137] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 302]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 146] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 302]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 171] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 21]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 180] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 21]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 189] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 21]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 198] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 21]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 207] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 21]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 216] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 21]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 225] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 21]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 171] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 309]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 180] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 309]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 189] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 309]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 198] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 309]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 207] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 309]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 216] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 309]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 225] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 309]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 172] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 22]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 181] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 22]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 190] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 22]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 199] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 22]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 208] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 22]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 217] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 22]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 226] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 22]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 172] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 310]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 181] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 310]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 190] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 310]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 199] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 310]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 208] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 310]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 217] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 310]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 226] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 310]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 173] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 23]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 182] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 23]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 191] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 23]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 200] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 23]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 209] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 23]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 218] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 23]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 227] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 23]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 173] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 311]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 182] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 311]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 191] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 311]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 200] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 311]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 209] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 311]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 218] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 311]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 227] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 311]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 252] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 30]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 261] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 30]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 270] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 30]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 279] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 30]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 288] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 30]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 297] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 30]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 306] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 30]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 252] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 318]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 261] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 318]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 270] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 318]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 279] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 318]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 288] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 318]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 297] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 318]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 306] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 318]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 253] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 31]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 262] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 31]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 271] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 31]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 280] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 31]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 289] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 31]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 298] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 31]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 307] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 31]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 253] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 319]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 262] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 319]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 271] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 319]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 280] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 319]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 289] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 319]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 298] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 319]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 307] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 319]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 254] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 32]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 263] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 32]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 272] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 32]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 281] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 32]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 290] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 32]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 299] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 32]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 308] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 32]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 254] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 320]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 263] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 320]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 272] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 320]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 281] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 320]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 290] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 320]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 299] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 320]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 308] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 320]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 333] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 39]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 342] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 39]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 351] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 39]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 360] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 39]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 369] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 39]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 378] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 39]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 387] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 39]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 333] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 327]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 342] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 327]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 351] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 327]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 360] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 327]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 369] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 327]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 378] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 327]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 387] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 327]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 334] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 40]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 343] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 40]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 352] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 40]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 361] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 40]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 370] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 40]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 379] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 40]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 388] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 40]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 334] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 328]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 343] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 328]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 352] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 328]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 361] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 328]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 370] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 328]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 379] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 328]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 388] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 328]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 335] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 41]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 344] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 41]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 353] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 41]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 362] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 41]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 371] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 41]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 380] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 41]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 389] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 41]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 335] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 329]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 344] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 329]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 353] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 329]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 362] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 329]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 371] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 329]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 380] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 329]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 389] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 329]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 414] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 48]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 423] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 48]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 432] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 48]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 441] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 48]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 450] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 48]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 459] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 48]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 468] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 48]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 414] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 336]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 423] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 336]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 432] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 336]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 441] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 336]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 450] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 336]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 459] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 336]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 468] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 336]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 415] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 49]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 424] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 49]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 433] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 49]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 442] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 49]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 451] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 49]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 460] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 49]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 469] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 49]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 415] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 337]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 424] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 337]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 433] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 337]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 442] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 337]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 451] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 337]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 460] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 337]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 469] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 337]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 416] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 50]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 425] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 50]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 434] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 50]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 443] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 50]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 452] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 50]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 461] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 50]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 470] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 50]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 416] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 338]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 425] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 338]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 434] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 338]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 443] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 338]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 452] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 338]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 461] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 338]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 470] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 338]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 495] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 57]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 504] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 57]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 513] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 57]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 522] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 57]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 531] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 57]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 540] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 57]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 549] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 57]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 495] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 345]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 504] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 345]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 513] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 345]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 522] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 345]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 531] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 345]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 540] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 345]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 549] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 345]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 496] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 58]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 505] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 58]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 514] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 58]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 523] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 58]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 532] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 58]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 541] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 58]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 550] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 58]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 496] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 346]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 505] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 346]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 514] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 346]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 523] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 346]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 532] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 346]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 541] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 346]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 550] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 346]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 497] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 59]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 506] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 59]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 515] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 59]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 524] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 59]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 533] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 59]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 542] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 59]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 551] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 59]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 497] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 347]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 506] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 347]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 515] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 347]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 524] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 347]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 533] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 347]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 542] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 347]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 551] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 347]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 576] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 66]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 585] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 66]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 594] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 66]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 603] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 66]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 612] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 66]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 621] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 66]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 630] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 66]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 576] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 354]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 585] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 354]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 594] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 354]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 603] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 354]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 612] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 354]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 621] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 354]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 630] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 354]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 577] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 67]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 586] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 67]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 595] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 67]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 604] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 67]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 613] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 67]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 622] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 67]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 631] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 67]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 577] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 355]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 586] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 355]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 595] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 355]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 604] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 355]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 613] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 355]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 622] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 355]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 631] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 355]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 578] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 68]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 587] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 68]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 596] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 68]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 605] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 68]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 614] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 68]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 623] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 68]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 632] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 68]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 578] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 356]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 587] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 356]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 596] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 356]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 605] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 356]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 614] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 356]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 623] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 356]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 632] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 356]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 18] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 6]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 27] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 6]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 36] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 6]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 45] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 6]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 54] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 6]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 63] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 6]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 72] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 6]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 18] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 294]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 27] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 294]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 36] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 294]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 45] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 294]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 54] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 294]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 63] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 294]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 72] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 294]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 19] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 7]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 28] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 7]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 37] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 7]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 46] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 7]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 55] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 7]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 64] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 7]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 73] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 7]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 19] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 295]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 28] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 295]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 37] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 295]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 46] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 295]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 55] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 295]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 64] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 295]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 73] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 295]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 20] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 8]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 29] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 8]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 38] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 8]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 47] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 8]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 56] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 8]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 65] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 8]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 74] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 8]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 20] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 296]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 29] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 296]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 38] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 296]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 47] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 296]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 56] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 296]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 65] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 296]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 74] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 296]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 99] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 15]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 108] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 15]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 117] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 15]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 126] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 15]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 135] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 15]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 144] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 15]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 153] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 15]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 99] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 303]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 108] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 303]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 117] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 303]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 126] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 303]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 135] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 303]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 144] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 303]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 153] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 303]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 100] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 16]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 109] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 16]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 118] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 16]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 127] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 16]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 136] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 16]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 145] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 16]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 154] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 16]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 100] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 304]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 109] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 304]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 118] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 304]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 127] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 304]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 136] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 304]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 145] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 304]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 154] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 304]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 101] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 17]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 110] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 17]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 119] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 17]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 128] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 17]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 137] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 17]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 146] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 17]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 155] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 17]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 101] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 305]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 110] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 305]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 119] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 305]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 128] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 305]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 137] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 305]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 146] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 305]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 155] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 305]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 180] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 24]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 189] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 24]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 198] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 24]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 207] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 24]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 216] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 24]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 225] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 24]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 234] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 24]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 180] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 312]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 189] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 312]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 198] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 312]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 207] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 312]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 216] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 312]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 225] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 312]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 234] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 312]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 181] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 25]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 190] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 25]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 199] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 25]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 208] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 25]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 217] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 25]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 226] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 25]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 235] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 25]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 181] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 313]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 190] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 313]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 199] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 313]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 208] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 313]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 217] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 313]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 226] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 313]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 235] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 313]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 182] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 26]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 191] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 26]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 200] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 26]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 209] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 26]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 218] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 26]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 227] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 26]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 236] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 26]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 182] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 314]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 191] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 314]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 200] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 314]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 209] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 314]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 218] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 314]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 227] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 314]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 236] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 314]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 261] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 33]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 270] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 33]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 279] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 33]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 288] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 33]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 297] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 33]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 306] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 33]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 315] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 33]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 261] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 321]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 270] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 321]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 279] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 321]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 288] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 321]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 297] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 321]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 306] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 321]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 315] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 321]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 262] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 34]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 271] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 34]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 280] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 34]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 289] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 34]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 298] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 34]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 307] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 34]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 316] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 34]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 262] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 322]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 271] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 322]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 280] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 322]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 289] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 322]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 298] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 322]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 307] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 322]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 316] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 322]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 263] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 35]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 272] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 35]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 281] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 35]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 290] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 35]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 299] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 35]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 308] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 35]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 317] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 35]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 263] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 323]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 272] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 323]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 281] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 323]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 290] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 323]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 299] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 323]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 308] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 323]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 317] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 323]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 342] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 42]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 351] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 42]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 360] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 42]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 369] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 42]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 378] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 42]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 387] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 42]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 396] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 42]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 342] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 330]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 351] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 330]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 360] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 330]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 369] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 330]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 378] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 330]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 387] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 330]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 396] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 330]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 343] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 43]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 352] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 43]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 361] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 43]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 370] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 43]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 379] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 43]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 388] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 43]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 397] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 43]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 343] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 331]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 352] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 331]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 361] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 331]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 370] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 331]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 379] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 331]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 388] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 331]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 397] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 331]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 344] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 44]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 353] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 44]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 362] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 44]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 371] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 44]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 380] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 44]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 389] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 44]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 398] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 44]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 344] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 332]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 353] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 332]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 362] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 332]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 371] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 332]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 380] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 332]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 389] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 332]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 398] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 332]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 423] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 51]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 432] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 51]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 441] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 51]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 450] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 51]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 459] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 51]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 468] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 51]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 477] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 51]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 423] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 339]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 432] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 339]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 441] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 339]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 450] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 339]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 459] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 339]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 468] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 339]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 477] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 339]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 424] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 52]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 433] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 52]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 442] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 52]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 451] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 52]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 460] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 52]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 469] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 52]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 478] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 52]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 424] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 340]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 433] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 340]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 442] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 340]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 451] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 340]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 460] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 340]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 469] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 340]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 478] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 340]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 425] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 53]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 434] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 53]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 443] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 53]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 452] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 53]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 461] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 53]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 470] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 53]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 479] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 53]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 425] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 341]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 434] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 341]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 443] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 341]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 452] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 341]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 461] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 341]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 470] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 341]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 479] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 341]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 504] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 60]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 513] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 60]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 522] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 60]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 531] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 60]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 540] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 60]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 549] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 60]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 558] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 60]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 504] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 348]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 513] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 348]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 522] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 348]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 531] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 348]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 540] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 348]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 549] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 348]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 558] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 348]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 505] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 61]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 514] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 61]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 523] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 61]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 532] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 61]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 541] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 61]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 550] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 61]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 559] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 61]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 505] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 349]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 514] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 349]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 523] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 349]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 532] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 349]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 541] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 349]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 550] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 349]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 559] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 349]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 506] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 62]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 515] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 62]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 524] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 62]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 533] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 62]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 542] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 62]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 551] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 62]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 560] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 62]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 506] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 350]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 515] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 350]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 524] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 350]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 533] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 350]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 542] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 350]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 551] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 350]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 560] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 350]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 585] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 69]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 594] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 69]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 603] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 69]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 612] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 69]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 621] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 69]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 630] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 69]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 639] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 69]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 585] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 357]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 594] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 357]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 603] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 357]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 612] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 357]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 621] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 357]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 630] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 357]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 639] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 357]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 586] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 70]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 595] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 70]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 604] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 70]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 613] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 70]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 622] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 70]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 631] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 70]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 640] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 70]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 586] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 358]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 595] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 358]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 604] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 358]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 613] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 358]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 622] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 358]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 631] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 358]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 640] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 358]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 587] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 71]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 596] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 71]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 605] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 71]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 614] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 71]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 623] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 71]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 632] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 71]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 641] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 71]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 587] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 359]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 596] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 359]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 605] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 359]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 614] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 359]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 623] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 359]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 632] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 359]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 648 + threadIdx_x % 7 + 641] * kernel_shared_1[threadIdx_x // 7 * 576 + rc_outer_inner * 72 + 359]
for i1_inner in range(2):
compute_1 = T.Buffer((25088,), data=compute.data)
bias_1 = T.Buffer((512,), data=bias.data)
- compute_1[blockIdx_x * 1568 + threadIdx_x // 49 * 98 + i1_inner * 49 + threadIdx_x % 49] = T.max(conv2d_nchw_1[i1_inner] + bias_1[blockIdx_x * 32 + threadIdx_x // 49 * 2 + i1_inner], T.float32(0))
+ compute_1[blockIdx_x * 784 + threadIdx_x // 7 * 98 + i1_inner * 49 + threadIdx_x % 7] = T.max(conv2d_nchw_1[i1_inner] + bias_1[blockIdx_x * 16 + threadIdx_x // 7 * 2 + i1_inner], T.float32(0))
+ compute_1[blockIdx_x * 784 + threadIdx_x // 7 * 98 + i1_inner * 49 + threadIdx_x % 7 + 7] = T.max(conv2d_nchw_1[i1_inner + 2] + bias_1[blockIdx_x * 16 + threadIdx_x // 7 * 2 + i1_inner], T.float32(0))
+ compute_1[blockIdx_x * 784 + threadIdx_x // 7 * 98 + i1_inner * 49 + threadIdx_x % 7 + 14] = T.max(conv2d_nchw_1[i1_inner + 4] + bias_1[blockIdx_x * 16 + threadIdx_x // 7 * 2 + i1_inner], T.float32(0))
+ compute_1[blockIdx_x * 784 + threadIdx_x // 7 * 98 + i1_inner * 49 + threadIdx_x % 7 + 21] = T.max(conv2d_nchw_1[i1_inner + 6] + bias_1[blockIdx_x * 16 + threadIdx_x // 7 * 2 + i1_inner], T.float32(0))
+ compute_1[blockIdx_x * 784 + threadIdx_x // 7 * 98 + i1_inner * 49 + threadIdx_x % 7 + 28] = T.max(conv2d_nchw_1[i1_inner + 8] + bias_1[blockIdx_x * 16 + threadIdx_x // 7 * 2 + i1_inner], T.float32(0))
+ compute_1[blockIdx_x * 784 + threadIdx_x // 7 * 98 + i1_inner * 49 + threadIdx_x % 7 + 35] = T.max(conv2d_nchw_1[i1_inner + 10] + bias_1[blockIdx_x * 16 + threadIdx_x // 7 * 2 + i1_inner], T.float32(0))
+ compute_1[blockIdx_x * 784 + threadIdx_x // 7 * 98 + i1_inner * 49 + threadIdx_x % 7 + 42] = T.max(conv2d_nchw_1[i1_inner + 12] + bias_1[blockIdx_x * 16 + threadIdx_x // 7 * 2 + i1_inner], T.float32(0))
</pre></div>
</div>
</div>
@@ -569,7 +1853,7 @@ class Module:
<span class="p">)</span>
</pre></div>
</div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Execution time of this operator: 0.405 ms
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Execution time of this operator: 0.234 ms
</pre></div>
</div>
</div>
@@ -600,32 +1884,32 @@ conv2d_nchw_nn_o_o_o_i, conv2d_nchw_nn_o_o_i = s[conv2d_nchw].split(conv2d_nchw_
conv2d_nchw_nn_o_o_o_o, conv2d_nchw_nn_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_nn_o_o_o_i, factor=1)
conv2d_nchw_ff_o_i, conv2d_nchw_ff_i = s[conv2d_nchw].split(conv2d_nchw_ff, factor=2)
conv2d_nchw_ff_o_o_i, conv2d_nchw_ff_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_i, factor=1)
-conv2d_nchw_ff_o_o_o_i, conv2d_nchw_ff_o_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_o_i, factor=16)
+conv2d_nchw_ff_o_o_o_i, conv2d_nchw_ff_o_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_o_i, factor=8)
conv2d_nchw_ff_o_o_o_o, conv2d_nchw_ff_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_o_o_i, factor=1)
conv2d_nchw_yy_o_i, conv2d_nchw_yy_i = s[conv2d_nchw].split(conv2d_nchw_yy, factor=1)
conv2d_nchw_yy_o_o_i, conv2d_nchw_yy_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_i, factor=1)
-conv2d_nchw_yy_o_o_o_i, conv2d_nchw_yy_o_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_o_i, factor=7)
-conv2d_nchw_yy_o_o_o_o, conv2d_nchw_yy_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_o_o_i, factor=1)
+conv2d_nchw_yy_o_o_o_i, conv2d_nchw_yy_o_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_o_i, factor=1)
+conv2d_nchw_yy_o_o_o_o, conv2d_nchw_yy_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_o_o_i, factor=7)
conv2d_nchw_xx_o_i, conv2d_nchw_xx_i = s[conv2d_nchw].split(conv2d_nchw_xx, factor=1)
conv2d_nchw_xx_o_o_i, conv2d_nchw_xx_o_i = s[conv2d_nchw].split(conv2d_nchw_xx_o_i, factor=1)
conv2d_nchw_xx_o_o_o_i, conv2d_nchw_xx_o_o_i = s[conv2d_nchw].split(conv2d_nchw_xx_o_o_i, factor=7)
conv2d_nchw_xx_o_o_o_o, conv2d_nchw_xx_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_xx_o_o_o_i, factor=1)
-conv2d_nchw_rc_o_i, conv2d_nchw_rc_i = s[conv2d_nchw].split(conv2d_nchw_rc, factor=1)
-conv2d_nchw_rc_o_o, conv2d_nchw_rc_o_i = s[conv2d_nchw].split(conv2d_nchw_rc_o_i, factor=16)
+conv2d_nchw_rc_o_i, conv2d_nchw_rc_i = s[conv2d_nchw].split(conv2d_nchw_rc, factor=8)
+conv2d_nchw_rc_o_o, conv2d_nchw_rc_o_i = s[conv2d_nchw].split(conv2d_nchw_rc_o_i, factor=4)
conv2d_nchw_ry_o_i, conv2d_nchw_ry_i = s[conv2d_nchw].split(conv2d_nchw_ry, factor=1)
-conv2d_nchw_ry_o_o, conv2d_nchw_ry_o_i = s[conv2d_nchw].split(conv2d_nchw_ry_o_i, factor=1)
-conv2d_nchw_rx_o_i, conv2d_nchw_rx_i = s[conv2d_nchw].split(conv2d_nchw_rx, factor=1)
+conv2d_nchw_ry_o_o, conv2d_nchw_ry_o_i = s[conv2d_nchw].split(conv2d_nchw_ry_o_i, factor=3)
+conv2d_nchw_rx_o_i, conv2d_nchw_rx_i = s[conv2d_nchw].split(conv2d_nchw_rx, factor=3)
conv2d_nchw_rx_o_o, conv2d_nchw_rx_o_i = s[conv2d_nchw].split(conv2d_nchw_rx_o_i, factor=1)
s[conv2d_nchw].reorder(conv2d_nchw_nn_o_o_o_o, conv2d_nchw_ff_o_o_o_o, conv2d_nchw_yy_o_o_o_o, conv2d_nchw_xx_o_o_o_o, conv2d_nchw_nn_o_o_o_i, conv2d_nchw_ff_o_o_o_i, conv2d_nchw_yy_o_o_o_i, conv2d_nchw_xx_o_o_o_i, conv2d_nchw_nn_o_o_i, conv2d_nchw_ff_o_o_i, conv2d_nchw_yy_o_o_i, conv2d_nchw_xx_o_o_i, conv2d_nchw_rc_o_o, conv2d_nchw_ry_o_o, conv2d_nchw_rx_o_o, conv2d_nchw_rc_o_i, conv2d_nchw_ry_o_i, conv2d_nchw_rx_o_i, conv2d_nchw_nn_o_i, conv2d_nchw_ff_o_i, conv2d_nchw_yy_o_i, conv2d_nc [...]
compute_i0_o_i, compute_i0_i = s[compute].split(compute_i0, factor=1)
compute_i0_o_o_i, compute_i0_o_i = s[compute].split(compute_i0_o_i, factor=1)
compute_i0_o_o_o, compute_i0_o_o_i = s[compute].split(compute_i0_o_o_i, factor=1)
compute_i1_o_i, compute_i1_i = s[compute].split(compute_i1, factor=2)
-compute_i1_o_o_i, compute_i1_o_i = s[compute].split(compute_i1_o_i, factor=16)
+compute_i1_o_o_i, compute_i1_o_i = s[compute].split(compute_i1_o_i, factor=8)
compute_i1_o_o_o, compute_i1_o_o_i = s[compute].split(compute_i1_o_o_i, factor=1)
compute_i2_o_i, compute_i2_i = s[compute].split(compute_i2, factor=1)
-compute_i2_o_o_i, compute_i2_o_i = s[compute].split(compute_i2_o_i, factor=7)
-compute_i2_o_o_o, compute_i2_o_o_i = s[compute].split(compute_i2_o_o_i, factor=1)
+compute_i2_o_o_i, compute_i2_o_i = s[compute].split(compute_i2_o_i, factor=1)
+compute_i2_o_o_o, compute_i2_o_o_i = s[compute].split(compute_i2_o_o_i, factor=7)
compute_i3_o_i, compute_i3_i = s[compute].split(compute_i3, factor=1)
compute_i3_o_o_i, compute_i3_o_i = s[compute].split(compute_i3_o_i, factor=7)
compute_i3_o_o_o, compute_i3_o_o_i = s[compute].split(compute_i3_o_o_i, factor=1)
@@ -647,14 +1931,14 @@ s[compute].bind(compute_i0_o_i_i1_o_i_fused_i2_o_i_fused_i3_o_i_fused, te.thread
kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused = s[kernel_shared].fuse(kernel_shared_ax0, kernel_shared_ax1, kernel_shared_ax2, kernel_shared_ax3)
kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_i = s[kernel_shared].split(kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused, factor=1)
s[kernel_shared].vectorize(kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_i)
-kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_o, kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i = s[kernel_shared].split(kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, factor=784)
+kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_o, kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i = s[kernel_shared].split(kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, factor=56)
s[kernel_shared].bind(kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i, te.thread_axis("threadIdx.x"))
pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused = s[pad_temp_shared].fuse(pad_temp_shared_ax0, pad_temp_shared_ax1, pad_temp_shared_ax2, pad_temp_shared_ax3)
-pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_i = s[pad_temp_shared].split(pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused, factor=4)
+pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_i = s[pad_temp_shared].split(pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused, factor=1)
s[pad_temp_shared].vectorize(pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_i)
-pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_o, pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i = s[pad_temp_shared].split(pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, factor=784)
+pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_o, pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i = s[pad_temp_shared].split(pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, factor=56)
s[pad_temp_shared].bind(pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i, te.thread_axis("threadIdx.x"))
-s[conv2d_nchw].pragma(conv2d_nchw_nn_o_o_o_o, "auto_unroll_max_step", 0)
+s[conv2d_nchw].pragma(conv2d_nchw_nn_o_o_o_o, "auto_unroll_max_step", 1024)
s[conv2d_nchw].pragma(conv2d_nchw_nn_o_o_o_o, "unroll_explicit", True)
CUDA source code:
@@ -672,36 +1956,1180 @@ CUDA source code:
#define int64_t long long
#define uint64_t unsigned long long
#endif
-extern "C" __global__ void __launch_bounds__(784) default_function_kernel0(float* __restrict__ data, float* __restrict__ kernel, float* __restrict__ compute, float* __restrict__ bias) {
- float conv2d_nchw[2];
- __shared__ float pad_temp_shared[784];
- __shared__ float kernel_shared[512];
- for (int ff_inner_init = 0; ff_inner_init < 2; ++ff_inner_init) {
- conv2d_nchw[ff_inner_init] = 0.000000e+00f;
- }
- for (int rc_outer_outer = 0; rc_outer_outer < 32; ++rc_outer_outer) {
- for (int ry_outer_outer = 0; ry_outer_outer < 3; ++ry_outer_outer) {
- for (int rx_outer_outer = 0; rx_outer_outer < 3; ++rx_outer_outer) {
- __syncthreads();
- for (int ax0_ax1_fused_ax2_fused_ax3_fused_inner_s = 0; ax0_ax1_fused_ax2_fused_ax3_fused_inner_s < 4; ++ax0_ax1_fused_ax2_fused_ax3_fused_inner_s) {
- if (((int)threadIdx.x) < 196) {
- pad_temp_shared[((((int)threadIdx.x) * 4) + ax0_ax1_fused_ax2_fused_ax3_fused_inner_s)] = (((((1 <= (((((((int)threadIdx.x) * 4) + ax0_ax1_fused_ax2_fused_ax3_fused_inner_s) % 49) / 7) + ry_outer_outer)) && ((((((((int)threadIdx.x) * 4) + ax0_ax1_fused_ax2_fused_ax3_fused_inner_s) % 49) / 7) + ry_outer_outer) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 4) + ax0_ax1_fused_ax2_fused_ax3_fused_inner_s) % 7)))) && ((rx_outer_outer + ( [...]
- }
- }
- if (((int)threadIdx.x) < 512) {
- kernel_shared[((int)threadIdx.x)] = kernel[((((((((int)blockIdx.x) * 147456) + ((((int)threadIdx.x) >> 4) * 4608)) + (rc_outer_outer * 144)) + ((((int)threadIdx.x) & 15) * 9)) + (ry_outer_outer * 3)) + rx_outer_outer)];
- }
- __syncthreads();
- for (int rc_outer_inner = 0; rc_outer_inner < 16; ++rc_outer_inner) {
- for (int ff_inner = 0; ff_inner < 2; ++ff_inner) {
- conv2d_nchw[ff_inner] = (conv2d_nchw[ff_inner] + (pad_temp_shared[((rc_outer_inner * 49) + (((int)threadIdx.x) % 49))] * kernel_shared[((((((int)threadIdx.x) / 49) * 32) + (ff_inner * 16)) + rc_outer_inner)]));
- }
- }
- }
+extern "C" __global__ void __launch_bounds__(56) default_function_kernel0(float* __restrict__ data, float* __restrict__ kernel, float* __restrict__ compute, float* __restrict__ bias) {
+ float conv2d_nchw[14];
+ __shared__ float pad_temp_shared[2592];
+ __shared__ float kernel_shared[4608];
+ conv2d_nchw[0] = 0.000000e+00f;
+ conv2d_nchw[2] = 0.000000e+00f;
+ conv2d_nchw[4] = 0.000000e+00f;
+ conv2d_nchw[6] = 0.000000e+00f;
+ conv2d_nchw[8] = 0.000000e+00f;
+ conv2d_nchw[10] = 0.000000e+00f;
+ conv2d_nchw[12] = 0.000000e+00f;
+ conv2d_nchw[1] = 0.000000e+00f;
+ conv2d_nchw[3] = 0.000000e+00f;
+ conv2d_nchw[5] = 0.000000e+00f;
+ conv2d_nchw[7] = 0.000000e+00f;
+ conv2d_nchw[9] = 0.000000e+00f;
+ conv2d_nchw[11] = 0.000000e+00f;
+ conv2d_nchw[13] = 0.000000e+00f;
+ for (int rc_outer_outer = 0; rc_outer_outer < 16; ++rc_outer_outer) {
+ __syncthreads();
+ pad_temp_shared[((int)threadIdx.x)] = ((((9 <= ((int)threadIdx.x)) && (1 <= (((int)threadIdx.x) % 9))) && ((((int)threadIdx.x) % 9) < 8)) ? data[((((rc_outer_outer * 1568) + ((((int)threadIdx.x) / 9) * 7)) + (((int)threadIdx.x) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 56)] = (((((9 <= ((((int)threadIdx.x) + 56) % 81)) && (((((int)threadIdx.x) + 56) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 2) % 9))) && (((((int)threadIdx.x) + 2) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 56) / 81) * 49)) + ((((((int)threadIdx.x) + 56) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 2) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 112)] = (((((9 <= ((((int)threadIdx.x) + 31) % 81)) && (((((int)threadIdx.x) + 31) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 4) % 9))) && (((((int)threadIdx.x) + 4) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 112) / 81) * 49)) + ((((((int)threadIdx.x) + 31) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 4) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 168)] = ((((3 <= ((int)threadIdx.x)) && (1 <= ((((int)threadIdx.x) + 6) % 9))) && (((((int)threadIdx.x) + 6) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 168) / 81) * 49)) + (((((int)threadIdx.x) + 6) / 9) * 7)) + ((((int)threadIdx.x) + 6) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 224)] = (((((9 <= ((((int)threadIdx.x) + 62) % 81)) && (((((int)threadIdx.x) + 62) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 8) % 9))) && (((((int)threadIdx.x) + 8) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 224) / 81) * 49)) + ((((((int)threadIdx.x) + 62) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 8) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 280)] = (((((9 <= ((((int)threadIdx.x) + 37) % 81)) && (((((int)threadIdx.x) + 37) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 1) % 9))) && (((((int)threadIdx.x) + 1) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 280) / 81) * 49)) + ((((((int)threadIdx.x) + 37) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 1) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 336)] = (((1 <= ((((int)threadIdx.x) + 3) % 9)) && (((((int)threadIdx.x) + 3) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 336) / 81) * 49)) + (((((int)threadIdx.x) + 12) / 9) * 7)) + ((((int)threadIdx.x) + 3) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 392)] = (((((9 <= ((((int)threadIdx.x) + 68) % 81)) && (((((int)threadIdx.x) + 68) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 5) % 9))) && (((((int)threadIdx.x) + 5) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 392) / 81) * 49)) + ((((((int)threadIdx.x) + 68) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 5) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 448)] = (((((9 <= ((((int)threadIdx.x) + 43) % 81)) && (((((int)threadIdx.x) + 43) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 7) % 9))) && (((((int)threadIdx.x) + 7) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 448) / 81) * 49)) + ((((((int)threadIdx.x) + 43) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 7) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 504)] = ((((((int)threadIdx.x) < 54) && (1 <= (((int)threadIdx.x) % 9))) && ((((int)threadIdx.x) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 504) / 81) * 49)) + ((((int)threadIdx.x) / 9) * 7)) + (((int)threadIdx.x) % 9)) + 6)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 560)] = (((((9 <= ((((int)threadIdx.x) + 74) % 81)) && (((((int)threadIdx.x) + 74) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 2) % 9))) && (((((int)threadIdx.x) + 2) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 560) / 81) * 49)) + ((((((int)threadIdx.x) + 74) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 2) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 616)] = (((((9 <= ((((int)threadIdx.x) + 49) % 81)) && (((((int)threadIdx.x) + 49) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 4) % 9))) && (((((int)threadIdx.x) + 4) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 616) / 81) * 49)) + ((((((int)threadIdx.x) + 49) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 4) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 672)] = ((((((int)threadIdx.x) < 48) && (1 <= ((((int)threadIdx.x) + 6) % 9))) && (((((int)threadIdx.x) + 6) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 672) / 81) * 49)) + (((((int)threadIdx.x) + 24) / 9) * 7)) + ((((int)threadIdx.x) + 6) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 728)] = (((((9 <= ((((int)threadIdx.x) + 80) % 81)) && (((((int)threadIdx.x) + 80) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 8) % 9))) && (((((int)threadIdx.x) + 8) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 728) / 81) * 49)) + ((((((int)threadIdx.x) + 80) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 8) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 784)] = (((((9 <= ((((int)threadIdx.x) + 55) % 81)) && (((((int)threadIdx.x) + 55) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 1) % 9))) && (((((int)threadIdx.x) + 1) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 784) / 81) * 49)) + ((((((int)threadIdx.x) + 55) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 1) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 840)] = (((((9 <= ((((int)threadIdx.x) + 30) % 81)) && (((((int)threadIdx.x) + 30) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 3) % 9))) && (((((int)threadIdx.x) + 3) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 840) / 81) * 49)) + ((((((int)threadIdx.x) + 30) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 3) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 896)] = ((((4 <= ((int)threadIdx.x)) && (1 <= ((((int)threadIdx.x) + 5) % 9))) && (((((int)threadIdx.x) + 5) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 896) / 81) * 49)) + (((((int)threadIdx.x) + 5) / 9) * 7)) + ((((int)threadIdx.x) + 5) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 952)] = (((((9 <= ((((int)threadIdx.x) + 61) % 81)) && (((((int)threadIdx.x) + 61) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 7) % 9))) && (((((int)threadIdx.x) + 7) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 952) / 81) * 49)) + ((((((int)threadIdx.x) + 61) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 7) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 1008)] = (((((1 <= (((((int)threadIdx.x) / 9) + 4) % 9)) && (((((int)threadIdx.x) + 36) % 81) < 72)) && (1 <= (((int)threadIdx.x) % 9))) && ((((int)threadIdx.x) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 1008) / 81) * 49)) + ((((((int)threadIdx.x) / 9) + 4) % 9) * 7)) + (((int)threadIdx.x) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 1064)] = (((1 <= ((((int)threadIdx.x) + 2) % 9)) && (((((int)threadIdx.x) + 2) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 1064) / 81) * 49)) + (((((int)threadIdx.x) + 11) / 9) * 7)) + ((((int)threadIdx.x) + 2) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 1120)] = (((((9 <= ((((int)threadIdx.x) + 67) % 81)) && (((((int)threadIdx.x) + 67) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 4) % 9))) && (((((int)threadIdx.x) + 4) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 1120) / 81) * 49)) + ((((((int)threadIdx.x) + 67) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 4) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 1176)] = (((((9 <= ((((int)threadIdx.x) + 42) % 81)) && (((((int)threadIdx.x) + 42) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 6) % 9))) && (((((int)threadIdx.x) + 6) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 1176) / 81) * 49)) + ((((((int)threadIdx.x) + 42) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 6) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 1232)] = ((((((int)threadIdx.x) < 55) && (1 <= ((((int)threadIdx.x) + 8) % 9))) && (((((int)threadIdx.x) + 8) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 1232) / 81) * 49)) + (((((int)threadIdx.x) + 17) / 9) * 7)) + ((((int)threadIdx.x) + 8) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 1288)] = (((((9 <= ((((int)threadIdx.x) + 73) % 81)) && (((((int)threadIdx.x) + 73) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 1) % 9))) && (((((int)threadIdx.x) + 1) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 1288) / 81) * 49)) + ((((((int)threadIdx.x) + 73) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 1) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 1344)] = (((((9 <= ((((int)threadIdx.x) + 48) % 81)) && (((((int)threadIdx.x) + 48) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 3) % 9))) && (((((int)threadIdx.x) + 3) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 1344) / 81) * 49)) + ((((((int)threadIdx.x) + 48) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 3) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 1400)] = ((((((int)threadIdx.x) < 49) && (1 <= ((((int)threadIdx.x) + 5) % 9))) && (((((int)threadIdx.x) + 5) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 1400) / 81) * 49)) + (((((int)threadIdx.x) + 23) / 9) * 7)) + ((((int)threadIdx.x) + 5) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 1456)] = (((((9 <= ((((int)threadIdx.x) + 79) % 81)) && (((((int)threadIdx.x) + 79) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 7) % 9))) && (((((int)threadIdx.x) + 7) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 1456) / 81) * 49)) + ((((((int)threadIdx.x) + 79) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 7) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 1512)] = (((((1 <= (((((int)threadIdx.x) / 9) + 6) % 9)) && (((((int)threadIdx.x) + 54) % 81) < 72)) && (1 <= (((int)threadIdx.x) % 9))) && ((((int)threadIdx.x) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 1512) / 81) * 49)) + ((((((int)threadIdx.x) / 9) + 6) % 9) * 7)) + (((int)threadIdx.x) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 1568)] = (((((9 <= ((((int)threadIdx.x) + 29) % 81)) && (((((int)threadIdx.x) + 29) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 2) % 9))) && (((((int)threadIdx.x) + 2) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 1568) / 81) * 49)) + ((((((int)threadIdx.x) + 29) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 2) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 1624)] = ((((5 <= ((int)threadIdx.x)) && (1 <= ((((int)threadIdx.x) + 4) % 9))) && (((((int)threadIdx.x) + 4) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 1624) / 81) * 49)) + (((((int)threadIdx.x) + 4) / 9) * 7)) + ((((int)threadIdx.x) + 4) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 1680)] = (((((9 <= ((((int)threadIdx.x) + 60) % 81)) && (((((int)threadIdx.x) + 60) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 6) % 9))) && (((((int)threadIdx.x) + 6) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 1680) / 81) * 49)) + ((((((int)threadIdx.x) + 60) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 6) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 1736)] = (((((9 <= ((((int)threadIdx.x) + 35) % 81)) && (((((int)threadIdx.x) + 35) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 8) % 9))) && (((((int)threadIdx.x) + 8) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 1736) / 81) * 49)) + ((((((int)threadIdx.x) + 35) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 8) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 1792)] = (((1 <= ((((int)threadIdx.x) + 1) % 9)) && (((((int)threadIdx.x) + 1) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 1792) / 81) * 49)) + (((((int)threadIdx.x) + 10) / 9) * 7)) + ((((int)threadIdx.x) + 1) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 1848)] = (((((9 <= ((((int)threadIdx.x) + 66) % 81)) && (((((int)threadIdx.x) + 66) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 3) % 9))) && (((((int)threadIdx.x) + 3) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 1848) / 81) * 49)) + ((((((int)threadIdx.x) + 66) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 3) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 1904)] = (((((9 <= ((((int)threadIdx.x) + 41) % 81)) && (((((int)threadIdx.x) + 41) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 5) % 9))) && (((((int)threadIdx.x) + 5) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 1904) / 81) * 49)) + ((((((int)threadIdx.x) + 41) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 5) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 1960)] = (((1 <= ((((int)threadIdx.x) + 7) % 9)) && (((((int)threadIdx.x) + 7) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 1960) / 81) * 49)) + (((((int)threadIdx.x) + 16) / 9) * 7)) + ((((int)threadIdx.x) + 7) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 2016)] = (((((1 <= (((((int)threadIdx.x) / 9) + 8) % 9)) && (((((int)threadIdx.x) + 72) % 81) < 72)) && (1 <= (((int)threadIdx.x) % 9))) && ((((int)threadIdx.x) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 2016) / 81) * 49)) + ((((((int)threadIdx.x) / 9) + 8) % 9) * 7)) + (((int)threadIdx.x) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 2072)] = (((((9 <= ((((int)threadIdx.x) + 47) % 81)) && (((((int)threadIdx.x) + 47) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 2) % 9))) && (((((int)threadIdx.x) + 2) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 2072) / 81) * 49)) + ((((((int)threadIdx.x) + 47) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 2) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 2128)] = ((((((int)threadIdx.x) < 50) && (1 <= ((((int)threadIdx.x) + 4) % 9))) && (((((int)threadIdx.x) + 4) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 2128) / 81) * 49)) + (((((int)threadIdx.x) + 22) / 9) * 7)) + ((((int)threadIdx.x) + 4) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 2184)] = (((((9 <= ((((int)threadIdx.x) + 78) % 81)) && (((((int)threadIdx.x) + 78) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 6) % 9))) && (((((int)threadIdx.x) + 6) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 2184) / 81) * 49)) + ((((((int)threadIdx.x) + 78) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 6) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 2240)] = (((((9 <= ((((int)threadIdx.x) + 53) % 81)) && (((((int)threadIdx.x) + 53) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 8) % 9))) && (((((int)threadIdx.x) + 8) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 2240) / 81) * 49)) + ((((((int)threadIdx.x) + 53) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 8) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 2296)] = (((((9 <= ((((int)threadIdx.x) + 28) % 81)) && (((((int)threadIdx.x) + 28) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 1) % 9))) && (((((int)threadIdx.x) + 1) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 2296) / 81) * 49)) + ((((((int)threadIdx.x) + 28) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 1) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 2352)] = ((((6 <= ((int)threadIdx.x)) && (1 <= ((((int)threadIdx.x) + 3) % 9))) && (((((int)threadIdx.x) + 3) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 2352) / 81) * 49)) + (((((int)threadIdx.x) + 3) / 9) * 7)) + ((((int)threadIdx.x) + 3) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 2408)] = (((((9 <= ((((int)threadIdx.x) + 59) % 81)) && (((((int)threadIdx.x) + 59) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 5) % 9))) && (((((int)threadIdx.x) + 5) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 2408) / 81) * 49)) + ((((((int)threadIdx.x) + 59) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 5) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 2464)] = (((((9 <= ((((int)threadIdx.x) + 34) % 81)) && (((((int)threadIdx.x) + 34) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 7) % 9))) && (((((int)threadIdx.x) + 7) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 2464) / 81) * 49)) + ((((((int)threadIdx.x) + 34) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 7) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 2520)] = (((1 <= (((int)threadIdx.x) % 9)) && ((((int)threadIdx.x) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 2520) / 81) * 49)) + ((((int)threadIdx.x) / 9) * 7)) + (((int)threadIdx.x) % 9)) - 1)] : 0.000000e+00f);
+ if (((int)threadIdx.x) < 16) {
+ pad_temp_shared[(((int)threadIdx.x) + 2576)] = ((((((int)threadIdx.x) < 7) && (1 <= ((((int)threadIdx.x) + 2) % 9))) && (((((int)threadIdx.x) + 2) % 9) < 8)) ? data[(((((rc_outer_outer * 1568) + (((((int)threadIdx.x) + 2576) / 81) * 49)) + (((((int)threadIdx.x) + 65) / 9) * 7)) + ((int)threadIdx.x)) - 6)] : 0.000000e+00f);
+ }
+ kernel_shared[((int)threadIdx.x)] = kernel[(((((int)blockIdx.x) * 73728) + (rc_outer_outer * 288)) + ((int)threadIdx.x))];
+ kernel_shared[(((int)threadIdx.x) + 56)] = kernel[((((((int)blockIdx.x) * 73728) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 56)];
+ kernel_shared[(((int)threadIdx.x) + 112)] = kernel[((((((int)blockIdx.x) * 73728) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 112)];
+ kernel_shared[(((int)threadIdx.x) + 168)] = kernel[((((((int)blockIdx.x) * 73728) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 168)];
+ kernel_shared[(((int)threadIdx.x) + 224)] = kernel[((((((int)blockIdx.x) * 73728) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 224)];
+ kernel_shared[(((int)threadIdx.x) + 280)] = kernel[((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 280) / 288) * 4608)) + (rc_outer_outer * 288)) + ((((int)threadIdx.x) + 280) % 288))];
+ kernel_shared[(((int)threadIdx.x) + 336)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 336) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 48)];
+ kernel_shared[(((int)threadIdx.x) + 392)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 392) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 104)];
+ kernel_shared[(((int)threadIdx.x) + 448)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 448) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 160)];
+ kernel_shared[(((int)threadIdx.x) + 504)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 504) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 216)];
+ kernel_shared[(((int)threadIdx.x) + 560)] = kernel[((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 560) / 288) * 4608)) + (rc_outer_outer * 288)) + ((((int)threadIdx.x) + 272) % 288))];
+ kernel_shared[(((int)threadIdx.x) + 616)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 616) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 40)];
+ kernel_shared[(((int)threadIdx.x) + 672)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 672) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 96)];
+ kernel_shared[(((int)threadIdx.x) + 728)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 728) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 152)];
+ kernel_shared[(((int)threadIdx.x) + 784)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 784) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 208)];
+ kernel_shared[(((int)threadIdx.x) + 840)] = kernel[((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 840) / 288) * 4608)) + (rc_outer_outer * 288)) + ((((int)threadIdx.x) + 264) % 288))];
+ kernel_shared[(((int)threadIdx.x) + 896)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 896) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 32)];
+ kernel_shared[(((int)threadIdx.x) + 952)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 952) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 88)];
+ kernel_shared[(((int)threadIdx.x) + 1008)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1008) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 144)];
+ kernel_shared[(((int)threadIdx.x) + 1064)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1064) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 200)];
+ kernel_shared[(((int)threadIdx.x) + 1120)] = kernel[((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1120) / 288) * 4608)) + (rc_outer_outer * 288)) + ((((int)threadIdx.x) + 256) % 288))];
+ kernel_shared[(((int)threadIdx.x) + 1176)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1176) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 24)];
+ kernel_shared[(((int)threadIdx.x) + 1232)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1232) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 80)];
+ kernel_shared[(((int)threadIdx.x) + 1288)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1288) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 136)];
+ kernel_shared[(((int)threadIdx.x) + 1344)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1344) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 192)];
+ kernel_shared[(((int)threadIdx.x) + 1400)] = kernel[((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1400) / 288) * 4608)) + (rc_outer_outer * 288)) + ((((int)threadIdx.x) + 248) % 288))];
+ kernel_shared[(((int)threadIdx.x) + 1456)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1456) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 16)];
+ kernel_shared[(((int)threadIdx.x) + 1512)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1512) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 72)];
+ kernel_shared[(((int)threadIdx.x) + 1568)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1568) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 128)];
+ kernel_shared[(((int)threadIdx.x) + 1624)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1624) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 184)];
+ kernel_shared[(((int)threadIdx.x) + 1680)] = kernel[((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1680) / 288) * 4608)) + (rc_outer_outer * 288)) + ((((int)threadIdx.x) + 240) % 288))];
+ kernel_shared[(((int)threadIdx.x) + 1736)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1736) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 8)];
+ kernel_shared[(((int)threadIdx.x) + 1792)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1792) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 64)];
+ kernel_shared[(((int)threadIdx.x) + 1848)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1848) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 120)];
+ kernel_shared[(((int)threadIdx.x) + 1904)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1904) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 176)];
+ kernel_shared[(((int)threadIdx.x) + 1960)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1960) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 232)];
+ kernel_shared[(((int)threadIdx.x) + 2016)] = kernel[((((((int)blockIdx.x) * 73728) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 32256)];
+ kernel_shared[(((int)threadIdx.x) + 2072)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2072) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 56)];
+ kernel_shared[(((int)threadIdx.x) + 2128)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2128) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 112)];
+ kernel_shared[(((int)threadIdx.x) + 2184)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2184) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 168)];
+ kernel_shared[(((int)threadIdx.x) + 2240)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2240) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 224)];
+ kernel_shared[(((int)threadIdx.x) + 2296)] = kernel[((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2296) / 288) * 4608)) + (rc_outer_outer * 288)) + ((((int)threadIdx.x) + 280) % 288))];
+ kernel_shared[(((int)threadIdx.x) + 2352)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2352) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 48)];
+ kernel_shared[(((int)threadIdx.x) + 2408)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2408) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 104)];
+ kernel_shared[(((int)threadIdx.x) + 2464)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2464) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 160)];
+ kernel_shared[(((int)threadIdx.x) + 2520)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2520) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 216)];
+ kernel_shared[(((int)threadIdx.x) + 2576)] = kernel[((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2576) / 288) * 4608)) + (rc_outer_outer * 288)) + ((((int)threadIdx.x) + 272) % 288))];
+ kernel_shared[(((int)threadIdx.x) + 2632)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2632) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 40)];
+ kernel_shared[(((int)threadIdx.x) + 2688)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2688) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 96)];
+ kernel_shared[(((int)threadIdx.x) + 2744)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2744) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 152)];
+ kernel_shared[(((int)threadIdx.x) + 2800)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2800) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 208)];
+ kernel_shared[(((int)threadIdx.x) + 2856)] = kernel[((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2856) / 288) * 4608)) + (rc_outer_outer * 288)) + ((((int)threadIdx.x) + 264) % 288))];
+ kernel_shared[(((int)threadIdx.x) + 2912)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2912) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 32)];
+ kernel_shared[(((int)threadIdx.x) + 2968)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2968) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 88)];
+ kernel_shared[(((int)threadIdx.x) + 3024)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3024) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 144)];
+ kernel_shared[(((int)threadIdx.x) + 3080)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3080) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 200)];
+ kernel_shared[(((int)threadIdx.x) + 3136)] = kernel[((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3136) / 288) * 4608)) + (rc_outer_outer * 288)) + ((((int)threadIdx.x) + 256) % 288))];
+ kernel_shared[(((int)threadIdx.x) + 3192)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3192) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 24)];
+ kernel_shared[(((int)threadIdx.x) + 3248)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3248) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 80)];
+ kernel_shared[(((int)threadIdx.x) + 3304)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3304) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 136)];
+ kernel_shared[(((int)threadIdx.x) + 3360)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3360) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 192)];
+ kernel_shared[(((int)threadIdx.x) + 3416)] = kernel[((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3416) / 288) * 4608)) + (rc_outer_outer * 288)) + ((((int)threadIdx.x) + 248) % 288))];
+ kernel_shared[(((int)threadIdx.x) + 3472)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3472) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 16)];
+ kernel_shared[(((int)threadIdx.x) + 3528)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3528) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 72)];
+ kernel_shared[(((int)threadIdx.x) + 3584)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3584) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 128)];
+ kernel_shared[(((int)threadIdx.x) + 3640)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3640) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 184)];
+ kernel_shared[(((int)threadIdx.x) + 3696)] = kernel[((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3696) / 288) * 4608)) + (rc_outer_outer * 288)) + ((((int)threadIdx.x) + 240) % 288))];
+ kernel_shared[(((int)threadIdx.x) + 3752)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3752) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 8)];
+ kernel_shared[(((int)threadIdx.x) + 3808)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3808) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 64)];
+ kernel_shared[(((int)threadIdx.x) + 3864)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3864) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 120)];
+ kernel_shared[(((int)threadIdx.x) + 3920)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3920) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 176)];
+ kernel_shared[(((int)threadIdx.x) + 3976)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3976) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 232)];
+ kernel_shared[(((int)threadIdx.x) + 4032)] = kernel[((((((int)blockIdx.x) * 73728) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 64512)];
+ kernel_shared[(((int)threadIdx.x) + 4088)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 4088) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 56)];
+ kernel_shared[(((int)threadIdx.x) + 4144)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 4144) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 112)];
+ kernel_shared[(((int)threadIdx.x) + 4200)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 4200) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 168)];
+ kernel_shared[(((int)threadIdx.x) + 4256)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 4256) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 224)];
+ kernel_shared[(((int)threadIdx.x) + 4312)] = kernel[((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 4312) / 288) * 4608)) + (rc_outer_outer * 288)) + ((((int)threadIdx.x) + 280) % 288))];
+ kernel_shared[(((int)threadIdx.x) + 4368)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 4368) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 48)];
+ kernel_shared[(((int)threadIdx.x) + 4424)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 4424) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 104)];
+ kernel_shared[(((int)threadIdx.x) + 4480)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 4480) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 160)];
+ kernel_shared[(((int)threadIdx.x) + 4536)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 4536) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 216)];
+ if (((int)threadIdx.x) < 16) {
+ kernel_shared[(((int)threadIdx.x) + 4592)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 4592) / 288) * 4608)) + (rc_outer_outer * 288)) + ((int)threadIdx.x)) + 272)];
+ }
+ __syncthreads();
+ for (int rc_outer_inner = 0; rc_outer_inner < 4; ++rc_outer_inner) {
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((rc_outer_inner * 648) + (((int)threadIdx.x) % 7))] * kernel_shared[(((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72))]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[(((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72))]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[(((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72))]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[(((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72))]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[(((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72))]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[(((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72))]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[(((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72))]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((rc_outer_inner * 648) + (((int)threadIdx.x) % 7))] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 288)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 288)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 288)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 288)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 288)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 288)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 288)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 1)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 1)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 1)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 28)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 1)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 37)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 1)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 46)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 1)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 1)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 289)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 289)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 289)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 28)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 289)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 37)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 289)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 46)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 289)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 289)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 2)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 2)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 2)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 29)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 2)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 38)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 2)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 47)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 2)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 56)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 2)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 290)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 290)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 290)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 29)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 290)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 38)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 290)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 47)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 290)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 56)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 290)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 81)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 9)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 9)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 9)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 9)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 9)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 9)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 9)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 81)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 297)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 297)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 297)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 297)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 297)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 297)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 297)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 82)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 10)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 10)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 10)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 109)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 10)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 118)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 10)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 10)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 136)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 10)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 82)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 298)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 298)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 298)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 109)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 298)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 118)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 298)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 298)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 136)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 298)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 83)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 11)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 11)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 11)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 110)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 11)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 119)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 11)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 11)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 137)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 11)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 83)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 299)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 299)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 299)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 110)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 299)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 119)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 299)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 299)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 137)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 299)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 162)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 18)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 18)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 18)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 18)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 18)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 18)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 18)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 162)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 306)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 306)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 306)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 306)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 306)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 306)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 306)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 163)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 19)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 19)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 19)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 19)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 199)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 19)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 208)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 19)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 217)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 19)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 163)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 307)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 307)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 307)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 307)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 199)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 307)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 208)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 307)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 217)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 307)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 164)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 20)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 20)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 20)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 20)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 200)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 20)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 209)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 20)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 218)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 20)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 164)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 308)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 308)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 308)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 308)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 200)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 308)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 209)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 308)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 218)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 308)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 243)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 27)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 27)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 27)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 27)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 27)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 27)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 27)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 243)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 315)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 315)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 315)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 315)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 315)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 315)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 315)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 244)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 28)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 28)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 28)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 271)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 28)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 280)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 28)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 289)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 28)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 298)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 28)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 244)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 316)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 316)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 316)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 271)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 316)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 280)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 316)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 289)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 316)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 298)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 316)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 245)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 29)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 29)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 29)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 272)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 29)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 281)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 29)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 290)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 29)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 299)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 29)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 245)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 317)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 317)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 317)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 272)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 317)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 281)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 317)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 290)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 317)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 299)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 317)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 324)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 36)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 333)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 36)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 342)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 36)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 351)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 36)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 360)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 36)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 369)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 36)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 36)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 324)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 324)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 333)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 324)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 342)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 324)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 351)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 324)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 360)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 324)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 369)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 324)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 324)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 325)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 37)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 334)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 37)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 343)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 37)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 352)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 37)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 361)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 37)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 370)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 37)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 379)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 37)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 325)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 325)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 334)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 325)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 343)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 325)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 352)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 325)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 361)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 325)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 370)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 325)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 379)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 325)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 326)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 38)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 335)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 38)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 344)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 38)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 353)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 38)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 362)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 38)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 371)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 38)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 380)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 38)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 326)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 326)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 335)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 326)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 344)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 326)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 353)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 326)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 362)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 326)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 371)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 326)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 380)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 326)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 405)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 45)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 414)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 45)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 423)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 45)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 432)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 45)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 45)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 450)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 45)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 459)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 45)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 405)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 333)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 414)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 333)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 423)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 333)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 432)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 333)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 333)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 450)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 333)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 459)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 333)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 406)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 46)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 415)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 46)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 424)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 46)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 433)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 46)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 442)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 46)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 451)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 46)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 460)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 46)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 406)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 334)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 415)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 334)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 424)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 334)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 433)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 334)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 442)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 334)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 451)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 334)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 460)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 334)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 407)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 47)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 416)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 47)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 425)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 47)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 434)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 47)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 443)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 47)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 452)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 47)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 461)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 47)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 407)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 335)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 416)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 335)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 425)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 335)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 434)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 335)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 443)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 335)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 452)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 335)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 461)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 335)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 486)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 54)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 495)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 54)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 504)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 54)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 513)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 54)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 522)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 54)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 531)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 54)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 540)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 54)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 486)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 342)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 495)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 342)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 504)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 342)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 513)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 342)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 522)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 342)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 531)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 342)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 540)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 342)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 487)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 55)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 496)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 55)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 505)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 55)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 514)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 55)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 523)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 55)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 532)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 55)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 541)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 55)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 487)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 343)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 496)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 343)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 505)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 343)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 514)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 343)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 523)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 343)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 532)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 343)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 541)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 343)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 488)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 56)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 497)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 56)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 506)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 56)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 515)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 56)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 524)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 56)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 533)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 56)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 542)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 56)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 488)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 344)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 497)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 344)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 506)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 344)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 515)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 344)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 524)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 344)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 533)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 344)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 542)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 344)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 567)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 63)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 576)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 63)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 585)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 63)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 594)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 63)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 603)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 63)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 612)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 63)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 621)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 63)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 567)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 351)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 576)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 351)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 585)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 351)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 594)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 351)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 603)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 351)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 612)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 351)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 621)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 351)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 568)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 64)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 577)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 64)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 586)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 64)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 595)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 64)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 604)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 64)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 613)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 64)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 622)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 64)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 568)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 352)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 577)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 352)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 586)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 352)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 595)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 352)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 604)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 352)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 613)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 352)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 622)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 352)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 569)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 65)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 578)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 65)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 587)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 65)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 596)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 65)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 605)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 65)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 614)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 65)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 623)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 65)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 569)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 353)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 578)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 353)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 587)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 353)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 596)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 353)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 605)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 353)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 614)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 353)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 623)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 353)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 3)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 3)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 3)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 3)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 3)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 3)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 3)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 291)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 291)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 291)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 291)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 291)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 291)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 291)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 4)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 4)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 28)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 4)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 37)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 4)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 46)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 4)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 4)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 4)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 292)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 292)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 28)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 292)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 37)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 292)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 46)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 292)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 292)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 292)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 5)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 5)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 29)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 5)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 38)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 5)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 47)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 5)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 56)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 5)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 5)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 293)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 293)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 29)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 293)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 38)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 293)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 47)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 293)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 56)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 293)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 293)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 12)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 12)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 12)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 12)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 12)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 12)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 144)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 12)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 300)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 300)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 300)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 300)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 300)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 300)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 144)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 300)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 13)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 13)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 109)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 13)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 118)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 13)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 13)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 136)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 13)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 145)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 13)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 301)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 301)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 109)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 301)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 118)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 301)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 301)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 136)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 301)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 145)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 301)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 14)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 14)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 110)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 14)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 119)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 14)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 14)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 137)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 14)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 146)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 14)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 302)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 302)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 110)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 302)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 119)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 302)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 302)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 137)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 302)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 146)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 302)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 21)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 21)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 21)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 21)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 21)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 21)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 225)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 21)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 309)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 309)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 309)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 309)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 309)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 309)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 225)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 309)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 22)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 22)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 22)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 199)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 22)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 208)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 22)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 217)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 22)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 226)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 22)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 310)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 310)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 310)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 199)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 310)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 208)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 310)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 217)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 310)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 226)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 310)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 23)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 23)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 23)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 200)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 23)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 209)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 23)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 218)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 23)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 227)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 23)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 311)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 311)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 311)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 200)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 311)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 209)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 311)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 218)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 311)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 227)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 311)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 30)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 30)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 30)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 30)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 30)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 30)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 306)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 30)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 318)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 318)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 318)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 318)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 318)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 318)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 306)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 318)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 31)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 31)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 271)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 31)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 280)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 31)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 289)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 31)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 298)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 31)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 307)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 31)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 319)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 319)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 271)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 319)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 280)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 319)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 289)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 319)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 298)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 319)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 307)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 319)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 32)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 32)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 272)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 32)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 281)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 32)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 290)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 32)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 299)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 32)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 308)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 32)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 320)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 320)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 272)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 320)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 281)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 320)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 290)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 320)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 299)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 320)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 308)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 320)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 333)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 39)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 342)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 39)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 351)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 39)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 360)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 39)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 369)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 39)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 39)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 387)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 39)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 333)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 327)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 342)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 327)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 351)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 327)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 360)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 327)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 369)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 327)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 327)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 387)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 327)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 334)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 40)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 343)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 40)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 352)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 40)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 361)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 40)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 370)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 40)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 379)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 40)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 388)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 40)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 334)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 328)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 343)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 328)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 352)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 328)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 361)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 328)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 370)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 328)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 379)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 328)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 388)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 328)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 335)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 41)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 344)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 41)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 353)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 41)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 362)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 41)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 371)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 41)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 380)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 41)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 389)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 41)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 335)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 329)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 344)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 329)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 353)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 329)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 362)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 329)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 371)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 329)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 380)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 329)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 389)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 329)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 414)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 48)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 423)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 48)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 432)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 48)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 48)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 450)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 48)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 459)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 48)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 468)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 48)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 414)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 336)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 423)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 336)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 432)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 336)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 336)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 450)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 336)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 459)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 336)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 468)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 336)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 415)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 49)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 424)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 49)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 433)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 49)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 442)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 49)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 451)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 49)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 460)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 49)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 469)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 49)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 415)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 337)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 424)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 337)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 433)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 337)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 442)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 337)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 451)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 337)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 460)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 337)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 469)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 337)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 416)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 50)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 425)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 50)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 434)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 50)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 443)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 50)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 452)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 50)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 461)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 50)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 470)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 50)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 416)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 338)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 425)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 338)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 434)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 338)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 443)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 338)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 452)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 338)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 461)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 338)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 470)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 338)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 495)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 57)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 504)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 57)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 513)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 57)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 522)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 57)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 531)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 57)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 540)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 57)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 549)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 57)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 495)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 345)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 504)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 345)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 513)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 345)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 522)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 345)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 531)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 345)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 540)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 345)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 549)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 345)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 496)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 58)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 505)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 58)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 514)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 58)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 523)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 58)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 532)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 58)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 541)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 58)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 550)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 58)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 496)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 346)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 505)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 346)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 514)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 346)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 523)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 346)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 532)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 346)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 541)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 346)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 550)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 346)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 497)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 59)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 506)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 59)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 515)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 59)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 524)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 59)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 533)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 59)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 542)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 59)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 551)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 59)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 497)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 347)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 506)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 347)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 515)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 347)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 524)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 347)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 533)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 347)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 542)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 347)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 551)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 347)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 576)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 66)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 585)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 66)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 594)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 66)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 603)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 66)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 612)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 66)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 621)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 66)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 630)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 66)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 576)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 354)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 585)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 354)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 594)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 354)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 603)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 354)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 612)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 354)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 621)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 354)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 630)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 354)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 577)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 67)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 586)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 67)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 595)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 67)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 604)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 67)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 613)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 67)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 622)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 67)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 631)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 67)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 577)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 355)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 586)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 355)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 595)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 355)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 604)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 355)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 613)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 355)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 622)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 355)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 631)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 355)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 578)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 68)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 587)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 68)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 596)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 68)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 605)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 68)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 614)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 68)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 623)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 68)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 632)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 68)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 578)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 356)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 587)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 356)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 596)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 356)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 605)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 356)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 614)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 356)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 623)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 356)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 632)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 356)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 6)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 6)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 6)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 6)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 6)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 6)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 72)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 6)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 294)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 294)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 294)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 294)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 294)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 294)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 72)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 294)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 7)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 28)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 7)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 37)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 7)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 46)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 7)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 7)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 7)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 73)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 7)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 295)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 28)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 295)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 37)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 295)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 46)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 295)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 295)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 295)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 73)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 295)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 8)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 29)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 8)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 38)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 8)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 47)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 8)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 56)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 8)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 8)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 74)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 8)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 296)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 29)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 296)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 38)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 296)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 47)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 296)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 56)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 296)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 296)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 74)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 296)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 15)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 15)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 15)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 15)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 15)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 144)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 15)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 153)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 15)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 303)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 303)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 303)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 303)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 303)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 144)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 303)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 153)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 303)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 16)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 109)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 16)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 118)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 16)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 16)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 136)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 16)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 145)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 16)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 154)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 16)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 304)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 109)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 304)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 118)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 304)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 304)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 136)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 304)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 145)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 304)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 154)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 304)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 17)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 110)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 17)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 119)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 17)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 17)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 137)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 17)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 146)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 17)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 155)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 17)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 305)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 110)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 305)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 119)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 305)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 305)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 137)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 305)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 146)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 305)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 155)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 305)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 24)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 24)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 24)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 24)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 24)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 225)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 24)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 234)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 24)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 312)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 312)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 312)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 312)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 312)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 225)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 312)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 234)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 312)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 25)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 25)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 199)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 25)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 208)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 25)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 217)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 25)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 226)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 25)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 235)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 25)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 313)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 313)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 199)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 313)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 208)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 313)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 217)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 313)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 226)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 313)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 235)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 313)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 26)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 26)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 200)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 26)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 209)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 26)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 218)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 26)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 227)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 26)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 236)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 26)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 314)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 314)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 200)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 314)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 209)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 314)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 218)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 314)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 227)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 314)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 236)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 314)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 33)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 33)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 33)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 33)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 33)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 306)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 33)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 33)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 321)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 321)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 321)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 321)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 321)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 306)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 321)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 321)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 34)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 271)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 34)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 280)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 34)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 289)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 34)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 298)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 34)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 307)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 34)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 316)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 34)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 322)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 271)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 322)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 280)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 322)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 289)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 322)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 298)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 322)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 307)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 322)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 316)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 322)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 35)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 272)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 35)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 281)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 35)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 290)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 35)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 299)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 35)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 308)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 35)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 317)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 35)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 323)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 272)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 323)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 281)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 323)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 290)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 323)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 299)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 323)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 308)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 323)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 317)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 323)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 342)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 42)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 351)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 42)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 360)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 42)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 369)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 42)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 42)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 387)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 42)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 396)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 42)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 342)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 330)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 351)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 330)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 360)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 330)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 369)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 330)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 330)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 387)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 330)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 396)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 330)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 343)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 43)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 352)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 43)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 361)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 43)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 370)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 43)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 379)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 43)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 388)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 43)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 397)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 43)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 343)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 331)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 352)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 331)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 361)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 331)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 370)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 331)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 379)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 331)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 388)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 331)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 397)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 331)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 344)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 44)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 353)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 44)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 362)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 44)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 371)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 44)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 380)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 44)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 389)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 44)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 398)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 44)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 344)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 332)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 353)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 332)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 362)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 332)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 371)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 332)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 380)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 332)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 389)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 332)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 398)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 332)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 423)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 51)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 432)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 51)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 51)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 450)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 51)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 459)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 51)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 468)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 51)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 477)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 51)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 423)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 339)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 432)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 339)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 339)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 450)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 339)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 459)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 339)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 468)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 339)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 477)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 339)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 424)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 52)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 433)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 52)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 442)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 52)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 451)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 52)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 460)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 52)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 469)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 52)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 478)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 52)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 424)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 340)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 433)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 340)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 442)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 340)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 451)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 340)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 460)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 340)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 469)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 340)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 478)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 340)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 425)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 53)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 434)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 53)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 443)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 53)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 452)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 53)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 461)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 53)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 470)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 53)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 479)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 53)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 425)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 341)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 434)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 341)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 443)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 341)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 452)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 341)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 461)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 341)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 470)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 341)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 479)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 341)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 504)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 60)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 513)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 60)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 522)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 60)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 531)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 60)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 540)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 60)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 549)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 60)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 558)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 60)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 504)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 348)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 513)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 348)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 522)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 348)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 531)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 348)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 540)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 348)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 549)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 348)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 558)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 348)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 505)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 61)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 514)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 61)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 523)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 61)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 532)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 61)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 541)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 61)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 550)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 61)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 559)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 61)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 505)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 349)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 514)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 349)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 523)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 349)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 532)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 349)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 541)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 349)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 550)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 349)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 559)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 349)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 506)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 62)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 515)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 62)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 524)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 62)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 533)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 62)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 542)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 62)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 551)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 62)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 560)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 62)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 506)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 350)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 515)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 350)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 524)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 350)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 533)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 350)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 542)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 350)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 551)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 350)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 560)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 350)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 585)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 69)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 594)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 69)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 603)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 69)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 612)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 69)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 621)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 69)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 630)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 69)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 639)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 69)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 585)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 357)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 594)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 357)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 603)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 357)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 612)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 357)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 621)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 357)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 630)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 357)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 639)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 357)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 586)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 70)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 595)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 70)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 604)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 70)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 613)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 70)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 622)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 70)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 631)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 70)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 640)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 70)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 586)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 358)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 595)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 358)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 604)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 358)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 613)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 358)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 622)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 358)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 631)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 358)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 640)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 358)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 587)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 71)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 596)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 71)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 605)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 71)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 614)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 71)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 623)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 71)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 632)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 71)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 641)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 71)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 587)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 359)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 596)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 359)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 605)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 359)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 614)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 359)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 623)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 359)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 632)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 359)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 648) + (((int)threadIdx.x) % 7)) + 641)] * kernel_shared[((((((int)threadIdx.x) / 7) * 576) + (rc_outer_inner * 72)) + 359)]));
}
}
for (int i1_inner = 0; i1_inner < 2; ++i1_inner) {
- compute[((((((int)blockIdx.x) * 1568) + ((((int)threadIdx.x) / 49) * 98)) + (i1_inner * 49)) + (((int)threadIdx.x) % 49))] = max((conv2d_nchw[i1_inner] + bias[(((((int)blockIdx.x) * 32) + ((((int)threadIdx.x) / 49) * 2)) + i1_inner)]), 0.000000e+00f);
+ compute[((((((int)blockIdx.x) * 784) + ((((int)threadIdx.x) / 7) * 98)) + (i1_inner * 49)) + (((int)threadIdx.x) % 7))] = max((conv2d_nchw[i1_inner] + bias[(((((int)blockIdx.x) * 16) + ((((int)threadIdx.x) / 7) * 2)) + i1_inner)]), 0.000000e+00f);
+ compute[(((((((int)blockIdx.x) * 784) + ((((int)threadIdx.x) / 7) * 98)) + (i1_inner * 49)) + (((int)threadIdx.x) % 7)) + 7)] = max((conv2d_nchw[(i1_inner + 2)] + bias[(((((int)blockIdx.x) * 16) + ((((int)threadIdx.x) / 7) * 2)) + i1_inner)]), 0.000000e+00f);
+ compute[(((((((int)blockIdx.x) * 784) + ((((int)threadIdx.x) / 7) * 98)) + (i1_inner * 49)) + (((int)threadIdx.x) % 7)) + 14)] = max((conv2d_nchw[(i1_inner + 4)] + bias[(((((int)blockIdx.x) * 16) + ((((int)threadIdx.x) / 7) * 2)) + i1_inner)]), 0.000000e+00f);
+ compute[(((((((int)blockIdx.x) * 784) + ((((int)threadIdx.x) / 7) * 98)) + (i1_inner * 49)) + (((int)threadIdx.x) % 7)) + 21)] = max((conv2d_nchw[(i1_inner + 6)] + bias[(((((int)blockIdx.x) * 16) + ((((int)threadIdx.x) / 7) * 2)) + i1_inner)]), 0.000000e+00f);
+ compute[(((((((int)blockIdx.x) * 784) + ((((int)threadIdx.x) / 7) * 98)) + (i1_inner * 49)) + (((int)threadIdx.x) % 7)) + 28)] = max((conv2d_nchw[(i1_inner + 8)] + bias[(((((int)blockIdx.x) * 16) + ((((int)threadIdx.x) / 7) * 2)) + i1_inner)]), 0.000000e+00f);
+ compute[(((((((int)blockIdx.x) * 784) + ((((int)threadIdx.x) / 7) * 98)) + (i1_inner * 49)) + (((int)threadIdx.x) % 7)) + 35)] = max((conv2d_nchw[(i1_inner + 10)] + bias[(((((int)blockIdx.x) * 16) + ((((int)threadIdx.x) / 7) * 2)) + i1_inner)]), 0.000000e+00f);
+ compute[(((((((int)blockIdx.x) * 784) + ((((int)threadIdx.x) / 7) * 98)) + (i1_inner * 49)) + (((int)threadIdx.x) % 7)) + 42)] = max((conv2d_nchw[(i1_inner + 12)] + bias[(((((int)blockIdx.x) * 16) + ((((int)threadIdx.x) / 7) * 2)) + i1_inner)]), 0.000000e+00f);
}
}
</pre></div>
@@ -736,7 +3164,7 @@ In the example below we resume the status and do more 5 trials.</p>
Get devices for measurement successfully!
</pre></div>
</div>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 6 minutes 47.248 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 6 minutes 12.870 seconds)</p>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-tune-with-autoscheduler-tune-conv2d-layer-cuda-py">
<div class="sphx-glr-download sphx-glr-download-python docutils container">
<p><a class="reference download internal" download="" href="../../_downloads/e3e540f3b477c0c52d8eb73e674e8ffd/tune_conv2d_layer_cuda.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">tune_conv2d_layer_cuda.py</span></code></a></p>
diff --git a/docs/how_to/tune_with_autoscheduler/tune_network_cuda.html b/docs/how_to/tune_with_autoscheduler/tune_network_cuda.html
index cfd24b9c0f..5c5a489b30 100644
--- a/docs/how_to/tune_with_autoscheduler/tune_network_cuda.html
+++ b/docs/how_to/tune_with_autoscheduler/tune_network_cuda.html
@@ -921,7 +921,7 @@ so we can read the log file and load the best schedules.</p>
Evaluate inference time cost...
Execution time summary:
mean (ms) median (ms) max (ms) min (ms) std (ms)
- 7.9360 7.9369 7.9395 7.9317 0.0032
+ 7.8932 7.8938 7.9052 7.8806 0.0101
</pre></div>
</div>
</div>
@@ -943,7 +943,7 @@ to learn how to use the RPC Tracker and RPC Server.
To use the RPC Tracker in auto-scheduler, replace the runner in <code class="code docutils literal notranslate"><span class="pre">TuningOptions</span></code>
with <a class="reference internal" href="../../reference/api/python/auto_scheduler.html#tvm.auto_scheduler.RPCRunner" title="tvm.auto_scheduler.RPCRunner"><code class="xref any py py-class docutils literal notranslate"><span class="pre">auto_scheduler.RPCRunner</span></code></a>.</p></li>
</ol>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 11.065 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 6.524 seconds)</p>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-tune-with-autoscheduler-tune-network-cuda-py">
<div class="sphx-glr-download sphx-glr-download-python docutils container">
<p><a class="reference download internal" download="" href="../../_downloads/eafe360d52540634c9eea0fa89e804bd/tune_network_cuda.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">tune_network_cuda.py</span></code></a></p>
diff --git a/docs/how_to/tune_with_autoscheduler/tune_network_x86.html b/docs/how_to/tune_with_autoscheduler/tune_network_x86.html
index b52e14d331..08e83eaf08 100644
--- a/docs/how_to/tune_with_autoscheduler/tune_network_x86.html
+++ b/docs/how_to/tune_with_autoscheduler/tune_network_x86.html
@@ -940,7 +940,7 @@ so we can read the log file and load the best schedules.</p>
Evaluate inference time cost...
Execution time summary:
mean (ms) median (ms) max (ms) min (ms) std (ms)
- 764.1290 765.7904 765.8056 760.7911 2.3603
+ 746.5936 746.7246 747.6161 745.4401 0.8932
</pre></div>
</div>
</div>
@@ -962,7 +962,7 @@ to learn how to use the RPC Tracker and RPC Server.
To use the RPC Tracker in auto-scheduler, replace the runner in <code class="code docutils literal notranslate"><span class="pre">TuningOptions</span></code>
with <a class="reference internal" href="../../reference/api/python/auto_scheduler.html#tvm.auto_scheduler.RPCRunner" title="tvm.auto_scheduler.RPCRunner"><code class="xref any py py-class docutils literal notranslate"><span class="pre">auto_scheduler.RPCRunner</span></code></a>.</p></li>
</ol>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 48.279 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 40.793 seconds)</p>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-tune-with-autoscheduler-tune-network-x86-py">
<div class="sphx-glr-download sphx-glr-download-python docutils container">
<p><a class="reference download internal" download="" href="../../_downloads/e416b94ca1090b0897c0f6e0df95b911/tune_network_x86.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">tune_network_x86.py</span></code></a></p>
diff --git a/docs/how_to/tune_with_autoscheduler/tune_sparse_x86.html b/docs/how_to/tune_with_autoscheduler/tune_sparse_x86.html
index db0ec465a5..a20a9f0543 100644
--- a/docs/how_to/tune_with_autoscheduler/tune_sparse_x86.html
+++ b/docs/how_to/tune_with_autoscheduler/tune_sparse_x86.html
@@ -637,12 +637,12 @@ class Module:
@T.prim_func
def main(placeholder: T.Buffer((128, 256), "float32"), placeholder_1: T.Buffer((4916, 16, 1), "float32"), placeholder_2: T.Buffer((4916,), "int32"), placeholder_3: T.Buffer((33,), "int32"), placeholder_4: T.Buffer((128, 512), "float32"), compute: T.Buffer((128, 512), "float32")):
T.func_attr({"from_legacy_te_schedule": True, "global_symbol": "main", "tir.noalias": True})
- for i0_outer_i1_outer_fused in T.parallel(32):
- compute_1 = T.allocate([2048], "float32", "global")
- compute_2 = T.Buffer((2048,), data=compute_1)
- for i_outer_inner in range(8):
+ for i0_outer_i1_outer_fused in T.parallel(128):
+ compute_1 = T.allocate([512], "float32", "global")
+ compute_2 = T.Buffer((512,), data=compute_1)
+ for nb_j_inner in range(2):
for i_inner_init in range(16):
- cse_var_1: T.int32 = i_outer_inner * 256 + i_inner_init * 16
+ cse_var_1: T.int32 = i_inner_init * 32 + nb_j_inner * 16
compute_2[cse_var_1] = T.float32(0)
compute_2[cse_var_1 + 1] = T.float32(0)
compute_2[cse_var_1 + 2] = T.float32(0)
@@ -659,64 +659,52 @@ class Module:
compute_2[cse_var_1 + 13] = T.float32(0)
compute_2[cse_var_1 + 14] = T.float32(0)
compute_2[cse_var_1 + 15] = T.float32(0)
- for elem_idx, i_inner in T.grid(placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused], 16):
+ for elem_idx, i_inner in T.grid(T.Let(placeholder_5[cse_var_2 + 1] - placeholder_5[cse_var_2], where={cse_var_2: i0_outer_i1_outer_fused % 16 * 2 + nb_j_inner}), 16):
+ cse_var_2 = T.int32()
placeholder_5 = T.Buffer((33,), "int32", data=placeholder_3.data)
+ cse_var_21: T.int32 = elem_idx * 16
+ cse_var_20: T.int32 = i_inner * 32 + nb_j_inner * 16
+ cse_var_19: T.int32 = i0_outer_i1_outer_fused % 16 * 2 + nb_j_inner
+ cse_var_18: T.int32 = i0_outer_i1_outer_fused // 16 * 4096 + i_inner * 256
+ cse_var_17: T.int32 = cse_var_20 + 9
+ cse_var_16: T.int32 = cse_var_20 + 8
+ cse_var_15: T.int32 = cse_var_20 + 7
+ cse_var_14: T.int32 = cse_var_20 + 6
+ cse_var_13: T.int32 = cse_var_20 + 5
+ cse_var_12: T.int32 = cse_var_20 + 4
+ cse_var_11: T.int32 = cse_var_20 + 3
+ cse_var_10: T.int32 = cse_var_20 + 2
+ cse_var_9: T.int32 = cse_var_20 + 15
+ cse_var_8: T.int32 = cse_var_20 + 14
+ cse_var_7: T.int32 = cse_var_20 + 13
+ cse_var_6: T.int32 = cse_var_20 + 12
+ cse_var_5: T.int32 = cse_var_20 + 11
+ cse_var_4: T.int32 = cse_var_20 + 10
+ cse_var_3: T.int32 = cse_var_20 + 1
placeholder_6 = T.Buffer((78656,), data=placeholder_1.data)
placeholder_7 = T.Buffer((32768,), data=placeholder.data)
placeholder_8 = T.Buffer((4916,), "int32", data=placeholder_2.data)
- if T.likely(elem_idx < placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused]):
- cse_var_2: T.int32 = i_outer_inner * 256 + i_inner * 16
- compute_2[cse_var_2] = compute_2[cse_var_2] + placeholder_6[placeholder_5[i0_outer_i1_outer_fused] * 16 + elem_idx * 16] * T.max(placeholder_7[i_outer_inner * 4096 + i_inner * 256 + placeholder_8[placeholder_5[i0_outer_i1_outer_fused] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused]):
- cse_var_3: T.int32 = i_outer_inner * 256 + i_inner * 16 + 1
- compute_2[cse_var_3] = compute_2[cse_var_3] + placeholder_6[placeholder_5[i0_outer_i1_outer_fused] * 16 + elem_idx * 16 + 1] * T.max(placeholder_7[i_outer_inner * 4096 + i_inner * 256 + placeholder_8[placeholder_5[i0_outer_i1_outer_fused] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused]):
- cse_var_4: T.int32 = i_outer_inner * 256 + i_inner * 16 + 2
- compute_2[cse_var_4] = compute_2[cse_var_4] + placeholder_6[placeholder_5[i0_outer_i1_outer_fused] * 16 + elem_idx * 16 + 2] * T.max(placeholder_7[i_outer_inner * 4096 + i_inner * 256 + placeholder_8[placeholder_5[i0_outer_i1_outer_fused] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused]):
- cse_var_5: T.int32 = i_outer_inner * 256 + i_inner * 16 + 3
- compute_2[cse_var_5] = compute_2[cse_var_5] + placeholder_6[placeholder_5[i0_outer_i1_outer_fused] * 16 + elem_idx * 16 + 3] * T.max(placeholder_7[i_outer_inner * 4096 + i_inner * 256 + placeholder_8[placeholder_5[i0_outer_i1_outer_fused] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused]):
- cse_var_6: T.int32 = i_outer_inner * 256 + i_inner * 16 + 4
- compute_2[cse_var_6] = compute_2[cse_var_6] + placeholder_6[placeholder_5[i0_outer_i1_outer_fused] * 16 + elem_idx * 16 + 4] * T.max(placeholder_7[i_outer_inner * 4096 + i_inner * 256 + placeholder_8[placeholder_5[i0_outer_i1_outer_fused] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused]):
- cse_var_7: T.int32 = i_outer_inner * 256 + i_inner * 16 + 5
- compute_2[cse_var_7] = compute_2[cse_var_7] + placeholder_6[placeholder_5[i0_outer_i1_outer_fused] * 16 + elem_idx * 16 + 5] * T.max(placeholder_7[i_outer_inner * 4096 + i_inner * 256 + placeholder_8[placeholder_5[i0_outer_i1_outer_fused] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused]):
- cse_var_8: T.int32 = i_outer_inner * 256 + i_inner * 16 + 6
- compute_2[cse_var_8] = compute_2[cse_var_8] + placeholder_6[placeholder_5[i0_outer_i1_outer_fused] * 16 + elem_idx * 16 + 6] * T.max(placeholder_7[i_outer_inner * 4096 + i_inner * 256 + placeholder_8[placeholder_5[i0_outer_i1_outer_fused] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused]):
- cse_var_9: T.int32 = i_outer_inner * 256 + i_inner * 16 + 7
- compute_2[cse_var_9] = compute_2[cse_var_9] + placeholder_6[placeholder_5[i0_outer_i1_outer_fused] * 16 + elem_idx * 16 + 7] * T.max(placeholder_7[i_outer_inner * 4096 + i_inner * 256 + placeholder_8[placeholder_5[i0_outer_i1_outer_fused] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused]):
- cse_var_10: T.int32 = i_outer_inner * 256 + i_inner * 16 + 8
- compute_2[cse_var_10] = compute_2[cse_var_10] + placeholder_6[placeholder_5[i0_outer_i1_outer_fused] * 16 + elem_idx * 16 + 8] * T.max(placeholder_7[i_outer_inner * 4096 + i_inner * 256 + placeholder_8[placeholder_5[i0_outer_i1_outer_fused] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused]):
- cse_var_11: T.int32 = i_outer_inner * 256 + i_inner * 16 + 9
- compute_2[cse_var_11] = compute_2[cse_var_11] + placeholder_6[placeholder_5[i0_outer_i1_outer_fused] * 16 + elem_idx * 16 + 9] * T.max(placeholder_7[i_outer_inner * 4096 + i_inner * 256 + placeholder_8[placeholder_5[i0_outer_i1_outer_fused] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused]):
- cse_var_12: T.int32 = i_outer_inner * 256 + i_inner * 16 + 10
- compute_2[cse_var_12] = compute_2[cse_var_12] + placeholder_6[placeholder_5[i0_outer_i1_outer_fused] * 16 + elem_idx * 16 + 10] * T.max(placeholder_7[i_outer_inner * 4096 + i_inner * 256 + placeholder_8[placeholder_5[i0_outer_i1_outer_fused] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused]):
- cse_var_13: T.int32 = i_outer_inner * 256 + i_inner * 16 + 11
- compute_2[cse_var_13] = compute_2[cse_var_13] + placeholder_6[placeholder_5[i0_outer_i1_outer_fused] * 16 + elem_idx * 16 + 11] * T.max(placeholder_7[i_outer_inner * 4096 + i_inner * 256 + placeholder_8[placeholder_5[i0_outer_i1_outer_fused] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused]):
- cse_var_14: T.int32 = i_outer_inner * 256 + i_inner * 16 + 12
- compute_2[cse_var_14] = compute_2[cse_var_14] + placeholder_6[placeholder_5[i0_outer_i1_outer_fused] * 16 + elem_idx * 16 + 12] * T.max(placeholder_7[i_outer_inner * 4096 + i_inner * 256 + placeholder_8[placeholder_5[i0_outer_i1_outer_fused] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused]):
- cse_var_15: T.int32 = i_outer_inner * 256 + i_inner * 16 + 13
- compute_2[cse_var_15] = compute_2[cse_var_15] + placeholder_6[placeholder_5[i0_outer_i1_outer_fused] * 16 + elem_idx * 16 + 13] * T.max(placeholder_7[i_outer_inner * 4096 + i_inner * 256 + placeholder_8[placeholder_5[i0_outer_i1_outer_fused] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused]):
- cse_var_16: T.int32 = i_outer_inner * 256 + i_inner * 16 + 14
- compute_2[cse_var_16] = compute_2[cse_var_16] + placeholder_6[placeholder_5[i0_outer_i1_outer_fused] * 16 + elem_idx * 16 + 14] * T.max(placeholder_7[i_outer_inner * 4096 + i_inner * 256 + placeholder_8[placeholder_5[i0_outer_i1_outer_fused] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused]):
- cse_var_17: T.int32 = i_outer_inner * 256 + i_inner * 16 + 15
- compute_2[cse_var_17] = compute_2[cse_var_17] + placeholder_6[placeholder_5[i0_outer_i1_outer_fused] * 16 + elem_idx * 16 + 15] * T.max(placeholder_7[i_outer_inner * 4096 + i_inner * 256 + placeholder_8[placeholder_5[i0_outer_i1_outer_fused] + elem_idx]], T.float32(0))
- for i0_inner in range(128):
- cse_var_18: T.int32 = i0_inner * 512 + i0_outer_i1_outer_fused * 16
+ compute_2[cse_var_20] = compute_2[cse_var_20] + placeholder_6[placeholder_5[cse_var_19] * 16 + cse_var_21] * T.max(placeholder_7[cse_var_18 + placeholder_8[placeholder_5[cse_var_19] + elem_idx]], T.float32(0))
+ compute_2[cse_var_3] = compute_2[cse_var_3] + placeholder_6[placeholder_5[cse_var_19] * 16 + cse_var_21 + 1] * T.max(placeholder_7[cse_var_18 + placeholder_8[placeholder_5[cse_var_19] + elem_idx]], T.float32(0))
+ compute_2[cse_var_10] = compute_2[cse_var_10] + placeholder_6[placeholder_5[cse_var_19] * 16 + cse_var_21 + 2] * T.max(placeholder_7[cse_var_18 + placeholder_8[placeholder_5[cse_var_19] + elem_idx]], T.float32(0))
+ compute_2[cse_var_11] = compute_2[cse_var_11] + placeholder_6[placeholder_5[cse_var_19] * 16 + cse_var_21 + 3] * T.max(placeholder_7[cse_var_18 + placeholder_8[placeholder_5[cse_var_19] + elem_idx]], T.float32(0))
+ compute_2[cse_var_12] = compute_2[cse_var_12] + placeholder_6[placeholder_5[cse_var_19] * 16 + cse_var_21 + 4] * T.max(placeholder_7[cse_var_18 + placeholder_8[placeholder_5[cse_var_19] + elem_idx]], T.float32(0))
+ compute_2[cse_var_13] = compute_2[cse_var_13] + placeholder_6[placeholder_5[cse_var_19] * 16 + cse_var_21 + 5] * T.max(placeholder_7[cse_var_18 + placeholder_8[placeholder_5[cse_var_19] + elem_idx]], T.float32(0))
+ compute_2[cse_var_14] = compute_2[cse_var_14] + placeholder_6[placeholder_5[cse_var_19] * 16 + cse_var_21 + 6] * T.max(placeholder_7[cse_var_18 + placeholder_8[placeholder_5[cse_var_19] + elem_idx]], T.float32(0))
+ compute_2[cse_var_15] = compute_2[cse_var_15] + placeholder_6[placeholder_5[cse_var_19] * 16 + cse_var_21 + 7] * T.max(placeholder_7[cse_var_18 + placeholder_8[placeholder_5[cse_var_19] + elem_idx]], T.float32(0))
+ compute_2[cse_var_16] = compute_2[cse_var_16] + placeholder_6[placeholder_5[cse_var_19] * 16 + cse_var_21 + 8] * T.max(placeholder_7[cse_var_18 + placeholder_8[placeholder_5[cse_var_19] + elem_idx]], T.float32(0))
+ compute_2[cse_var_17] = compute_2[cse_var_17] + placeholder_6[placeholder_5[cse_var_19] * 16 + cse_var_21 + 9] * T.max(placeholder_7[cse_var_18 + placeholder_8[placeholder_5[cse_var_19] + elem_idx]], T.float32(0))
+ compute_2[cse_var_4] = compute_2[cse_var_4] + placeholder_6[placeholder_5[cse_var_19] * 16 + cse_var_21 + 10] * T.max(placeholder_7[cse_var_18 + placeholder_8[placeholder_5[cse_var_19] + elem_idx]], T.float32(0))
+ compute_2[cse_var_5] = compute_2[cse_var_5] + placeholder_6[placeholder_5[cse_var_19] * 16 + cse_var_21 + 11] * T.max(placeholder_7[cse_var_18 + placeholder_8[placeholder_5[cse_var_19] + elem_idx]], T.float32(0))
+ compute_2[cse_var_6] = compute_2[cse_var_6] + placeholder_6[placeholder_5[cse_var_19] * 16 + cse_var_21 + 12] * T.max(placeholder_7[cse_var_18 + placeholder_8[placeholder_5[cse_var_19] + elem_idx]], T.float32(0))
+ compute_2[cse_var_7] = compute_2[cse_var_7] + placeholder_6[placeholder_5[cse_var_19] * 16 + cse_var_21 + 13] * T.max(placeholder_7[cse_var_18 + placeholder_8[placeholder_5[cse_var_19] + elem_idx]], T.float32(0))
+ compute_2[cse_var_8] = compute_2[cse_var_8] + placeholder_6[placeholder_5[cse_var_19] * 16 + cse_var_21 + 14] * T.max(placeholder_7[cse_var_18 + placeholder_8[placeholder_5[cse_var_19] + elem_idx]], T.float32(0))
+ compute_2[cse_var_9] = compute_2[cse_var_9] + placeholder_6[placeholder_5[cse_var_19] * 16 + cse_var_21 + 15] * T.max(placeholder_7[cse_var_18 + placeholder_8[placeholder_5[cse_var_19] + elem_idx]], T.float32(0))
+ for i0_inner in range(16):
+ cse_var_22: T.int32 = i0_outer_i1_outer_fused // 16 * 8192 + i0_inner * 512 + i0_outer_i1_outer_fused % 16 * 32
compute_3 = T.Buffer((65536,), data=compute.data)
placeholder_5 = T.Buffer((65536,), data=placeholder_4.data)
- compute_3[cse_var_18:cse_var_18 + 16] = T.max(compute_2[i0_inner * 16:i0_inner * 16 + 16] + placeholder_5[cse_var_18:cse_var_18 + 16], T.Broadcast(T.float32(0), 16))
+ compute_3[cse_var_22:cse_var_22 + 32] = T.max(compute_2[i0_inner * 32:i0_inner * 32 + 32] + placeholder_5[cse_var_22:cse_var_22 + 32], T.Broadcast(T.float32(0), 32))
</pre></div>
</div>
</div>
@@ -750,7 +738,7 @@ class Module:
<span class="p">)</span>
</pre></div>
</div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Execution time of this operator: 1.700 ms
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Execution time of this operator: 1.670 ms
</pre></div>
</div>
<div class="admonition note">
diff --git a/docs/how_to/tune_with_autotvm/sg_execution_times.html b/docs/how_to/tune_with_autotvm/sg_execution_times.html
index 4f9291f451..c8c6d748bd 100644
--- a/docs/how_to/tune_with_autotvm/sg_execution_times.html
+++ b/docs/how_to/tune_with_autotvm/sg_execution_times.html
@@ -345,7 +345,7 @@
<div class="section" id="computation-times">
<span id="sphx-glr-how-to-tune-with-autotvm-sg-execution-times"></span><h1>Computation times<a class="headerlink" href="#computation-times" title="Permalink to this headline">¶</a></h1>
-<p><strong>00:35.388</strong> total execution time for <strong>how_to_tune_with_autotvm</strong> files:</p>
+<p><strong>00:45.727</strong> total execution time for <strong>how_to_tune_with_autotvm</strong> files:</p>
<table class="docutils align-default">
<colgroup>
<col style="width: 84%" />
@@ -354,11 +354,11 @@
</colgroup>
<tbody>
<tr class="row-odd"><td><p><a class="reference internal" href="tune_conv2d_cuda.html#sphx-glr-how-to-tune-with-autotvm-tune-conv2d-cuda-py"><span class="std std-ref">Tuning High Performance Convolution on NVIDIA GPUs</span></a> (<code class="docutils literal notranslate"><span class="pre">tune_conv2d_cuda.py</span></code>)</p></td>
-<td><p>00:35.352</p></td>
+<td><p>00:45.695</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="tune_relay_x86.html#sphx-glr-how-to-tune-with-autotvm-tune-relay-x86-py"><span class="std std-ref">Auto-tuning a Convolutional Network for x86 CPU</span></a> (<code class="docutils literal notranslate"><span class="pre">tune_relay_x86.py</span></code>)</p></td>
-<td><p>00:00.021</p></td>
+<td><p>00:00.019</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="tune_relay_cuda.html#sphx-glr-how-to-tune-with-autotvm-tune-relay-cuda-py"><span class="std std-ref">Auto-tuning a Convolutional Network for NVIDIA GPU</span></a> (<code class="docutils literal notranslate"><span class="pre">tune_relay_cuda.py</span></code>)</p></td>
@@ -366,11 +366,11 @@
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="tune_relay_arm.html#sphx-glr-how-to-tune-with-autotvm-tune-relay-arm-py"><span class="std std-ref">Auto-tuning a Convolutional Network for ARM CPU</span></a> (<code class="docutils literal notranslate"><span class="pre">tune_relay_arm.py</span></code>)</p></td>
-<td><p>00:00.005</p></td>
+<td><p>00:00.004</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="tune_relay_mobile_gpu.html#sphx-glr-how-to-tune-with-autotvm-tune-relay-mobile-gpu-py"><span class="std std-ref">Auto-tuning a Convolutional Network for Mobile GPU</span></a> (<code class="docutils literal notranslate"><span class="pre">tune_relay_mobile_gpu.py</span></code>)</p></td>
-<td><p>00:00.005</p></td>
+<td><p>00:00.004</p></td>
<td><p>0.0 MB</p></td>
</tr>
</tbody>
diff --git a/docs/how_to/tune_with_autotvm/tune_conv2d_cuda.html b/docs/how_to/tune_with_autotvm/tune_conv2d_cuda.html
index 6a85a8ad6b..d6f67f7356 100644
--- a/docs/how_to/tune_with_autotvm/tune_conv2d_cuda.html
+++ b/docs/how_to/tune_with_autotvm/tune_conv2d_cuda.html
@@ -695,501 +695,162 @@ Traceback (most recent call last):
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 1, 1, 16]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 64, 1]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,3314684
-No: 2 GFLOPS: 0.00/0.00 result: Traceback (most recent call last):
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
- func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
- func = build(s, args, target=target, runtime=runtime)
- File "/workspace/python/tvm/driver/build_module.py", line 227, in build
- input_mod = lower(inputs, args, name=name, binds=binds)
- File "/workspace/python/tvm/driver/build_module.py", line 134, in lower
- return ffi.lower_schedule(inp, args, name, binds, simple_mode)
- File "tvm/_ffi/_cython/./packed_func.pxi", line 331, in tvm._ffi._cy3.core.PackedFuncBase.__call__
- File "tvm/_ffi/_cython/./packed_func.pxi", line 276, in tvm._ffi._cy3.core.FuncCall
- File "tvm/_ffi/_cython/./base.pxi", line 181, in tvm._ffi._cy3.core.CHECK_CALL
-tvm._ffi.base.TVMError: Traceback (most recent call last):
- 24: TVMFuncCall
- at ../src/runtime/c_runtime_api.cc:477
- 23: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 22: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 21: operator()
- at ../include/tvm/runtime/packed_func.h:1734
- 20: unpack_call<tvm::IRModule, 5, tvm::<lambda(tvm::te::Schedule, const tvm::runtime::Array<tvm::runtime::ObjectRef>&, const tvm::runtime::String&, const tvm::runtime::Map<tvm::te::Tensor, tvm::tir::Buffer>&, bool)> >
- at ../include/tvm/runtime/packed_func.h:1674
- 19: run<>
- at ../include/tvm/runtime/packed_func.h:1634
- 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1649
- 13: operator()
- at ../src/driver/driver_api.cc:402
- 12: tvm::LowerSchedule(tvm::te::Schedule, tvm::runtime::Array<tvm::runtime::ObjectRef, void> const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::unordered_map<tvm::te::Tensor, tvm::tir::Buffer, std::hash<tvm::te::Tensor>, std::equal_to<tvm::te::Tensor>, std::allocator<std::pair<tvm::te::Tensor const, tvm::tir::Buffer> > > const&, tvm::GlobalVarSupply, bool)
- at ../src/driver/driver_api.cc:388
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:283
- 10: tvm::transform::Pass::operator()(tvm::IRModule) const
- at ../src/ir/transform.cc:258
- 9: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 8: tvm::transform::SequentialNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:451
- 7: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 6: tvm::tir::transform::PrimFuncPassNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/tir/ir/transform.cc:100
- 5: tvm::runtime::TypedPackedFunc<tvm::tir::PrimFunc (tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext)>::operator()(tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext) const
- at ../include/tvm/runtime/packed_func.h:1753
- 4: tvm::tir::PrimFunc tvm::runtime::detail::typed_packed_call_dispatcher<tvm::tir::PrimFunc>::run<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::runtime::PackedFunc const&, tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&)
- at ../include/tvm/runtime/packed_func.h:1697
- 3: tvm::runtime::TVMRetValue tvm::runtime::PackedFunc::operator()<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&) const
- at ../include/tvm/runtime/packed_func.h:1621
- 2: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 1: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 0: operator()
- at ../src/runtime/c_runtime_api.cc:534
- File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
- raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel
-
-Traceback (most recent call last):
- 24: TVMFuncCall
- at ../src/runtime/c_runtime_api.cc:477
- 23: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 22: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 21: operator()
- at ../include/tvm/runtime/packed_func.h:1734
- 20: unpack_call<tvm::IRModule, 5, tvm::<lambda(tvm::te::Schedule, const tvm::runtime::Array<tvm::runtime::ObjectRef>&, const tvm::runtime::String&, const tvm::runtime::Map<tvm::te::Tensor, tvm::tir::Buffer>&, bool)> >
- at ../include/tvm/runtime/packed_func.h:1674
- 19: run<>
- at ../include/tvm/runtime/packed_func.h:1634
- 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1649
- 13: operator()
- at ../src/driver/driver_api.cc:402
- 12: tvm::LowerSchedule(tvm::te::Schedule, tvm::runtime::Array<tvm::runtime::ObjectRef, void> const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::unordered_map<tvm::te::Tensor, tvm::tir::Buffer, std::hash<tvm::te::Tensor>, std::equal_to<tvm::te::Tensor>, std::allocator<std::pair<tvm::te::Tensor const, tvm::tir::Buffer> > > const&, tvm::GlobalVarSupply, bool)
- at ../src/driver/driver_api.cc:388
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:283
- 10: tvm::transform::Pass::operator()(tvm::IRModule) const
- at ../src/ir/transform.cc:258
- 9: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 8: tvm::transform::SequentialNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:451
- 7: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 6: tvm::tir::transform::PrimFuncPassNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/tir/ir/transform.cc:100
- 5: tvm::runtime::TypedPackedFunc<tvm::tir::PrimFunc (tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext)>::operator()(tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext) const
- at ../include/tvm/runtime/packed_func.h:1753
- 4: tvm::tir::PrimFunc tvm::runtime::detail::typed_packed_call_dispatcher<tvm::tir::PrimFunc>::run<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::runtime::PackedFunc const&, tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&)
- at ../include/tvm/runtime/packed_func.h:1697
- 3: tvm::runtime::TVMRetValue tvm::runtime::PackedFunc::operator()<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&) const
- at ../include/tvm/runtime/packed_func.h:1621
- 2: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 1: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 0: operator()
- at ../src/runtime/c_runtime_api.cc:534
- File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
- raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 64, 1, 2]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 64, 2]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,2769201
-No: 3 GFLOPS: 0.00/0.00 result: Traceback (most recent call last):
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
- func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
- func = build(s, args, target=target, runtime=runtime)
- File "/workspace/python/tvm/driver/build_module.py", line 227, in build
- input_mod = lower(inputs, args, name=name, binds=binds)
- File "/workspace/python/tvm/driver/build_module.py", line 134, in lower
- return ffi.lower_schedule(inp, args, name, binds, simple_mode)
- File "tvm/_ffi/_cython/./packed_func.pxi", line 331, in tvm._ffi._cy3.core.PackedFuncBase.__call__
- File "tvm/_ffi/_cython/./packed_func.pxi", line 276, in tvm._ffi._cy3.core.FuncCall
- File "tvm/_ffi/_cython/./base.pxi", line 181, in tvm._ffi._cy3.core.CHECK_CALL
-tvm._ffi.base.TVMError: Traceback (most recent call last):
- 24: TVMFuncCall
- at ../src/runtime/c_runtime_api.cc:477
- 23: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 22: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 21: operator()
- at ../include/tvm/runtime/packed_func.h:1734
- 20: unpack_call<tvm::IRModule, 5, tvm::<lambda(tvm::te::Schedule, const tvm::runtime::Array<tvm::runtime::ObjectRef>&, const tvm::runtime::String&, const tvm::runtime::Map<tvm::te::Tensor, tvm::tir::Buffer>&, bool)> >
- at ../include/tvm/runtime/packed_func.h:1674
- 19: run<>
- at ../include/tvm/runtime/packed_func.h:1634
- 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1649
- 13: operator()
- at ../src/driver/driver_api.cc:402
- 12: tvm::LowerSchedule(tvm::te::Schedule, tvm::runtime::Array<tvm::runtime::ObjectRef, void> const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::unordered_map<tvm::te::Tensor, tvm::tir::Buffer, std::hash<tvm::te::Tensor>, std::equal_to<tvm::te::Tensor>, std::allocator<std::pair<tvm::te::Tensor const, tvm::tir::Buffer> > > const&, tvm::GlobalVarSupply, bool)
- at ../src/driver/driver_api.cc:388
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:283
- 10: tvm::transform::Pass::operator()(tvm::IRModule) const
- at ../src/ir/transform.cc:258
- 9: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 8: tvm::transform::SequentialNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:451
- 7: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 6: tvm::tir::transform::PrimFuncPassNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/tir/ir/transform.cc:100
- 5: tvm::runtime::TypedPackedFunc<tvm::tir::PrimFunc (tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext)>::operator()(tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext) const
- at ../include/tvm/runtime/packed_func.h:1753
- 4: tvm::tir::PrimFunc tvm::runtime::detail::typed_packed_call_dispatcher<tvm::tir::PrimFunc>::run<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::runtime::PackedFunc const&, tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&)
- at ../include/tvm/runtime/packed_func.h:1697
- 3: tvm::runtime::TVMRetValue tvm::runtime::PackedFunc::operator()<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&) const
- at ../include/tvm/runtime/packed_func.h:1621
- 2: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 1: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 0: operator()
- at ../src/runtime/c_runtime_api.cc:534
- File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
- raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel
-
-Traceback (most recent call last):
- 24: TVMFuncCall
- at ../src/runtime/c_runtime_api.cc:477
- 23: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 22: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 21: operator()
- at ../include/tvm/runtime/packed_func.h:1734
- 20: unpack_call<tvm::IRModule, 5, tvm::<lambda(tvm::te::Schedule, const tvm::runtime::Array<tvm::runtime::ObjectRef>&, const tvm::runtime::String&, const tvm::runtime::Map<tvm::te::Tensor, tvm::tir::Buffer>&, bool)> >
- at ../include/tvm/runtime/packed_func.h:1674
- 19: run<>
- at ../include/tvm/runtime/packed_func.h:1634
- 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1649
- 13: operator()
- at ../src/driver/driver_api.cc:402
- 12: tvm::LowerSchedule(tvm::te::Schedule, tvm::runtime::Array<tvm::runtime::ObjectRef, void> const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::unordered_map<tvm::te::Tensor, tvm::tir::Buffer, std::hash<tvm::te::Tensor>, std::equal_to<tvm::te::Tensor>, std::allocator<std::pair<tvm::te::Tensor const, tvm::tir::Buffer> > > const&, tvm::GlobalVarSupply, bool)
- at ../src/driver/driver_api.cc:388
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:283
- 10: tvm::transform::Pass::operator()(tvm::IRModule) const
- at ../src/ir/transform.cc:258
- 9: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 8: tvm::transform::SequentialNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:451
- 7: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 6: tvm::tir::transform::PrimFuncPassNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/tir/ir/transform.cc:100
- 5: tvm::runtime::TypedPackedFunc<tvm::tir::PrimFunc (tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext)>::operator()(tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext) const
- at ../include/tvm/runtime/packed_func.h:1753
- 4: tvm::tir::PrimFunc tvm::runtime::detail::typed_packed_call_dispatcher<tvm::tir::PrimFunc>::run<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::runtime::PackedFunc const&, tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&)
- at ../include/tvm/runtime/packed_func.h:1697
- 3: tvm::runtime::TVMRetValue tvm::runtime::PackedFunc::operator()<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&) const
- at ../include/tvm/runtime/packed_func.h:1621
- 2: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 1: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 0: operator()
- at ../src/runtime/c_runtime_api.cc:534
- File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
- raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 1, 4, 32]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 4, 32]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,8280114
-No: 4 GFLOPS: 0.00/0.00 result: Traceback (most recent call last):
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
- func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
- func = build(s, args, target=target, runtime=runtime)
- File "/workspace/python/tvm/driver/build_module.py", line 227, in build
- input_mod = lower(inputs, args, name=name, binds=binds)
- File "/workspace/python/tvm/driver/build_module.py", line 134, in lower
- return ffi.lower_schedule(inp, args, name, binds, simple_mode)
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 1, 1, 256]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 512, 1]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,4681596
+No: 2 GFLOPS: 42.72/42.72 result: MeasureResult(costs=(0.005419384523809524,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.903918981552124, timestamp=1679085229.6098208) [('tile_f', [-1, 16, 32, 1]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 1, 2]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,231264
+No: 3 GFLOPS: 0.00/42.72 result: Traceback (most recent call last):
+ File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 742, in __call__
+ yield remote, remote.load_module(os.path.split(build_result.filename)[1])
+ File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 706, in run_through_rpc
+ costs = time_f(*args).results
+ File "/workspace/python/tvm/runtime/module.py", line 357, in evaluator
+ blob = feval(*args)
File "tvm/_ffi/_cython/./packed_func.pxi", line 331, in tvm._ffi._cy3.core.PackedFuncBase.__call__
- File "tvm/_ffi/_cython/./packed_func.pxi", line 276, in tvm._ffi._cy3.core.FuncCall
+ File "tvm/_ffi/_cython/./packed_func.pxi", line 262, in tvm._ffi._cy3.core.FuncCall
+ File "tvm/_ffi/_cython/./packed_func.pxi", line 251, in tvm._ffi._cy3.core.FuncCall3
File "tvm/_ffi/_cython/./base.pxi", line 181, in tvm._ffi._cy3.core.CHECK_CALL
tvm._ffi.base.TVMError: Traceback (most recent call last):
- 24: TVMFuncCall
+ 4: TVMFuncCall
at ../src/runtime/c_runtime_api.cc:477
- 23: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
+ 3: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
at ../include/tvm/runtime/packed_func.h:1217
- 22: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 21: operator()
- at ../include/tvm/runtime/packed_func.h:1734
- 20: unpack_call<tvm::IRModule, 5, tvm::<lambda(tvm::te::Schedule, const tvm::runtime::Array<tvm::runtime::ObjectRef>&, const tvm::runtime::String&, const tvm::runtime::Map<tvm::te::Tensor, tvm::tir::Buffer>&, bool)> >
- at ../include/tvm/runtime/packed_func.h:1674
- 19: run<>
- at ../include/tvm/runtime/packed_func.h:1634
- 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1649
- 13: operator()
- at ../src/driver/driver_api.cc:402
- 12: tvm::LowerSchedule(tvm::te::Schedule, tvm::runtime::Array<tvm::runtime::ObjectRef, void> const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::unordered_map<tvm::te::Tensor, tvm::tir::Buffer, std::hash<tvm::te::Tensor>, std::equal_to<tvm::te::Tensor>, std::allocator<std::pair<tvm::te::Tensor const, tvm::tir::Buffer> > > const&, tvm::GlobalVarSupply, bool)
- at ../src/driver/driver_api.cc:388
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:283
- 10: tvm::transform::Pass::operator()(tvm::IRModule) const
- at ../src/ir/transform.cc:258
- 9: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 8: tvm::transform::SequentialNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:451
- 7: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 6: tvm::tir::transform::PrimFuncPassNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/tir/ir/transform.cc:100
- 5: tvm::runtime::TypedPackedFunc<tvm::tir::PrimFunc (tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext)>::operator()(tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext) const
- at ../include/tvm/runtime/packed_func.h:1753
- 4: tvm::tir::PrimFunc tvm::runtime::detail::typed_packed_call_dispatcher<tvm::tir::PrimFunc>::run<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::runtime::PackedFunc const&, tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&)
- at ../include/tvm/runtime/packed_func.h:1697
- 3: tvm::runtime::TVMRetValue tvm::runtime::PackedFunc::operator()<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&) const
- at ../include/tvm/runtime/packed_func.h:1621
- 2: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 1: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 0: operator()
- at ../src/runtime/c_runtime_api.cc:534
- File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
- raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel
+ 2: tvm::runtime::RPCWrappedFunc::operator()(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
+ at ../src/runtime/rpc/rpc_module.cc:129
+ 1: tvm::runtime::RPCClientSession::CallFunc(void*, TVMValue const*, int const*, int, std::function<void (tvm::runtime::TVMArgs)> const&)
+ at ../src/runtime/rpc/rpc_endpoint.cc:1012
+ 0: tvm::runtime::RPCEndpoint::CallFunc(void*, TVMValue const*, int const*, int, std::function<void (tvm::runtime::TVMArgs)>)
+ at ../src/runtime/rpc/rpc_endpoint.cc:804
+ File "../src/runtime/rpc/rpc_endpoint.cc", line 804
+TVMError:
+---------------------------------------------------------------
+An error occurred during the execution of TVM.
+For more information, please see: https://tvm.apache.org/docs/errors.html
+---------------------------------------------------------------
+ Check failed: (code == RPCCode::kReturn) is false: code=kShutdown
+
+During handling of the above exception, another exception occurred:
Traceback (most recent call last):
- 24: TVMFuncCall
- at ../src/runtime/c_runtime_api.cc:477
- 23: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 22: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 21: operator()
- at ../include/tvm/runtime/packed_func.h:1734
- 20: unpack_call<tvm::IRModule, 5, tvm::<lambda(tvm::te::Schedule, const tvm::runtime::Array<tvm::runtime::ObjectRef>&, const tvm::runtime::String&, const tvm::runtime::Map<tvm::te::Tensor, tvm::tir::Buffer>&, bool)> >
- at ../include/tvm/runtime/packed_func.h:1674
- 19: run<>
- at ../include/tvm/runtime/packed_func.h:1634
- 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1649
- 13: operator()
- at ../src/driver/driver_api.cc:402
- 12: tvm::LowerSchedule(tvm::te::Schedule, tvm::runtime::Array<tvm::runtime::ObjectRef, void> const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::unordered_map<tvm::te::Tensor, tvm::tir::Buffer, std::hash<tvm::te::Tensor>, std::equal_to<tvm::te::Tensor>, std::allocator<std::pair<tvm::te::Tensor const, tvm::tir::Buffer> > > const&, tvm::GlobalVarSupply, bool)
- at ../src/driver/driver_api.cc:388
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:283
- 10: tvm::transform::Pass::operator()(tvm::IRModule) const
- at ../src/ir/transform.cc:258
- 9: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 8: tvm::transform::SequentialNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:451
- 7: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 6: tvm::tir::transform::PrimFuncPassNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/tir/ir/transform.cc:100
- 5: tvm::runtime::TypedPackedFunc<tvm::tir::PrimFunc (tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext)>::operator()(tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext) const
- at ../include/tvm/runtime/packed_func.h:1753
- 4: tvm::tir::PrimFunc tvm::runtime::detail::typed_packed_call_dispatcher<tvm::tir::PrimFunc>::run<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::runtime::PackedFunc const&, tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&)
- at ../include/tvm/runtime/packed_func.h:1697
- 3: tvm::runtime::TVMRetValue tvm::runtime::PackedFunc::operator()<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&) const
- at ../include/tvm/runtime/packed_func.h:1621
- 2: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 1: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 0: operator()
- at ../src/runtime/c_runtime_api.cc:534
- File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
- raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 512, 1, 1]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 8, 16]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,4583049
-No: 5 GFLOPS: 10.92/10.92 result: MeasureResult(costs=(0.02120587783333333,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.8181979656219482, timestamp=1679084112.4073758) [('tile_f', [-1, 1, 2, 32]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 2, 8]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,6681370
-No: 6 GFLOPS: 0.00/10.92 result: Traceback (most recent call last):
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
- func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
- func = build(s, args, target=target, runtime=runtime)
- File "/workspace/python/tvm/driver/build_module.py", line 227, in build
- input_mod = lower(inputs, args, name=name, binds=binds)
- File "/workspace/python/tvm/driver/build_module.py", line 134, in lower
- return ffi.lower_schedule(inp, args, name, binds, simple_mode)
- File "tvm/_ffi/_cython/./packed_func.pxi", line 331, in tvm._ffi._cy3.core.PackedFuncBase.__call__
- File "tvm/_ffi/_cython/./packed_func.pxi", line 276, in tvm._ffi._cy3.core.FuncCall
- File "tvm/_ffi/_cython/./base.pxi", line 181, in tvm._ffi._cy3.core.CHECK_CALL
+ File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 706, in run_through_rpc
+ costs = time_f(*args).results
+ File "/usr/lib/python3.7/contextlib.py", line 130, in __exit__
+ self.gen.throw(type, value, traceback)
+ File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 746, in __call__
+ remote.remove(build_result.filename)
+ File "/workspace/python/tvm/rpc/client.py", line 144, in remove
+ self._remote_funcs["remove"] = self.get_function("tvm.rpc.server.remove")
+ File "/workspace/python/tvm/rpc/client.py", line 72, in get_function
+ return self._sess.get_function(name)
+ File "/workspace/python/tvm/runtime/module.py", line 171, in get_function
+ self.handle, c_str(name), ctypes.c_int(query_imports), ctypes.byref(ret_handle)
+ File "/workspace/python/tvm/_ffi/base.py", line 348, in check_call
+ raise get_last_ffi_error()
tvm._ffi.base.TVMError: Traceback (most recent call last):
- 24: TVMFuncCall
- at ../src/runtime/c_runtime_api.cc:477
- 23: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 22: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 21: operator()
- at ../include/tvm/runtime/packed_func.h:1734
- 20: unpack_call<tvm::IRModule, 5, tvm::<lambda(tvm::te::Schedule, const tvm::runtime::Array<tvm::runtime::ObjectRef>&, const tvm::runtime::String&, const tvm::runtime::Map<tvm::te::Tensor, tvm::tir::Buffer>&, bool)> >
- at ../include/tvm/runtime/packed_func.h:1674
- 19: run<>
- at ../include/tvm/runtime/packed_func.h:1634
- 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1649
- 13: operator()
- at ../src/driver/driver_api.cc:402
- 12: tvm::LowerSchedule(tvm::te::Schedule, tvm::runtime::Array<tvm::runtime::ObjectRef, void> const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::unordered_map<tvm::te::Tensor, tvm::tir::Buffer, std::hash<tvm::te::Tensor>, std::equal_to<tvm::te::Tensor>, std::allocator<std::pair<tvm::te::Tensor const, tvm::tir::Buffer> > > const&, tvm::GlobalVarSupply, bool)
- at ../src/driver/driver_api.cc:388
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:283
- 10: tvm::transform::Pass::operator()(tvm::IRModule) const
- at ../src/ir/transform.cc:258
- 9: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 8: tvm::transform::SequentialNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:451
- 7: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 6: tvm::tir::transform::PrimFuncPassNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/tir/ir/transform.cc:100
- 5: tvm::runtime::TypedPackedFunc<tvm::tir::PrimFunc (tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext)>::operator()(tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext) const
- at ../include/tvm/runtime/packed_func.h:1753
- 4: tvm::tir::PrimFunc tvm::runtime::detail::typed_packed_call_dispatcher<tvm::tir::PrimFunc>::run<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::runtime::PackedFunc const&, tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&)
- at ../include/tvm/runtime/packed_func.h:1697
- 3: tvm::runtime::TVMRetValue tvm::runtime::PackedFunc::operator()<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&) const
+ 52: 0xffffffffffffffff
+ 51: _start
+ 50: __libc_start_main
+ 49: _Py_UnixMain
+ 48: 0x0000000000650da0
+ 47: 0x0000000000650afa
+ 46: _PyFunction_FastCallDict
+ 45: _PyEval_EvalCodeWithName
+ 44: _PyEval_EvalFrameDefault
+ 43: _PyFunction_FastCallKeywords
+ 42: _PyEval_EvalCodeWithName
+ 41: _PyEval_EvalFrameDefault
+ 40: _PyMethodDef_RawFastCallKeywords
+ 39: 0x0000000000546369
+ 38: _PyEval_EvalCodeWithName
+ 37: _PyEval_EvalFrameDefault
+ 36: _PyFunction_FastCallKeywords
+ 35: _PyEval_EvalCodeWithName
+ 34: _PyEval_EvalFrameDefault
+ 33: _PyFunction_FastCallDict
+ 32: _PyEval_EvalCodeWithName
+ 31: _PyEval_EvalFrameDefault
+ 30: _PyObject_FastCallDict
+ 29: 0x00000000004c06e1
+ 28: _PyFunction_FastCallDict
+ 27: _PyEval_EvalFrameDefault
+ 26: _PyMethodDescr_FastCallKeywords
+ 25: 0x00000000005dcb58
+ 24: 0x00000000005dc83f
+ 23: 0x00000000004ba127
+ 22: _PyEval_EvalFrameDefault
+ 21: _PyFunction_FastCallKeywords
+ 20: _PyEval_EvalFrameDefault
+ 19: _PyFunction_FastCallKeywords
+ 18: _PyEval_EvalFrameDefault
+ 17: _PyFunction_FastCallKeywords
+ 16: _PyEval_EvalCodeWithName
+ 15: _PyEval_EvalFrameDefault
+ 14: 0x0000000000537c30
+ 13: _PyObject_FastCallKeywords
+ 12: 0x00007f32449a8fa2
+ 11: _ctypes_callproc
+ 10: ffi_call
+ 9: ffi_call_unix64
+ 8: TVMModGetFunction
+ at ../src/runtime/c_runtime_api.cc:408
+ 7: tvm::runtime::ModuleNode::GetFunction(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, bool)
+ at ../src/runtime/module.cc:66
+ 6: tvm::runtime::RPCModuleNode::GetFunction(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, tvm::runtime::ObjectPtr<tvm::runtime::Object> const&)
+ at ../src/runtime/rpc/rpc_module.cc:185
+ 5: tvm::runtime::RPCClientSession::GetFunction(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&)
+ at ../src/runtime/rpc/rpc_endpoint.cc:1007
+ 4: tvm::runtime::TVMRetValue tvm::runtime::RPCEndpoint::SysCallRemote<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&>(tvm::runtime::RPCCode, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&)
+ at ../src/runtime/rpc/rpc_endpoint.h:223
+ 3: tvm::runtime::TVMRetValue tvm::runtime::PackedFunc::operator()<int, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&>(int&&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&) const
at ../include/tvm/runtime/packed_func.h:1621
2: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
at ../include/tvm/runtime/packed_func.h:1217
1: Call
at ../include/tvm/runtime/packed_func.h:1213
0: operator()
- at ../src/runtime/c_runtime_api.cc:534
- File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
- raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel
+ at ../src/runtime/rpc/rpc_endpoint.cc:684
+ File "../src/runtime/rpc/rpc_endpoint.cc", line 684
+TVMError:
+---------------------------------------------------------------
+An error occurred during the execution of TVM.
+For more information, please see: https://tvm.apache.org/docs/errors.html
+---------------------------------------------------------------
+ Check failed: (code == RPCCode::kReturn) is false: code=1
Traceback (most recent call last):
- 24: TVMFuncCall
- at ../src/runtime/c_runtime_api.cc:477
- 23: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 22: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 21: operator()
- at ../include/tvm/runtime/packed_func.h:1734
- 20: unpack_call<tvm::IRModule, 5, tvm::<lambda(tvm::te::Schedule, const tvm::runtime::Array<tvm::runtime::ObjectRef>&, const tvm::runtime::String&, const tvm::runtime::Map<tvm::te::Tensor, tvm::tir::Buffer>&, bool)> >
- at ../include/tvm/runtime/packed_func.h:1674
- 19: run<>
- at ../include/tvm/runtime/packed_func.h:1634
- 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1649
- 13: operator()
- at ../src/driver/driver_api.cc:402
- 12: tvm::LowerSchedule(tvm::te::Schedule, tvm::runtime::Array<tvm::runtime::ObjectRef, void> const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::unordered_map<tvm::te::Tensor, tvm::tir::Buffer, std::hash<tvm::te::Tensor>, std::equal_to<tvm::te::Tensor>, std::allocator<std::pair<tvm::te::Tensor const, tvm::tir::Buffer> > > const&, tvm::GlobalVarSupply, bool)
- at ../src/driver/driver_api.cc:388
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:283
- 10: tvm::transform::Pass::operator()(tvm::IRModule) const
- at ../src/ir/transform.cc:258
- 9: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 8: tvm::transform::SequentialNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:451
- 7: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 6: tvm::tir::transform::PrimFuncPassNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/tir/ir/transform.cc:100
- 5: tvm::runtime::TypedPackedFunc<tvm::tir::PrimFunc (tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext)>::operator()(tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext) const
- at ../include/tvm/runtime/packed_func.h:1753
- 4: tvm::tir::PrimFunc tvm::runtime::detail::typed_packed_call_dispatcher<tvm::tir::PrimFunc>::run<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::runtime::PackedFunc const&, tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&)
- at ../include/tvm/runtime/packed_func.h:1697
- 3: tvm::runtime::TVMRetValue tvm::runtime::PackedFunc::operator()<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&) const
- at ../include/tvm/runtime/packed_func.h:1621
- 2: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 1: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 0: operator()
- at ../src/runtime/c_runtime_api.cc:534
- File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
- raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 1, 128, 1]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 64, 2]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,5090849
-No: 7 GFLOPS: 0.00/10.92 result: Traceback (most recent call last):
+ 52: 0xffffffffffffffff
+ 51: _start
+ 50: __libc_start_main
+ 49: _Py_UnixMain
+ 48: 0x0000000000650da0
+ 47: 0x0000000000650afa
+ 46: _PyFunction_FastCallDict
+ 45: _PyEval_EvalCodeWithName
+ 44: _PyEval_EvalFrameDefault
+ 43: _PyFunction_FastCallKeywords
+ 42: _PyEval_EvalCodeWithName
+ 41: _PyEval_EvalFrameDefault
+ 40: _PyMethodDef_RawFastCallKeywords
+ 39: 0x0000000000546369
+ 38: _PyEval_EvalCodeWithName
+ 37: _PyEval_EvalFrameDefault
+ 36: _PyFunction_FastCallKeywords
+ 35: _PyEval_EvalCodeWithName
+ 34: _PyEval_EvalFrameDefault
+ 33: _PyFunction_FastCallDict
+ 32: _PyEval_EvalCodeWithName
+ 31: _PyEval_EvalFrameDefault
+ 30: _PyObject_FastCallDict
+ 29: 0x00000000004c06e1
+ 28: _PyFunction_FastCallDict
+ 27: _PyEval_EvalFrameDefault
+ 26: _PyMethodDescr_FastCallKeywords
+ 25: 0x00000000005dcb58
+ 24: 0x00000000005dc83f
+ 23: 0x00000000004ba127
+ 22: _PyEval_EvalFrameDefault
+ 21: _PyFunction_FastCallKeywords
+ 20: _PyEval_EvalFrameDefault
+ 19: _PyFunction_FastCall [('tile_f', [-1, 2, 1, 32]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 1, 4]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,6071966
+No: 4 GFLOPS: 0.00/42.72 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -1311,9 +972,9 @@ Traceback (most recent call last):
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 4, 8]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 2, 32]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,9825350
-No: 8 GFLOPS: 37.60/37.60 result: MeasureResult(costs=(0.006157010647058823,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.4753825664520264, timestamp=1679084113.3095083) [('tile_f', [-1, 16, 32, 1]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 2, 8]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,5326464
-No: 9 GFLOPS: 0.00/37.60 result: Traceback (most recent call last):
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 8, 2, 32]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 2, 8]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,6487993
+No: 5 GFLOPS: 47.66/47.66 result: MeasureResult(costs=(0.004857183833333334,), error_no=MeasureErrorNo.NO_ERROR, all_cost=7.572648048400879, timestamp=1679085244.793465) [('tile_f', [-1, 8, 2, 4]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 8, 2]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,7405091
+No: 6 GFLOPS: 0.00/47.66 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -1435,8 +1096,9 @@ Traceback (most recent call last):
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 4, 4, 4]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 4, 16]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,1677617
-No: 10 GFLOPS: 0.00/37.60 result: Traceback (most recent call last):
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 1, 64, 4]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 256, 2]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,1032813
+No: 7 GFLOPS: 123.69/123.69 result: MeasureResult(costs=(0.0018716147272727275,), error_no=MeasureErrorNo.NO_ERROR, all_cost=2.975389003753662, timestamp=1679085247.3761587) [('tile_f', [-1, 2, 32, 4]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 2, 16]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,8837311
+No: 8 GFLOPS: 0.00/123.69 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -1558,8 +1220,8 @@ Traceback (most recent call last):
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 1, 256, 2]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 16, 4]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,1826759
-No: 11 GFLOPS: 0.00/37.60 result: Traceback (most recent call last):
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 8, 8, 2]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 512, 1]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,1581662
+No: 9 GFLOPS: 0.00/123.69 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -1681,8 +1343,8 @@ Traceback (most recent call last):
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 64, 8, 1]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 64, 4]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,2798653
-No: 12 GFLOPS: 0.00/37.60 result: Traceback (most recent call last):
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 16, 16]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 2, 16]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,3608623
+No: 10 GFLOPS: 0.00/123.69 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -1804,9 +1466,9 @@ Traceback (most recent call last):
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 16, 4, 2]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 16, 4]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,4341336
-No: 13 GFLOPS: 248.82/248.82 result: MeasureResult(costs=(0.0009303794031007751,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.3959543704986572, timestamp=1679084116.8684433) [('tile_f', [-1, 2, 8, 4]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 8, 1]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,206482
-No: 14 GFLOPS: 0.00/248.82 result: Traceback (most recent call last):
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 1, 4, 2]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 128, 4]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,2804192
+No: 11 GFLOPS: 10.53/123.69 result: MeasureResult(costs=(0.021979039500000002,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.6817421913146973, timestamp=1679085249.2773573) [('tile_f', [-1, 2, 4, 64]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 1, 16]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,5349288
+No: 12 GFLOPS: 0.00/123.69 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -1928,8 +1590,8 @@ Traceback (most recent call last):
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 1, 1, 512]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 8, 1]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,3303519
-No: 15 GFLOPS: 0.00/248.82 result: Traceback (most recent call last):
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 64, 1]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 32, 16]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,8464326
+No: 13 GFLOPS: 0.00/123.69 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -2051,9 +1713,10 @@ Traceback (most recent call last):
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 1, 16, 8]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 4, 64]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,5783078
-No: 16 GFLOPS: 1.32/248.82 result: MeasureResult(costs=(0.1759165495,), error_no=MeasureErrorNo.NO_ERROR, all_cost=4.974991321563721, timestamp=1679084119.72121) [('tile_f', [-1, 2, 1, 16]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 16, 1]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,3889105
-No: 17 GFLOPS: 0.00/248.82 result: Traceback (most recent call last):
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 8, 4]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 8, 32]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,347942
+No: 14 GFLOPS: 27.41/123.69 result: MeasureResult(costs=(0.008445255833333333,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.5827915668487549, timestamp=1679085251.0693889) [('tile_f', [-1, 32, 1, 1]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 2, 1]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,6008205
+No: 15 GFLOPS: 151.48/151.48 result: MeasureResult(costs=(0.001528289495238095,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.777534008026123, timestamp=1679085252.1466382) [('tile_f', [-1, 2, 16, 1]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 4, 1]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,1363375
+No: 16 GFLOPS: 0.00/151.48 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -2175,8 +1838,9 @@ Traceback (most recent call last):
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 8, 8, 2]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 4, 128]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,2891762
-No: 18 GFLOPS: 0.00/248.82 result: Traceback (most recent call last):
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 4, 8, 8]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 8, 64]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,6364756
+No: 17 GFLOPS: 30.41/151.48 result: MeasureResult(costs=(0.007613814714285715,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.7912659645080566, timestamp=1679085254.1267264) [('tile_f', [-1, 16, 1, 1]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 4, 16]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,5936704
+No: 18 GFLOPS: 0.00/151.48 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -2298,8 +1962,8 @@ Traceback (most recent call last):
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 1, 2, 256]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 1, 512]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,6194538
-No: 19 GFLOPS: 0.00/248.82 result: Traceback (most recent call last):
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 1, 8, 8]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 8, 64]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,1139314
+No: 19 GFLOPS: 0.00/151.48 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -2421,8 +2085,8 @@ Traceback (most recent call last):
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 16, 2, 2]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 8, 64]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,1718048
-No: 20 GFLOPS: 9.77/248.82 result: MeasureResult(costs=(0.023703824333333335,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.8923025131225586, timestamp=1679084121.846795) [('tile_f', [-1, 8, 1, 8]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 2, 1]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,5814739
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 4, 64]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 16, 8]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,8628608
+No: 20 GFLOPS: 70.04/151.48 result: MeasureResult(costs=(0.0033050880416666665,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.5203571319580078, timestamp=1679085255.2458742) [('tile_f', [-1, 2, 2, 8]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 16, 1]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,5243404
</pre></div>
</div>
<p>Finally we can inspect the best config from log file, check correctness,
@@ -2461,9 +2125,9 @@ and measure running time.</p>
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Finish loading 20 records
Best config:
-[('tile_f', [-1, 2, 8, 4]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 8, 1]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,206482
+[('tile_f', [-1, 2, 16, 1]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 4, 1]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,1363375
Finish loading 20 records
-Time cost of this operator: 0.001283
+Time cost of this operator: 0.001930
</pre></div>
</div>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-tune-with-autotvm-tune-conv2d-cuda-py">
diff --git a/docs/how_to/work_with_microtvm/micro_autotune.html b/docs/how_to/work_with_microtvm/micro_autotune.html
index 1ff17d6bed..a07210fdfe 100644
--- a/docs/how_to/work_with_microtvm/micro_autotune.html
+++ b/docs/how_to/work_with_microtvm/micro_autotune.html
@@ -649,10 +649,10 @@ the tuned operator.</p>
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>########## Build without Autotuning ##########
Node Name Ops Time(us) Time(%) Shape Inputs Outputs Measurements(us)
--------- --- -------- ------- ----- ------ ------- ----------------
-tvmgen_default_fused_nn_contrib_conv2d_NCHWc tvmgen_default_fused_nn_contrib_conv2d_NCHWc 315.6 98.74 (1, 2, 10, 10, 3) 2 1 [315.6]
-tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 3.06 0.957 (1, 6, 10, 10) 1 1 [3.06]
-tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.967 0.303 (1, 1, 10, 10, 3) 1 1 [0.967]
-Total_time - 319.627 - - - - -
+tvmgen_default_fused_nn_contrib_conv2d_NCHWc tvmgen_default_fused_nn_contrib_conv2d_NCHWc 316.6 98.736 (1, 2, 10, 10, 3) 2 1 [316.6]
+tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 3.063 0.955 (1, 6, 10, 10) 1 1 [3.063]
+tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.99 0.309 (1, 1, 10, 10, 3) 1 1 [0.99]
+Total_time - 320.653 - - - - -
</pre></div>
</div>
</div>
@@ -704,13 +704,13 @@ Total_time -
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>########## Build with Autotuning ##########
Node Name Ops Time(us) Time(%) Shape Inputs Outputs Measurements(us)
--------- --- -------- ------- ----- ------ ------- ----------------
-tvmgen_default_fused_nn_contrib_conv2d_NCHWc tvmgen_default_fused_nn_contrib_conv2d_NCHWc 103.0 97.522 (1, 6, 10, 10, 1) 2 1 [103.0]
-tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 1.767 1.673 (1, 6, 10, 10) 1 1 [1.767]
-tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.85 0.805 (1, 3, 10, 10, 1) 1 1 [0.85]
-Total_time - 105.617 - - - - -
+tvmgen_default_fused_nn_contrib_conv2d_NCHWc tvmgen_default_fused_nn_contrib_conv2d_NCHWc 100.7 97.351 (1, 6, 10, 10, 1) 2 1 [100.7]
+tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 1.771 1.712 (1, 6, 10, 10) 1 1 [1.771]
+tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.969 0.937 (1, 1, 10, 10, 3) 1 1 [0.969]
+Total_time - 103.44 - - - - -
</pre></div>
</div>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 22.329 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 19.674 seconds)</p>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-work-with-microtvm-micro-autotune-py">
<div class="sphx-glr-download sphx-glr-download-python docutils container">
<p><a class="reference download internal" download="" href="../../_downloads/9ccca8fd489a1486ac71b55a55c320c5/micro_autotune.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">micro_autotune.py</span></code></a></p>
diff --git a/docs/how_to/work_with_microtvm/micro_pytorch.html b/docs/how_to/work_with_microtvm/micro_pytorch.html
index b7cae6cd8a..3fddde1b8a 100644
--- a/docs/how_to/work_with_microtvm/micro_pytorch.html
+++ b/docs/how_to/work_with_microtvm/micro_pytorch.html
@@ -460,8 +460,7 @@ download a cat image and preprocess it to use as the model input.</p>
Downloading: "https://download.pytorch.org/models/quantized/mobilenet_v2_qnnpack_37f702c5.pth" to /workspace/.cache/torch/hub/checkpoints/mobilenet_v2_qnnpack_37f702c5.pth
0%| | 0.00/3.42M [00:00<?, ?B/s]
- 61%|###### | 2.09M/3.42M [00:00<00:00, 10.1MB/s]
-100%|##########| 3.42M/3.42M [00:00<00:00, 16.0MB/s]
+100%|##########| 3.42M/3.42M [00:00<00:00, 160MB/s]
/workspace/python/tvm/relay/frontend/pytorch_utils.py:47: DeprecationWarning: distutils Version classes are deprecated. Use packaging.version instead.
return LooseVersion(torch_ver) > ver
/venv/apache-tvm-py3.7/lib/python3.7/site-packages/setuptools/_distutils/version.py:346: DeprecationWarning: distutils Version classes are deprecated. Use packaging.version instead.
@@ -587,7 +586,7 @@ via the host <cite>main.cc`</cite> or if a Zephyr emulated board is selected as
Torch top-1 id: 282, class name: tiger cat
</pre></div>
</div>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 21.807 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 16.632 seconds)</p>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-work-with-microtvm-micro-pytorch-py">
<div class="sphx-glr-download sphx-glr-download-python docutils container">
<p><a class="reference download internal" download="" href="../../_downloads/12b9ecc04c41abaa12022061771821d1/micro_pytorch.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">micro_pytorch.py</span></code></a></p>
diff --git a/docs/how_to/work_with_microtvm/micro_train.html b/docs/how_to/work_with_microtvm/micro_train.html
index b789ae9f4e..4f1e1cf3c0 100644
--- a/docs/how_to/work_with_microtvm/micro_train.html
+++ b/docs/how_to/work_with_microtvm/micro_train.html
@@ -529,7 +529,7 @@ take about <strong>2 minutes</strong> to download the Stanford Cars, while COCO
<a href="https://docs.python.org/3/library/shutil.html#shutil.move" title="shutil.move" class="sphx-glr-backref-module-shutil sphx-glr-backref-type-py-function"><span class="n">shutil</span><span class="o">.</span><span class="n">move</span></a><span class="p">(</span><span class="sa">f</span><span class="s2">"</span><span class="si">{</span><a href="https://docs.python.org/3/library/stdtypes.html#str" title="builtins.str" class="sphx-glr-backref-module-builtins sphx-glr-backref-typ [...]
</pre></div>
</div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>'/tmp/tmp9sbi5qof/images/random'
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>'/tmp/tmpk8njacwx/images/random'
</pre></div>
</div>
</div>
@@ -589,8 +589,8 @@ objects to other stuff? We can display some examples from our datasets using <co
<span class="n">plt</span><span class="o">.</span><span class="n">axis</span><span class="p">(</span><span class="s2">"off"</span><span class="p">)</span>
</pre></div>
</div>
-<img src="../../_images/sphx_glr_micro_train_001.png" srcset="../../_images/sphx_glr_micro_train_001.png" alt="[0.0, 1.0], [1.0, 0.0], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0], [1.0, 0.0], [1.0, 0.0], [1.0, 0.0], [1.0, 0.0]" class = "sphx-glr-single-img"/><div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>/tmp/tmp9sbi5qof/images/target contains 8144 images
-/tmp/tmp9sbi5qof/images/random contains 5000 images
+<img src="../../_images/sphx_glr_micro_train_001.png" srcset="../../_images/sphx_glr_micro_train_001.png" alt="[0.0, 1.0], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0], [1.0, 0.0], [1.0, 0.0], [0.0, 1.0], [1.0, 0.0], [0.0, 1.0], [1.0, 0.0]" class = "sphx-glr-single-img"/><div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>/tmp/tmpk8njacwx/images/target contains 8144 images
+/tmp/tmpk8njacwx/images/random contains 5000 images
</pre></div>
</div>
</div>
@@ -702,13 +702,13 @@ the time on our validation set).</p>
</pre></div>
</div>
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Epoch 1/3
-328/328 - 48s - loss: 0.2123 - accuracy: 0.9259 - val_loss: 0.1314 - val_accuracy: 0.9607 - 48s/epoch - 146ms/step
+328/328 - 47s - loss: 0.2252 - accuracy: 0.9217 - val_loss: 0.2626 - val_accuracy: 0.8992 - 47s/epoch - 143ms/step
Epoch 2/3
-328/328 - 43s - loss: 0.0985 - accuracy: 0.9648 - val_loss: 0.1363 - val_accuracy: 0.9562 - 43s/epoch - 132ms/step
+328/328 - 43s - loss: 0.0975 - accuracy: 0.9618 - val_loss: 0.0954 - val_accuracy: 0.9622 - 43s/epoch - 132ms/step
Epoch 3/3
-328/328 - 42s - loss: 0.0669 - accuracy: 0.9745 - val_loss: 0.0953 - val_accuracy: 0.9675 - 42s/epoch - 128ms/step
+328/328 - 43s - loss: 0.0641 - accuracy: 0.9760 - val_loss: 0.1629 - val_accuracy: 0.9366 - 43s/epoch - 131ms/step
-<keras.callbacks.History object at 0x7fb002be9450>
+<keras.callbacks.History object at 0x7f3be8b52f90>
</pre></div>
</div>
</div>
@@ -972,7 +972,7 @@ as intended.</p>
<p>From here, we could modify the model to read live images from the camera - we have another
Arduino tutorial for how to do that <a class="reference external" href="https://github.com/guberti/tvm-arduino-demos/tree/master/examples/person_detection">on GitHub</a>. Alternatively, we could also
<a class="reference external" href="https://tvm.apache.org/docs/how_to/work_with_microtvm/micro_autotune.html">use TVM’s autotuning capabilities</a> to dramatically improve the model’s performance.</p>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 4 minutes 42.728 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 4 minutes 35.152 seconds)</p>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-work-with-microtvm-micro-train-py">
<div class="sphx-glr-download sphx-glr-download-python docutils container">
<p><a class="reference download internal" download="" href="../../_downloads/b52cec46baf4f78d6bcd94cbe269c8a6/micro_train.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">micro_train.py</span></code></a></p>
diff --git a/docs/how_to/work_with_microtvm/sg_execution_times.html b/docs/how_to/work_with_microtvm/sg_execution_times.html
index a6eaca67fc..b6baa2e931 100644
--- a/docs/how_to/work_with_microtvm/sg_execution_times.html
+++ b/docs/how_to/work_with_microtvm/sg_execution_times.html
@@ -345,7 +345,7 @@
<div class="section" id="computation-times">
<span id="sphx-glr-how-to-work-with-microtvm-sg-execution-times"></span><h1>Computation times<a class="headerlink" href="#computation-times" title="Permalink to this headline">¶</a></h1>
-<p><strong>07:54.442</strong> total execution time for <strong>how_to_work_with_microtvm</strong> files:</p>
+<p><strong>07:36.586</strong> total execution time for <strong>how_to_work_with_microtvm</strong> files:</p>
<table class="docutils align-default">
<colgroup>
<col style="width: 82%" />
@@ -354,34 +354,34 @@
</colgroup>
<tbody>
<tr class="row-odd"><td><p><a class="reference internal" href="micro_train.html#sphx-glr-how-to-work-with-microtvm-micro-train-py"><span class="std std-ref">5. Training Vision Models for microTVM on Arduino</span></a> (<code class="docutils literal notranslate"><span class="pre">micro_train.py</span></code>)</p></td>
-<td><p>04:42.728</p></td>
+<td><p>04:35.152</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="micro_autotune.html#sphx-glr-how-to-work-with-microtvm-micro-autotune-py"><span class="std std-ref">6. Model Tuning with microTVM</span></a> (<code class="docutils literal notranslate"><span class="pre">micro_autotune.py</span></code>)</p></td>
-<td><p>01:22.329</p></td>
+<td><p>01:19.674</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="micro_pytorch.html#sphx-glr-how-to-work-with-microtvm-micro-pytorch-py"><span class="std std-ref">4. microTVM PyTorch Tutorial</span></a> (<code class="docutils literal notranslate"><span class="pre">micro_pytorch.py</span></code>)</p></td>
-<td><p>01:21.807</p></td>
+<td><p>01:16.632</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="micro_aot.html#sphx-glr-how-to-work-with-microtvm-micro-aot-py"><span class="std std-ref">3. microTVM Ahead-of-Time (AOT) Compilation</span></a> (<code class="docutils literal notranslate"><span class="pre">micro_aot.py</span></code>)</p></td>
-<td><p>00:10.649</p></td>
+<td><p>00:09.934</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="micro_custom_ide.html#sphx-glr-how-to-work-with-microtvm-micro-custom-ide-py"><span class="std std-ref">9. Bring microTVM to your own development environment</span></a> (<code class="docutils literal notranslate"><span class="pre">micro_custom_ide.py</span></code>)</p></td>
-<td><p>00:09.153</p></td>
+<td><p>00:08.059</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="micro_tflite.html#sphx-glr-how-to-work-with-microtvm-micro-tflite-py"><span class="std std-ref">2. microTVM TFLite Tutorial</span></a> (<code class="docutils literal notranslate"><span class="pre">micro_tflite.py</span></code>)</p></td>
-<td><p>00:07.776</p></td>
+<td><p>00:07.135</p></td>
... 3400 lines suppressed ...