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 2023/04/16 21:01:41 UTC
[tvm-site] branch asf-site updated: deploying docs (apache/tvm@0d51fbbecde0cefc4da8ef5c171c2fbd22655e30)
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 b6883ceae9 deploying docs (apache/tvm@0d51fbbecde0cefc4da8ef5c171c2fbd22655e30)
b6883ceae9 is described below
commit b6883ceae97d6f0650f847cebc5a04f93630369a
Author: tvm-bot <95...@users.noreply.github.com>
AuthorDate: Sun Apr 16 21:01:34 2023 +0000
deploying docs (apache/tvm@0d51fbbecde0cefc4da8ef5c171c2fbd22655e30)
---
docs/_images/sphx_glr_micro_train_001.png | Bin 329926 -> 336918 bytes
docs/_images/sphx_glr_micro_train_thumb.png | Bin 23812 -> 24568 bytes
.../how_to/compile_models/from_darknet.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 | 2853 +++++---------------
.../tune_network_cuda.rst.txt | 4 +-
.../tune_network_x86.rst.txt | 4 +-
.../tune_sparse_x86.rst.txt | 55 +-
.../tune_with_autotvm/sg_execution_times.rst.txt | 10 +-
.../tune_with_autotvm/tune_conv2d_cuda.rst.txt | 560 +---
.../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 | 14 +-
.../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 | 16 +-
.../tutorials/autotvm/sg_execution_times.rst.txt | 4 +-
.../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 | 6 +-
.../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 | 59 +-
.../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 | 51 +-
docs/commit_hash | 2 +-
docs/how_to/compile_models/from_darknet.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 | 13 +-
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 | 39 +-
.../deploy_models/deploy_model_on_android.html | 2 +-
.../deploy_object_detection_pytorch.html | 51 +-
docs/how_to/deploy_models/deploy_prequantized.html | 10 +-
.../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 | 38 +-
docs/how_to/deploy_models/sg_execution_times.html | 22 +-
.../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 | 2853 +++++---------------
.../tune_with_autoscheduler/tune_network_cuda.html | 4 +-
.../tune_with_autoscheduler/tune_network_x86.html | 4 +-
.../tune_with_autoscheduler/tune_sparse_x86.html | 55 +-
.../tune_with_autotvm/sg_execution_times.html | 10 +-
.../how_to/tune_with_autotvm/tune_conv2d_cuda.html | 560 +---
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 | 14 +-
.../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 | 16 +-
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 | 4 +-
.../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 | 6 +-
docs/tutorial/auto_scheduler_matmul_x86.html | 4 +-
docs/tutorial/autotvm_matmul_x86.html | 20 +-
docs/tutorial/autotvm_relay_x86.html | 273 +-
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 | 47 +-
129 files changed, 2340 insertions(+), 6347 deletions(-)
diff --git a/docs/_images/sphx_glr_micro_train_001.png b/docs/_images/sphx_glr_micro_train_001.png
index cf6a64d19f..a7d0b269c0 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 5378603442..9c19979ef3 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 5487f678c9..be70175056 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 19.028 seconds)
+ **Total running time of the script:** ( 1 minutes 20.948 seconds)
.. _sphx_glr_download_how_to_compile_models_from_darknet.py:
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 821241f386..4c87438585 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.zip14f4d7c9-c966-4257-bf78-b7900fd652e8 from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/resnet18_v1-a0666292.zip...
+ Downloading /workspace/.mxnet/models/resnet18_v1-a0666292.zipfe19e3f0-490e-4e8c-b4a5-cffbb6c3277d 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 b02f286fde..5af13cddbd 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, 58.8MB/s]
41%|#### | 16.9M/41.5M [00:00<00:00, 76.0MB/s]
59%|#####8 | 24.4M/41.5M [00:00<00:00, 75.8MB/s]
77%|#######7 | 32.0M/41.5M [00:00<00:00, 71.2MB/s]
94%|#########3| 38.9M/41.5M [00:00<00:00, 62.1MB/s]
100%|##########| 41.5M/41.5M [00:00<00:00, 61.4MB/s]
+
0%| | 0.00/41.5M [00:00<?, ?B/s]
19%|#9 | 7.99M/41.5M [00:00<00:00, 47.9MB/s]
39%|###8 | 16.0M/41.5M [00:00<00:00, 49.3MB/s]
62%|######2 | 25.8M/41.5M [00:00<00:00, 66.9MB/s]
88%|########7 | 36.3M/41.5M [00:00<00:00, 80.8MB/s]
100%|##########| 41.5M/41.5M [00:00<00:00, 65.0MB/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 c928cb351c..1ee86bbc2a 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%|#7 | 7.99M/44.7M [00:00<00:00, 73.4MB/s]
34%|###3 | 15.0M/44.7M [00:00<00:00, 57.3MB/s]
49%|####8 | 21.8M/44.7M [00:00<00:00, 62.7MB/s]
63%|######2 | 28.0M/44.7M [00:00<00:00, 56.1MB/s]
86%|########5 | 38.3M/44.7M [00:00<00:00, 48.8MB/s]
99%|#########8| 44.2M/44.7M [00:00<00:00, 50.4MB/s]
100%|##########| 44.7M/44.7M [00:00<00:00, 53.7MB/s]
+
0%| | 0.00/44.7M [00:00<?, ?B/s]
18%|#7 | 7.99M/44.7M [00:00<00:00, 77.3MB/s]
36%|###5 | 16.0M/44.7M [00:00<00:00, 80.4MB/s]
58%|#####8 | 26.1M/44.7M [00:00<00:00, 83.8MB/s]
76%|#######6 | 34.0M/44.7M [00:00<00:00, 83.1MB/s]
96%|#########6| 43.0M/44.7M [00:00<00:00, 86.8MB/s]
100%|##########| 44.7M/44.7M [00:00<00:00, 87.0MB/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 fe655a49e0..4c59446a28 100644
--- a/docs/_sources/how_to/compile_models/from_tensorflow.rst.txt
+++ b/docs/_sources/how_to/compile_models/from_tensorflow.rst.txt
@@ -430,7 +430,7 @@ Run the corresponding model on tensorflow
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 32.584 seconds)
+ **Total running time of the script:** ( 1 minutes 32.337 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 d361c456c9..847408ac5f 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
=================
-**06:46.135** total execution time for **how_to_compile_models** files:
+**06:57.086** total execution time for **how_to_compile_models** files:
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_tensorflow.py` (``from_tensorflow.py``) | 01:32.584 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_tensorflow.py` (``from_tensorflow.py``) | 01:32.337 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_darknet.py` (``from_darknet.py``) | 01:19.028 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_darknet.py` (``from_darknet.py``) | 01:20.948 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_paddle.py` (``from_paddle.py``) | 00:56.006 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_paddle.py` (``from_paddle.py``) | 00:58.825 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_oneflow.py` (``from_oneflow.py``) | 00:37.565 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_oneflow.py` (``from_oneflow.py``) | 00:38.579 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_coreml.py` (``from_coreml.py``) | 00:31.604 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_coreml.py` (``from_coreml.py``) | 00:33.235 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_mxnet.py` (``from_mxnet.py``) | 00:29.816 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_mxnet.py` (``from_mxnet.py``) | 00:30.773 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_tflite.py` (``from_tflite.py``) | 00:28.016 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_tflite.py` (``from_tflite.py``) | 00:28.641 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_pytorch.py` (``from_pytorch.py``) | 00:26.519 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_pytorch.py` (``from_pytorch.py``) | 00:26.838 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_keras.py` (``from_keras.py``) | 00:22.363 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_keras.py` (``from_keras.py``) | 00:24.163 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_onnx.py` (``from_onnx.py``) | 00:02.633 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_onnx.py` (``from_onnx.py``) | 00:02.747 | 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 3b08d45d0c..6c7e37d9de 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
@@ -673,7 +673,7 @@ 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)
- 2639.3694 2638.2291 2644.0245 2636.1597 2.7886
+ 2679.1327 2678.2908 2682.9453 2676.3431 2.4263
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 05fdf66d4e..6d010476ee 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
7249920/102967424 [=>............................] - ETA: 0s
10575872/102967424 [==>...........................] - ETA: 1s
13828096/102967424 [===>..........................] - ETA: 1s
16482304/102967424 [===>..........................] - ETA: 1s
25616384/102967424 [======>.......................] - ETA: 1s
35422208/102967424 [=========>....................] - ETA: 0s
46432256/102967424 [============>.................] - ETA: 0s
52477952/102967424 [==============>...............] - ETA: 0s
58228736/102967424 [===============>..............] - ETA: 0s
63225856/102967424 [=================>............] - ETA: 0s
68239360/102967424 [==================>...........] - ETA: 0s
73261056/102967424 [====================>.........] - ETA: 0s
78217216/102967424 [=====================>........] - ETA: 0s
83255296/102967424 [=======================>......] - ETA: 0s
84746240/102967424 [=======================>......] -
ETA: 0s
92995584/102967424 [==========================>...] - ETA: 0s
98312192/102967424 [===========================>..] - ETA: 0s
102850560/102967424 [============================>.] - ETA: 0s
102967424/102967424 [==============================] - 1s 0us/step
+
8192/102967424 [..............................] - ETA: 0s
8380416/102967424 [=>............................] - ETA: 1s
15024128/102967424 [===>..........................] - ETA: 1s
16769024/102967424 [===>..........................] - ETA: 1s
23412736/102967424 [=====>........................] - ETA: 1s
25157632/102967424 [======>.......................] - ETA: 1s
33546240/102967424 [========>.....................] - ETA: 1s
40189952/102967424 [==========>...................] - ETA: 1s
41934848/102967424 [===========>..................] - ETA: 1s
50323456/102967424 [=============>................] - ETA: 0s
58712064/102967424 [================>.............] - ETA: 0s
65355776/102967424 [==================>...........] - ETA: 0s
67100672/102967424 [==================>...........] - ETA: 0s
69296128/102967424 [===================>..........] - ETA: 0s
75489280/102967424 [====================>.........] - ETA: 0s
82124800/102967424 [======================>.......] -
ETA: 0s
83877888/102967424 [=======================>......] - ETA: 0s
90521600/102967424 [=========================>....] - ETA: 0s
92266496/102967424 [=========================>....] - ETA: 0s
94191616/102967424 [==========================>...] - ETA: 0s
100646912/102967424 [============================>.] - ETA: 0s
102850560/102967424 [============================>.] - ETA: 0s
102967424/102967424 [==============================] - 2s 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 7ea64f9f80..a3702bdf19 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)
- 14.5475 14.4685 15.0415 14.4153 0.1812
+ 16.1558 15.9476 17.0428 15.8976 0.4230
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 8995f10bab..3a0afb4766 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]
5%|4 | 7.99M/170M [00:00<00:03, 52.9MB/s]
8%|8 | 14.3M/170M [00:00<00:02, 55.5MB/s]
12%|#1 | 19.6M/170M [00:00<00:03, 46.3MB/s]
14%|#4 | 24.1M/170M [00:00<00:03, 46.3MB/s]
19%|#8 | 32.0M/170M [00:00<00:03, 47.5MB/s]
24%|##3 | 40.0M/170M [00:00<00:02, 53.7MB/s]
28%|##8 | 48.0M/170M [00:00<00:02, 55.1MB/s]
33%|###2 | 56.0M/170M [00:01<00:02, 57.5MB/s]
37%|###6 | 62.3M/170M [00:01<00:02, 49.1MB/s]
40%|###9 | 67.2M/170M [00:01<00:02, 49.4MB/s]
42%|####2 | 72.0M/170M [00:01<00:02, 48.7MB/s]
47%|####7 | 80.0M/170M [00:01<00:02, 46.7MB/s]
50%|####9 | 84.5M/170M [00:01<00:02, 44.5MB/s]
52%|#####2 | 88.8M/170M [00:01<00:02, 40.2MB/s]
57%|#####6 | 96.0M/170M [00:02<00:01, 45.9MB/s]
61%|######1 | 104M/170M [00:02<00:01, 44.0MB/s]
66%|######6 | 112M/170M [00:02<00:01, 53.3MB/s
]
70%|######9 | 118M/170M [00:02<00:01, 53.7MB/s]
74%|#######3 | 125M/170M [00:02<00:00, 58.0MB/s]
77%|#######7 | 131M/170M [00:02<00:00, 47.9MB/s]
80%|######## | 136M/170M [00:02<00:00, 42.2MB/s]
85%|########5 | 145M/170M [00:03<00:00, 54.0MB/s]
89%|########9 | 152M/170M [00:03<00:00, 48.4MB/s]
93%|#########3| 158M/170M [00:03<00:00, 50.8MB/s]
96%|#########6| 164M/170M [00:03<00:00, 48.2MB/s]
99%|#########9| 169M/170M [00:03<00:00, 48.3MB/s]
100%|##########| 170M/170M [00:03<00:00, 49.5MB/s]
+
0%| | 0.00/170M [00:00<?, ?B/s]
5%|4 | 7.99M/170M [00:00<00:03, 48.5MB/s]
9%|9 | 16.0M/170M [00:00<00:03, 45.9MB/s]
14%|#4 | 24.0M/170M [00:00<00:03, 47.0MB/s]
19%|#8 | 32.0M/170M [00:00<00:02, 53.4MB/s]
24%|##3 | 40.0M/170M [00:00<00:02, 60.1MB/s]
30%|##9 | 50.6M/170M [00:00<00:01, 73.9MB/s]
38%|###7 | 64.0M/170M [00:00<00:01, 92.3MB/s]
43%|####3 | 73.5M/170M [00:01<00:01, 73.4MB/s]
48%|####7 | 81.4M/170M [00:01<00:01, 73.1MB/s]
52%|#####2 | 89.0M/170M [00:01<00:01, 64.4MB/s]
56%|#####6 | 95.8M/170M [00:01<00:01, 62.8MB/s]
60%|###### | 102M/170M [00:01<00:01, 48.4MB/s]
63%|######3 | 108M/170M [00:01<00:01, 46.4MB/s]
67%|######6 | 113M/170M [00:02<00:01, 48.7MB/s]
71%|####### | 120M/170M [00:02<00:00, 53.9MB/s]
74%|#######4 | 126M/170M [00:02<00:00, 56.7MB/s]
79%|#######9 | 134M/170M [00:02<00:00, 63.7MB/s]
85%|########4 | 144M/170M [00:02<00:00, 62.5MB/s]
89%|########9 | 152M/170M [00:02<00:00, 67.1MB/s]
94%|#########4| 160M/170M [00:02<00:00, 62.3MB/s]
98%|#########7| 166M/170M [00:02<00:00, 59.8MB/s]
100%|##########| 170M/170M [00:02<00:00, 60.1MB/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 24.688 seconds)
+ **Total running time of the script:** ( 3 minutes 38.754 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 e0075c71bb..59f33aa413 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]
47%|####6 | 6.30M/13.6M [00:00<00:00, 40.3MB/s]
75%|#######4 | 10.2M/13.6M [00:00<00:00, 32.6MB/s]
98%|#########7| 13.2M/13.6M [00:00<00:00, 31.6MB/s]
100%|##########| 13.6M/13.6M [00:00<00:00, 33.5MB/s]
+
0%| | 0.00/13.6M [00:00<?, ?B/s]
59%|#####8 | 7.99M/13.6M [00:00<00:00, 54.8MB/s]
100%|##########| 13.6M/13.6M [00:00<00:00, 79.6MB/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)
- 88.9007 88.8215 92.7114 88.4916 0.4311
+ 90.1986 90.1078 94.0665 89.9954 0.4309
@@ -458,7 +458,7 @@ TODO
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 15.619 seconds)
+ **Total running time of the script:** ( 1 minutes 19.336 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 8a5bf031c3..63ca58bcb2 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)
- 114.4671 114.2424 119.0063 113.4299 0.8437
+ 119.8069 119.4192 155.3561 118.6067 3.6082
@@ -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 34.480 seconds)
+ **Total running time of the script:** ( 2 minutes 36.460 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 b398c3d012..8d5e890748 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 39.726 seconds)
+ **Total running time of the script:** ( 1 minutes 38.680 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 4bc51a55b1..c0e1293089 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 | 4959/132723 [00:00<00:02, 49583.07KB/s]
10%|9 | 13134/132723 [00:00<00:01, 68502.69KB/s]
16%|#6 | 21402/132723 [00:00<00:01, 74975.06KB/s]
22%|##2 | 29630/132723 [00:00<00:01, 77855.61KB/s]
29%|##8 | 37962/132723 [00:00<00:01, 79825.08KB/s]
35%|###4 | 46265/132723 [00:00<00:01, 80913.76KB/s]
41%|####1 | 54510/132723 [00:00<00:00, 81413.65KB/s]
47%|####7 | 62777/132723 [00:00<00:00, 81811.70KB/s]
54%|#####3 | 71113/132723 [00:00<00:00, 82294.62KB/s]
60%|#####9 | 79443/132723 [00:01<00:00, 82604.16KB/s]
66%|######6 | 87771/132723 [00:01<00:00, 82810.17KB/s]
72%|#######2 | 96053/132723 [00:01<00:00, 82780.65KB/s]
79%|#######8 | 104332/132723 [00:01<00:00, 79493.09KB/s]
85%|########4 | 112309/132723 [00:01<00:00, 70754.81KB/s]
90%|######### | 119568/132723 [00:01<00:00, 51841.35KB/s]
96%|########
#6| 127808/132723 [00:01<00:00, 58567.59KB/s]
100%|##########| 132723/132723 [00:01<00:00, 71581.58KB/s]
+
0%| | 0/132723 [00:00<?, ?KB/s]
4%|3 | 5216/132723 [00:00<00:02, 52110.85KB/s]
9%|9 | 12469/132723 [00:00<00:01, 64115.06KB/s]
15%|#4 | 19783/132723 [00:00<00:01, 68231.77KB/s]
20%|## | 27099/132723 [00:00<00:01, 70174.57KB/s]
26%|##5 | 34365/132723 [00:00<00:01, 71068.11KB/s]
31%|###1 | 41777/132723 [00:00<00:01, 72093.28KB/s]
37%|###6 | 49038/132723 [00:00<00:01, 72259.21KB/s]
42%|####2 | 56372/132723 [00:00<00:01, 72601.32KB/s]
48%|####8 | 63712/132723 [00:00<00:00, 72734.40KB/s]
54%|#####3 | 71159/132723 [00:01<00:00, 73268.73KB/s]
59%|#####9 | 78629/132723 [00:01<00:00, 73701.27KB/s]
65%|######4 | 86152/132723 [00:01<00:00, 74016.32KB/s]
71%|####### | 93749/132723 [00:01<00:00, 74603.37KB/s]
76%|#######6 | 101347/132723 [00:01<00:00, 75017.03KB/s]
82%|########2 | 108849/132723 [00:01<00:00, 74753.18KB/s]
88%|########7
| 116480/132723 [00:01<00:00, 75218.67KB/s]
93%|#########3| 124003/132723 [00:01<00:00, 74792.79KB/s]
99%|#########9| 131998/132723 [00:01<00:00, 76330.82KB/s]
100%|##########| 132723/132723 [00:01<00:00, 73109.71KB/s]
@@ -246,7 +246,7 @@ Display result
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 3 minutes 52.236 seconds)
+ **Total running time of the script:** ( 4 minutes 2.527 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 391bf053c0..07d78db05b 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
=================
-**16:10.207** total execution time for **how_to_deploy_models** files:
+**16:48.689** total execution time for **how_to_deploy_models** files:
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_ssd_gluoncv.py` (``deploy_ssd_gluoncv.py``) | 03:52.236 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_ssd_gluoncv.py` (``deploy_ssd_gluoncv.py``) | 04:02.527 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_object_detection_pytorch.py` (``deploy_object_detection_pytorch.py``) | 03:24.688 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_object_detection_pytorch.py` (``deploy_object_detection_pytorch.py``) | 03:38.754 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_prequantized_tflite.py` (``deploy_prequantized_tflite.py``) | 02:34.480 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_prequantized_tflite.py` (``deploy_prequantized_tflite.py``) | 02:36.460 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_quantized.py` (``deploy_quantized.py``) | 01:39.726 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_quantized.py` (``deploy_quantized.py``) | 01:38.680 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_prequantized.py` (``deploy_prequantized.py``) | 01:15.619 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_prequantized.py` (``deploy_prequantized.py``) | 01:19.336 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_adreno.py` (``deploy_model_on_adreno.py``) | 00:55.581 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_adreno.py` (``deploy_model_on_adreno.py``) | 00:57.116 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_adreno_tvmc.py` (``deploy_model_on_adreno_tvmc.py``) | 00:50.240 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_adreno_tvmc.py` (``deploy_model_on_adreno_tvmc.py``) | 00:53.472 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_android.py` (``deploy_model_on_android.py``) | 00:41.837 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_android.py` (``deploy_model_on_android.py``) | 00:43.972 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_nano.py` (``deploy_model_on_nano.py``) | 00:28.124 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_nano.py` (``deploy_model_on_nano.py``) | 00:29.303 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_rasp.py` (``deploy_model_on_rasp.py``) | 00:27.672 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_rasp.py` (``deploy_model_on_rasp.py``) | 00:29.063 | 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 7e49c30f64..216530d76f 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.zipb9b51972-6301-4f46-b7b0-f27570bb132d from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/mobilenet0.25-9f83e440.zip...
+ Downloading /workspace/.mxnet/models/mobilenet0.25-9f83e440.zip0953cac7-8696-4cb7-8883-8dfc06a203a6 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 afb2661817..7b39d7d92c 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
=================
-**00:52.436** total execution time for **how_to_extend_tvm** files:
+**00:55.939** 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``) | 00:48.822 | 0.0 MB |
+| :ref:`sphx_glr_how_to_extend_tvm_bring_your_own_datatypes.py` (``bring_your_own_datatypes.py``) | 00:52.041 | 0.0 MB |
+-------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_extend_tvm_use_pass_instrument.py` (``use_pass_instrument.py``) | 00:02.596 | 0.0 MB |
+| :ref:`sphx_glr_how_to_extend_tvm_use_pass_instrument.py` (``use_pass_instrument.py``) | 00:02.794 | 0.0 MB |
+-------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_extend_tvm_use_pass_infra.py` (``use_pass_infra.py``) | 00:01.011 | 0.0 MB |
+| :ref:`sphx_glr_how_to_extend_tvm_use_pass_infra.py` (``use_pass_infra.py``) | 00:01.097 | 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 |
+| :ref:`sphx_glr_how_to_extend_tvm_low_level_custom_pass.py` (``low_level_custom_pass.py``) | 00:00.008 | 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 62dae68ee9..27ffa6a529 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: 22011us [22011us] (48.70%; 48.70%)
- FoldScaleAxis: 23182us [6us] (51.30%; 51.30%)
- FoldConstant: 23176us [1688us] (51.28%; 99.97%)
- InferType: 21488us [21488us] (47.55%; 92.72%)
+ InferType: 22554us [22554us] (47.83%; 47.83%)
+ FoldScaleAxis: 24604us [7us] (52.17%; 52.17%)
+ FoldConstant: 24597us [1674us] (52.16%; 99.97%)
+ InferType: 22923us [22923us] (48.61%; 93.19%)
@@ -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: 21512us [21512us] (48.59%; 48.59%)
- FoldScaleAxis: 22759us [4us] (51.41%; 51.41%)
- FoldConstant: 22754us [1566us] (51.40%; 99.98%)
- InferType: 21188us [21188us] (47.86%; 93.12%)
+ InferType: 22398us [22398us] (48.48%; 48.48%)
+ FoldScaleAxis: 23806us [6us] (51.52%; 51.52%)
+ FoldConstant: 23800us [1770us] (51.51%; 99.97%)
+ InferType: 22031us [22031us] (47.68%; 92.56%)
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 a6ca0b1412..efd11bac6f 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: 53.520542 ms
+ Convolution: 53.530464 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 99228f7cdb..7417a4e731 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: 11.141734 ms
+ conv2d with tensor core: 12.245008 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 4c6f0abc87..2d8d083b72 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: 0.015934
- Baseline: 3.237478
+ Numpy running time: 0.018680
+ Baseline: 3.376687
@@ -227,7 +227,7 @@ fill 32 * 32 * sizeof(float) which is 4KB in the cache whose total size is 32KB
.. code-block:: none
- Opt1: 0.288589
+ Opt1: 0.304087
@@ -318,7 +318,7 @@ In this tutorial, we chose to vectorize the inner loop row data since it is cach
.. code-block:: none
- Opt2: 0.323071
+ Opt2: 0.343316
@@ -406,7 +406,7 @@ the access pattern for A matrix is more cache friendly.
.. code-block:: none
- Opt3: 0.111656
+ Opt3: 0.124391
@@ -523,7 +523,7 @@ flattening.
.. code-block:: none
- Opt4: 0.107130
+ Opt4: 0.110405
@@ -635,7 +635,7 @@ write to C when all the block results are ready.
.. code-block:: none
- Opt5: 0.100487
+ Opt5: 0.111236
@@ -748,7 +748,7 @@ Furthermore, we can also utilize multi-core processors to do the thread-level pa
.. code-block:: none
- Opt6: 0.133279
+ Opt6: 0.147561
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 4d9e020bd8..045da5c624 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
=================
-**00:33.258** total execution time for **how_to_optimize_operators** files:
+**00:35.503** total execution time for **how_to_optimize_operators** files:
+-----------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_optimize_operators_opt_gemm.py` (``opt_gemm.py``) | 00:30.380 | 0.0 MB |
+| :ref:`sphx_glr_how_to_optimize_operators_opt_gemm.py` (``opt_gemm.py``) | 00:32.496 | 0.0 MB |
+-----------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_optimize_operators_opt_conv_tensorcore.py` (``opt_conv_tensorcore.py``) | 00:01.800 | 0.0 MB |
+| :ref:`sphx_glr_how_to_optimize_operators_opt_conv_tensorcore.py` (``opt_conv_tensorcore.py``) | 00:01.864 | 0.0 MB |
+-----------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_optimize_operators_opt_conv_cuda.py` (``opt_conv_cuda.py``) | 00:01.077 | 0.0 MB |
+| :ref:`sphx_glr_how_to_optimize_operators_opt_conv_cuda.py` (``opt_conv_cuda.py``) | 00:01.143 | 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 3ea5771180..a2b444cc77 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
=================
-**10:18.423** total execution time for **how_to_tune_with_autoscheduler** files:
+**10:29.512** 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``) | 06:30.519 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_conv2d_layer_cuda.py` (``tune_conv2d_layer_cuda.py``) | 06:17.263 | 0.0 MB |
+----------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_x86.py` (``tune_network_x86.py``) | 01:40.103 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_x86.py` (``tune_network_x86.py``) | 01:45.435 | 0.0 MB |
+----------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_cuda.py` (``tune_network_cuda.py``) | 01:09.444 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_cuda.py` (``tune_network_cuda.py``) | 01:12.706 | 0.0 MB |
+----------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_sparse_x86.py` (``tune_sparse_x86.py``) | 00:31.296 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_sparse_x86.py` (``tune_sparse_x86.py``) | 00:45.695 | 0.0 MB |
+----------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_arm.py` (``tune_network_arm.py``) | 00:13.816 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_arm.py` (``tune_network_arm.py``) | 00:14.585 | 0.0 MB |
+----------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_mali.py` (``tune_network_mali.py``) | 00:13.244 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_mali.py` (``tune_network_mali.py``) | 00:13.829 | 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 3f0adfd3a6..9fafff7eda 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
@@ -243,12 +243,12 @@ 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": T.bool(True), "global_symbol": "main", "tir.noalias": T.bool(True)})
- blockIdx_x = T.launch_thread("blockIdx.x", 32)
- conv2d_nchw = T.allocate([7], "float32", "local")
- pad_temp_shared = T.allocate([6272], "float32", "shared")
- kernel_shared = T.allocate([2048], "float32", "shared")
- threadIdx_x = T.launch_thread("threadIdx.x", 112)
- conv2d_nchw_1 = T.Buffer((1,), data=conv2d_nchw, scope="local", align=4)
+ blockIdx_x = T.launch_thread("blockIdx.x", 16)
+ conv2d_nchw = T.allocate([28], "float32", "local")
+ pad_temp_shared = T.allocate([1008], "float32", "shared")
+ kernel_shared = T.allocate([1536], "float32", "shared")
+ threadIdx_x = T.launch_thread("threadIdx.x", 56)
+ conv2d_nchw_1 = T.Buffer((28,), data=conv2d_nchw, scope="local")
conv2d_nchw_1[0] = T.float32(0)
conv2d_nchw_1[1] = T.float32(0)
conv2d_nchw_1[2] = T.float32(0)
@@ -256,1094 +256,360 @@ cooperative fetching, unrolling and operator fusion.
conv2d_nchw_1[4] = T.float32(0)
conv2d_nchw_1[5] = T.float32(0)
conv2d_nchw_1[6] = T.float32(0)
- for rc_outer_outer, ry_outer_outer, rx_outer_outer in T.grid(4, 3, 3):
+ conv2d_nchw_1[7] = T.float32(0)
+ conv2d_nchw_1[8] = T.float32(0)
+ conv2d_nchw_1[9] = T.float32(0)
+ conv2d_nchw_1[10] = T.float32(0)
+ conv2d_nchw_1[11] = T.float32(0)
+ conv2d_nchw_1[12] = T.float32(0)
+ conv2d_nchw_1[13] = T.float32(0)
+ conv2d_nchw_1[14] = T.float32(0)
+ conv2d_nchw_1[15] = T.float32(0)
+ conv2d_nchw_1[16] = T.float32(0)
+ conv2d_nchw_1[17] = T.float32(0)
+ conv2d_nchw_1[18] = T.float32(0)
+ conv2d_nchw_1[19] = T.float32(0)
+ conv2d_nchw_1[20] = T.float32(0)
+ conv2d_nchw_1[21] = T.float32(0)
+ conv2d_nchw_1[22] = T.float32(0)
+ conv2d_nchw_1[23] = T.float32(0)
+ conv2d_nchw_1[24] = T.float32(0)
+ conv2d_nchw_1[25] = T.float32(0)
+ conv2d_nchw_1[26] = T.float32(0)
+ conv2d_nchw_1[27] = T.float32(0)
+ for rc_outer_outer, ry_outer_outer in T.grid(32, 3):
+ cse_var_4: T.int32 = rc_outer_outer * 784
+ cse_var_3: T.int32 = ry_outer_outer * 7
+ cse_var_2: T.int32 = rc_outer_outer * 144
+ cse_var_1: T.int32 = ry_outer_outer * 3
threadIdx_x_1 = T.env_thread("threadIdx.x")
- pad_temp_shared_1 = T.Buffer((6272,), data=pad_temp_shared, scope="shared")
+ pad_temp_shared_1 = T.Buffer((1008,), data=pad_temp_shared, scope="shared")
data_1 = T.Buffer((25088,), data=data.data)
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1] = T.if_then_else(1 <= threadIdx_x_1 % 49 // 7 + ry_outer_outer and threadIdx_x_1 % 49 // 7 + ry_outer_outer < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer - 8], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 112] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 104], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 224] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 216], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 336] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 328], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 448] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 440], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 560] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 552], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 672] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 664], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 784] = T.if_then_else(1 <= threadIdx_x_1 % 49 // 7 + ry_outer_outer and threadIdx_x_1 % 49 // 7 + ry_outer_outer < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 776], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 896] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 888], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 1008] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 1000], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 1120] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 1112], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 1232] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 1224], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 1344] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 1336], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 1456] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 1448], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 1568] = T.if_then_else(1 <= threadIdx_x_1 % 49 // 7 + ry_outer_outer and threadIdx_x_1 % 49 // 7 + ry_outer_outer < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 1560], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 1680] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 1672], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 1792] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 1784], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 1904] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 1896], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 2016] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 2008], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 2128] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 2120], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 2240] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 2232], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 2352] = T.if_then_else(1 <= threadIdx_x_1 % 49 // 7 + ry_outer_outer and threadIdx_x_1 % 49 // 7 + ry_outer_outer < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 2344], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 2464] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 2456], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 2576] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 2568], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 2688] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 2680], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 2800] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 2792], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 2912] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 2904], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 3024] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 3016], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 3136] = T.if_then_else(1 <= threadIdx_x_1 % 49 // 7 + ry_outer_outer and threadIdx_x_1 % 49 // 7 + ry_outer_outer < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 3128], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 3248] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 3240], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 3360] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 3352], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 3472] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 3464], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 3584] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 3576], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 3696] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 3688], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 3808] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 3800], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 3920] = T.if_then_else(1 <= threadIdx_x_1 % 49 // 7 + ry_outer_outer and threadIdx_x_1 % 49 // 7 + ry_outer_outer < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 3912], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 4032] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 4024], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 4144] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 4136], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 4256] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 4248], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 4368] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 4360], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 4480] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 4472], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 4592] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 4584], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 4704] = T.if_then_else(1 <= threadIdx_x_1 % 49 // 7 + ry_outer_outer and threadIdx_x_1 % 49 // 7 + ry_outer_outer < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 4696], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 4816] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 4808], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 4928] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 4920], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 5040] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 5032], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 5152] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 5144], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 5264] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 5256], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 5376] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 5368], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 5488] = T.if_then_else(1 <= threadIdx_x_1 % 49 // 7 + ry_outer_outer and threadIdx_x_1 % 49 // 7 + ry_outer_outer < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 5480], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 5600] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 5592], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 5712] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 5704], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 5824] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 5816], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 5936] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 5928], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 6048] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 6040], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 6160] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 6152], T.float32(0))
- kernel_shared_1 = T.Buffer((2048,), data=kernel_shared, scope="shared")
- with T.launch_thread("threadIdx.x", 112) as threadIdx_x_2:
- kernel_1 = T.Buffer((2359296,), data=kernel.data)
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 1] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 9]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 2] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 18]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 3] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 27]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 4] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 36]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 5] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 45]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 6] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 54]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 7] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 63]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 8] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 72]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 9] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 81]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 10] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 90]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 11] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 99]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 12] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 108]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 13] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 117]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 14] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 126]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 15] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 135]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 16] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 144]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 17] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 153]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 18] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 162]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 19] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 171]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 20] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 180]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 21] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 189]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 22] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 198]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 23] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 207]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 24] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 216]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 25] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 225]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 26] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 234]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 27] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 243]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 28] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 252]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 29] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 261]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 30] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 270]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 31] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 279]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7] * kernel_shared_1[threadIdx_x // 7 * 128]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 7] * kernel_shared_1[threadIdx_x // 7 * 128]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 14] * kernel_shared_1[threadIdx_x // 7 * 128]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 21] * kernel_shared_1[threadIdx_x // 7 * 128]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 28] * kernel_shared_1[threadIdx_x // 7 * 128]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 35] * kernel_shared_1[threadIdx_x // 7 * 128]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 42] * kernel_shared_1[threadIdx_x // 7 * 128]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 49] * kernel_shared_1[threadIdx_x // 7 * 128 + 1]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 56] * kernel_shared_1[threadIdx_x // 7 * 128 + 1]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 63] * kernel_shared_1[threadIdx_x // 7 * 128 + 1]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 70] * kernel_shared_1[threadIdx_x // 7 * 128 + 1]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 77] * kernel_shared_1[threadIdx_x // 7 * 128 + 1]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 84] * kernel_shared_1[threadIdx_x // 7 * 128 + 1]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 91] * kernel_shared_1[threadIdx_x // 7 * 128 + 1]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 98] * kernel_shared_1[threadIdx_x // 7 * 128 + 2]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 105] * kernel_shared_1[threadIdx_x // 7 * 128 + 2]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 112] * kernel_shared_1[threadIdx_x // 7 * 128 + 2]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 119] * kernel_shared_1[threadIdx_x // 7 * 128 + 2]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 126] * kernel_shared_1[threadIdx_x // 7 * 128 + 2]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 133] * kernel_shared_1[threadIdx_x // 7 * 128 + 2]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 140] * kernel_shared_1[threadIdx_x // 7 * 128 + 2]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 147] * kernel_shared_1[threadIdx_x // 7 * 128 + 3]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 154] * kernel_shared_1[threadIdx_x // 7 * 128 + 3]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 161] * kernel_shared_1[threadIdx_x // 7 * 128 + 3]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 168] * kernel_shared_1[threadIdx_x // 7 * 128 + 3]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 175] * kernel_shared_1[threadIdx_x // 7 * 128 + 3]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 182] * kernel_shared_1[threadIdx_x // 7 * 128 + 3]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 189] * kernel_shared_1[threadIdx_x // 7 * 128 + 3]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 196] * kernel_shared_1[threadIdx_x // 7 * 128 + 4]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 203] * kernel_shared_1[threadIdx_x // 7 * 128 + 4]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 210] * kernel_shared_1[threadIdx_x // 7 * 128 + 4]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 217] * kernel_shared_1[threadIdx_x // 7 * 128 + 4]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 224] * kernel_shared_1[threadIdx_x // 7 * 128 + 4]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 231] * kernel_shared_1[threadIdx_x // 7 * 128 + 4]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 238] * kernel_shared_1[threadIdx_x // 7 * 128 + 4]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 245] * kernel_shared_1[threadIdx_x // 7 * 128 + 5]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 252] * kernel_shared_1[threadIdx_x // 7 * 128 + 5]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 259] * kernel_shared_1[threadIdx_x // 7 * 128 + 5]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 266] * kernel_shared_1[threadIdx_x // 7 * 128 + 5]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 273] * kernel_shared_1[threadIdx_x // 7 * 128 + 5]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 280] * kernel_shared_1[threadIdx_x // 7 * 128 + 5]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 287] * kernel_shared_1[threadIdx_x // 7 * 128 + 5]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 294] * kernel_shared_1[threadIdx_x // 7 * 128 + 6]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 301] * kernel_shared_1[threadIdx_x // 7 * 128 + 6]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 308] * kernel_shared_1[threadIdx_x // 7 * 128 + 6]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 315] * kernel_shared_1[threadIdx_x // 7 * 128 + 6]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 322] * kernel_shared_1[threadIdx_x // 7 * 128 + 6]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 329] * kernel_shared_1[threadIdx_x // 7 * 128 + 6]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 336] * kernel_shared_1[threadIdx_x // 7 * 128 + 6]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 343] * kernel_shared_1[threadIdx_x // 7 * 128 + 7]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 350] * kernel_shared_1[threadIdx_x // 7 * 128 + 7]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 357] * kernel_shared_1[threadIdx_x // 7 * 128 + 7]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 364] * kernel_shared_1[threadIdx_x // 7 * 128 + 7]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 371] * kernel_shared_1[threadIdx_x // 7 * 128 + 7]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 378] * kernel_shared_1[threadIdx_x // 7 * 128 + 7]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 385] * kernel_shared_1[threadIdx_x // 7 * 128 + 7]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 392] * kernel_shared_1[threadIdx_x // 7 * 128 + 8]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 399] * kernel_shared_1[threadIdx_x // 7 * 128 + 8]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 406] * kernel_shared_1[threadIdx_x // 7 * 128 + 8]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 413] * kernel_shared_1[threadIdx_x // 7 * 128 + 8]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 420] * kernel_shared_1[threadIdx_x // 7 * 128 + 8]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 427] * kernel_shared_1[threadIdx_x // 7 * 128 + 8]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 434] * kernel_shared_1[threadIdx_x // 7 * 128 + 8]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 441] * kernel_shared_1[threadIdx_x // 7 * 128 + 9]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 448] * kernel_shared_1[threadIdx_x // 7 * 128 + 9]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 455] * kernel_shared_1[threadIdx_x // 7 * 128 + 9]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 462] * kernel_shared_1[threadIdx_x // 7 * 128 + 9]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 469] * kernel_shared_1[threadIdx_x // 7 * 128 + 9]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 476] * kernel_shared_1[threadIdx_x // 7 * 128 + 9]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 483] * kernel_shared_1[threadIdx_x // 7 * 128 + 9]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 490] * kernel_shared_1[threadIdx_x // 7 * 128 + 10]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 497] * kernel_shared_1[threadIdx_x // 7 * 128 + 10]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 504] * kernel_shared_1[threadIdx_x // 7 * 128 + 10]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 511] * kernel_shared_1[threadIdx_x // 7 * 128 + 10]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 518] * kernel_shared_1[threadIdx_x // 7 * 128 + 10]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 525] * kernel_shared_1[threadIdx_x // 7 * 128 + 10]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 532] * kernel_shared_1[threadIdx_x // 7 * 128 + 10]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 539] * kernel_shared_1[threadIdx_x // 7 * 128 + 11]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 546] * kernel_shared_1[threadIdx_x // 7 * 128 + 11]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 553] * kernel_shared_1[threadIdx_x // 7 * 128 + 11]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 560] * kernel_shared_1[threadIdx_x // 7 * 128 + 11]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 567] * kernel_shared_1[threadIdx_x // 7 * 128 + 11]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 574] * kernel_shared_1[threadIdx_x // 7 * 128 + 11]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 581] * kernel_shared_1[threadIdx_x // 7 * 128 + 11]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 588] * kernel_shared_1[threadIdx_x // 7 * 128 + 12]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 595] * kernel_shared_1[threadIdx_x // 7 * 128 + 12]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 602] * kernel_shared_1[threadIdx_x // 7 * 128 + 12]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 609] * kernel_shared_1[threadIdx_x // 7 * 128 + 12]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 616] * kernel_shared_1[threadIdx_x // 7 * 128 + 12]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 623] * kernel_shared_1[threadIdx_x // 7 * 128 + 12]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 630] * kernel_shared_1[threadIdx_x // 7 * 128 + 12]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 637] * kernel_shared_1[threadIdx_x // 7 * 128 + 13]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 644] * kernel_shared_1[threadIdx_x // 7 * 128 + 13]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 651] * kernel_shared_1[threadIdx_x // 7 * 128 + 13]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 658] * kernel_shared_1[threadIdx_x // 7 * 128 + 13]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 665] * kernel_shared_1[threadIdx_x // 7 * 128 + 13]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 672] * kernel_shared_1[threadIdx_x // 7 * 128 + 13]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 679] * kernel_shared_1[threadIdx_x // 7 * 128 + 13]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 686] * kernel_shared_1[threadIdx_x // 7 * 128 + 14]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 693] * kernel_shared_1[threadIdx_x // 7 * 128 + 14]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 700] * kernel_shared_1[threadIdx_x // 7 * 128 + 14]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 707] * kernel_shared_1[threadIdx_x // 7 * 128 + 14]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 714] * kernel_shared_1[threadIdx_x // 7 * 128 + 14]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 721] * kernel_shared_1[threadIdx_x // 7 * 128 + 14]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 728] * kernel_shared_1[threadIdx_x // 7 * 128 + 14]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 735] * kernel_shared_1[threadIdx_x // 7 * 128 + 15]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 742] * kernel_shared_1[threadIdx_x // 7 * 128 + 15]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 749] * kernel_shared_1[threadIdx_x // 7 * 128 + 15]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 756] * kernel_shared_1[threadIdx_x // 7 * 128 + 15]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 763] * kernel_shared_1[threadIdx_x // 7 * 128 + 15]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 770] * kernel_shared_1[threadIdx_x // 7 * 128 + 15]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 777] * kernel_shared_1[threadIdx_x // 7 * 128 + 15]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 784] * kernel_shared_1[threadIdx_x // 7 * 128 + 16]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 791] * kernel_shared_1[threadIdx_x // 7 * 128 + 16]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 798] * kernel_shared_1[threadIdx_x // 7 * 128 + 16]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 805] * kernel_shared_1[threadIdx_x // 7 * 128 + 16]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 812] * kernel_shared_1[threadIdx_x // 7 * 128 + 16]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 819] * kernel_shared_1[threadIdx_x // 7 * 128 + 16]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 826] * kernel_shared_1[threadIdx_x // 7 * 128 + 16]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 833] * kernel_shared_1[threadIdx_x // 7 * 128 + 17]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 840] * kernel_shared_1[threadIdx_x // 7 * 128 + 17]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 847] * kernel_shared_1[threadIdx_x // 7 * 128 + 17]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 854] * kernel_shared_1[threadIdx_x // 7 * 128 + 17]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 861] * kernel_shared_1[threadIdx_x // 7 * 128 + 17]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 868] * kernel_shared_1[threadIdx_x // 7 * 128 + 17]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 875] * kernel_shared_1[threadIdx_x // 7 * 128 + 17]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 882] * kernel_shared_1[threadIdx_x // 7 * 128 + 18]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 889] * kernel_shared_1[threadIdx_x // 7 * 128 + 18]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 896] * kernel_shared_1[threadIdx_x // 7 * 128 + 18]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 903] * kernel_shared_1[threadIdx_x // 7 * 128 + 18]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 910] * kernel_shared_1[threadIdx_x // 7 * 128 + 18]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 917] * kernel_shared_1[threadIdx_x // 7 * 128 + 18]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 924] * kernel_shared_1[threadIdx_x // 7 * 128 + 18]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 931] * kernel_shared_1[threadIdx_x // 7 * 128 + 19]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 938] * kernel_shared_1[threadIdx_x // 7 * 128 + 19]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 945] * kernel_shared_1[threadIdx_x // 7 * 128 + 19]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 952] * kernel_shared_1[threadIdx_x // 7 * 128 + 19]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 959] * kernel_shared_1[threadIdx_x // 7 * 128 + 19]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 966] * kernel_shared_1[threadIdx_x // 7 * 128 + 19]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 973] * kernel_shared_1[threadIdx_x // 7 * 128 + 19]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 980] * kernel_shared_1[threadIdx_x // 7 * 128 + 20]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 987] * kernel_shared_1[threadIdx_x // 7 * 128 + 20]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 994] * kernel_shared_1[threadIdx_x // 7 * 128 + 20]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1001] * kernel_shared_1[threadIdx_x // 7 * 128 + 20]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1008] * kernel_shared_1[threadIdx_x // 7 * 128 + 20]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1015] * kernel_shared_1[threadIdx_x // 7 * 128 + 20]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1022] * kernel_shared_1[threadIdx_x // 7 * 128 + 20]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1029] * kernel_shared_1[threadIdx_x // 7 * 128 + 21]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1036] * kernel_shared_1[threadIdx_x // 7 * 128 + 21]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1043] * kernel_shared_1[threadIdx_x // 7 * 128 + 21]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1050] * kernel_shared_1[threadIdx_x // 7 * 128 + 21]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1057] * kernel_shared_1[threadIdx_x // 7 * 128 + 21]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1064] * kernel_shared_1[threadIdx_x // 7 * 128 + 21]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1071] * kernel_shared_1[threadIdx_x // 7 * 128 + 21]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1078] * kernel_shared_1[threadIdx_x // 7 * 128 + 22]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1085] * kernel_shared_1[threadIdx_x // 7 * 128 + 22]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1092] * kernel_shared_1[threadIdx_x // 7 * 128 + 22]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1099] * kernel_shared_1[threadIdx_x // 7 * 128 + 22]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1106] * kernel_shared_1[threadIdx_x // 7 * 128 + 22]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1113] * kernel_shared_1[threadIdx_x // 7 * 128 + 22]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1120] * kernel_shared_1[threadIdx_x // 7 * 128 + 22]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1127] * kernel_shared_1[threadIdx_x // 7 * 128 + 23]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1134] * kernel_shared_1[threadIdx_x // 7 * 128 + 23]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1141] * kernel_shared_1[threadIdx_x // 7 * 128 + 23]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1148] * kernel_shared_1[threadIdx_x // 7 * 128 + 23]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1155] * kernel_shared_1[threadIdx_x // 7 * 128 + 23]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1162] * kernel_shared_1[threadIdx_x // 7 * 128 + 23]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1169] * kernel_shared_1[threadIdx_x // 7 * 128 + 23]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1176] * kernel_shared_1[threadIdx_x // 7 * 128 + 24]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1183] * kernel_shared_1[threadIdx_x // 7 * 128 + 24]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1190] * kernel_shared_1[threadIdx_x // 7 * 128 + 24]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1197] * kernel_shared_1[threadIdx_x // 7 * 128 + 24]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1204] * kernel_shared_1[threadIdx_x // 7 * 128 + 24]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1211] * kernel_shared_1[threadIdx_x // 7 * 128 + 24]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1218] * kernel_shared_1[threadIdx_x // 7 * 128 + 24]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1225] * kernel_shared_1[threadIdx_x // 7 * 128 + 25]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1232] * kernel_shared_1[threadIdx_x // 7 * 128 + 25]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1239] * kernel_shared_1[threadIdx_x // 7 * 128 + 25]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1246] * kernel_shared_1[threadIdx_x // 7 * 128 + 25]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1253] * kernel_shared_1[threadIdx_x // 7 * 128 + 25]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1260] * kernel_shared_1[threadIdx_x // 7 * 128 + 25]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1267] * kernel_shared_1[threadIdx_x // 7 * 128 + 25]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1274] * kernel_shared_1[threadIdx_x // 7 * 128 + 26]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1281] * kernel_shared_1[threadIdx_x // 7 * 128 + 26]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1288] * kernel_shared_1[threadIdx_x // 7 * 128 + 26]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1295] * kernel_shared_1[threadIdx_x // 7 * 128 + 26]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1302] * kernel_shared_1[threadIdx_x // 7 * 128 + 26]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1309] * kernel_shared_1[threadIdx_x // 7 * 128 + 26]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1316] * kernel_shared_1[threadIdx_x // 7 * 128 + 26]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1323] * kernel_shared_1[threadIdx_x // 7 * 128 + 27]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1330] * kernel_shared_1[threadIdx_x // 7 * 128 + 27]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1337] * kernel_shared_1[threadIdx_x // 7 * 128 + 27]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1344] * kernel_shared_1[threadIdx_x // 7 * 128 + 27]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1351] * kernel_shared_1[threadIdx_x // 7 * 128 + 27]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1358] * kernel_shared_1[threadIdx_x // 7 * 128 + 27]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1365] * kernel_shared_1[threadIdx_x // 7 * 128 + 27]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1372] * kernel_shared_1[threadIdx_x // 7 * 128 + 28]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1379] * kernel_shared_1[threadIdx_x // 7 * 128 + 28]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1386] * kernel_shared_1[threadIdx_x // 7 * 128 + 28]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1393] * kernel_shared_1[threadIdx_x // 7 * 128 + 28]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1400] * kernel_shared_1[threadIdx_x // 7 * 128 + 28]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1407] * kernel_shared_1[threadIdx_x // 7 * 128 + 28]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1414] * kernel_shared_1[threadIdx_x // 7 * 128 + 28]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1421] * kernel_shared_1[threadIdx_x // 7 * 128 + 29]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1428] * kernel_shared_1[threadIdx_x // 7 * 128 + 29]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1435] * kernel_shared_1[threadIdx_x // 7 * 128 + 29]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1442] * kernel_shared_1[threadIdx_x // 7 * 128 + 29]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1449] * kernel_shared_1[threadIdx_x // 7 * 128 + 29]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1456] * kernel_shared_1[threadIdx_x // 7 * 128 + 29]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1463] * kernel_shared_1[threadIdx_x // 7 * 128 + 29]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1470] * kernel_shared_1[threadIdx_x // 7 * 128 + 30]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1477] * kernel_shared_1[threadIdx_x // 7 * 128 + 30]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1484] * kernel_shared_1[threadIdx_x // 7 * 128 + 30]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1491] * kernel_shared_1[threadIdx_x // 7 * 128 + 30]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1498] * kernel_shared_1[threadIdx_x // 7 * 128 + 30]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1505] * kernel_shared_1[threadIdx_x // 7 * 128 + 30]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1512] * kernel_shared_1[threadIdx_x // 7 * 128 + 30]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1519] * kernel_shared_1[threadIdx_x // 7 * 128 + 31]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1526] * kernel_shared_1[threadIdx_x // 7 * 128 + 31]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1533] * kernel_shared_1[threadIdx_x // 7 * 128 + 31]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1540] * kernel_shared_1[threadIdx_x // 7 * 128 + 31]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1547] * kernel_shared_1[threadIdx_x // 7 * 128 + 31]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1554] * kernel_shared_1[threadIdx_x // 7 * 128 + 31]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1561] * kernel_shared_1[threadIdx_x // 7 * 128 + 31]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1568] * kernel_shared_1[threadIdx_x // 7 * 128 + 32]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1575] * kernel_shared_1[threadIdx_x // 7 * 128 + 32]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1582] * kernel_shared_1[threadIdx_x // 7 * 128 + 32]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1589] * kernel_shared_1[threadIdx_x // 7 * 128 + 32]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1596] * kernel_shared_1[threadIdx_x // 7 * 128 + 32]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1603] * kernel_shared_1[threadIdx_x // 7 * 128 + 32]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1610] * kernel_shared_1[threadIdx_x // 7 * 128 + 32]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1617] * kernel_shared_1[threadIdx_x // 7 * 128 + 33]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1624] * kernel_shared_1[threadIdx_x // 7 * 128 + 33]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1631] * kernel_shared_1[threadIdx_x // 7 * 128 + 33]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1638] * kernel_shared_1[threadIdx_x // 7 * 128 + 33]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1645] * kernel_shared_1[threadIdx_x // 7 * 128 + 33]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1652] * kernel_shared_1[threadIdx_x // 7 * 128 + 33]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1659] * kernel_shared_1[threadIdx_x // 7 * 128 + 33]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1666] * kernel_shared_1[threadIdx_x // 7 * 128 + 34]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1673] * kernel_shared_1[threadIdx_x // 7 * 128 + 34]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1680] * kernel_shared_1[threadIdx_x // 7 * 128 + 34]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1687] * kernel_shared_1[threadIdx_x // 7 * 128 + 34]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1694] * kernel_shared_1[threadIdx_x // 7 * 128 + 34]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1701] * kernel_shared_1[threadIdx_x // 7 * 128 + 34]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1708] * kernel_shared_1[threadIdx_x // 7 * 128 + 34]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1715] * kernel_shared_1[threadIdx_x // 7 * 128 + 35]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1722] * kernel_shared_1[threadIdx_x // 7 * 128 + 35]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1729] * kernel_shared_1[threadIdx_x // 7 * 128 + 35]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1736] * kernel_shared_1[threadIdx_x // 7 * 128 + 35]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1743] * kernel_shared_1[threadIdx_x // 7 * 128 + 35]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1750] * kernel_shared_1[threadIdx_x // 7 * 128 + 35]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1757] * kernel_shared_1[threadIdx_x // 7 * 128 + 35]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1764] * kernel_shared_1[threadIdx_x // 7 * 128 + 36]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1771] * kernel_shared_1[threadIdx_x // 7 * 128 + 36]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1778] * kernel_shared_1[threadIdx_x // 7 * 128 + 36]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1785] * kernel_shared_1[threadIdx_x // 7 * 128 + 36]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1792] * kernel_shared_1[threadIdx_x // 7 * 128 + 36]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1799] * kernel_shared_1[threadIdx_x // 7 * 128 + 36]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1806] * kernel_shared_1[threadIdx_x // 7 * 128 + 36]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1813] * kernel_shared_1[threadIdx_x // 7 * 128 + 37]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1820] * kernel_shared_1[threadIdx_x // 7 * 128 + 37]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1827] * kernel_shared_1[threadIdx_x // 7 * 128 + 37]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1834] * kernel_shared_1[threadIdx_x // 7 * 128 + 37]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1841] * kernel_shared_1[threadIdx_x // 7 * 128 + 37]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1848] * kernel_shared_1[threadIdx_x // 7 * 128 + 37]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1855] * kernel_shared_1[threadIdx_x // 7 * 128 + 37]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1862] * kernel_shared_1[threadIdx_x // 7 * 128 + 38]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1869] * kernel_shared_1[threadIdx_x // 7 * 128 + 38]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1876] * kernel_shared_1[threadIdx_x // 7 * 128 + 38]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1883] * kernel_shared_1[threadIdx_x // 7 * 128 + 38]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1890] * kernel_shared_1[threadIdx_x // 7 * 128 + 38]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1897] * kernel_shared_1[threadIdx_x // 7 * 128 + 38]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1904] * kernel_shared_1[threadIdx_x // 7 * 128 + 38]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1911] * kernel_shared_1[threadIdx_x // 7 * 128 + 39]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1918] * kernel_shared_1[threadIdx_x // 7 * 128 + 39]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1925] * kernel_shared_1[threadIdx_x // 7 * 128 + 39]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1932] * kernel_shared_1[threadIdx_x // 7 * 128 + 39]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1939] * kernel_shared_1[threadIdx_x // 7 * 128 + 39]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1946] * kernel_shared_1[threadIdx_x // 7 * 128 + 39]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1953] * kernel_shared_1[threadIdx_x // 7 * 128 + 39]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1960] * kernel_shared_1[threadIdx_x // 7 * 128 + 40]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1967] * kernel_shared_1[threadIdx_x // 7 * 128 + 40]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1974] * kernel_shared_1[threadIdx_x // 7 * 128 + 40]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1981] * kernel_shared_1[threadIdx_x // 7 * 128 + 40]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1988] * kernel_shared_1[threadIdx_x // 7 * 128 + 40]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1995] * kernel_shared_1[threadIdx_x // 7 * 128 + 40]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2002] * kernel_shared_1[threadIdx_x // 7 * 128 + 40]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2009] * kernel_shared_1[threadIdx_x // 7 * 128 + 41]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2016] * kernel_shared_1[threadIdx_x // 7 * 128 + 41]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2023] * kernel_shared_1[threadIdx_x // 7 * 128 + 41]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2030] * kernel_shared_1[threadIdx_x // 7 * 128 + 41]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2037] * kernel_shared_1[threadIdx_x // 7 * 128 + 41]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2044] * kernel_shared_1[threadIdx_x // 7 * 128 + 41]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2051] * kernel_shared_1[threadIdx_x // 7 * 128 + 41]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2058] * kernel_shared_1[threadIdx_x // 7 * 128 + 42]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2065] * kernel_shared_1[threadIdx_x // 7 * 128 + 42]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2072] * kernel_shared_1[threadIdx_x // 7 * 128 + 42]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2079] * kernel_shared_1[threadIdx_x // 7 * 128 + 42]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2086] * kernel_shared_1[threadIdx_x // 7 * 128 + 42]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2093] * kernel_shared_1[threadIdx_x // 7 * 128 + 42]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2100] * kernel_shared_1[threadIdx_x // 7 * 128 + 42]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2107] * kernel_shared_1[threadIdx_x // 7 * 128 + 43]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2114] * kernel_shared_1[threadIdx_x // 7 * 128 + 43]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2121] * kernel_shared_1[threadIdx_x // 7 * 128 + 43]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2128] * kernel_shared_1[threadIdx_x // 7 * 128 + 43]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2135] * kernel_shared_1[threadIdx_x // 7 * 128 + 43]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2142] * kernel_shared_1[threadIdx_x // 7 * 128 + 43]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2149] * kernel_shared_1[threadIdx_x // 7 * 128 + 43]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2156] * kernel_shared_1[threadIdx_x // 7 * 128 + 44]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2163] * kernel_shared_1[threadIdx_x // 7 * 128 + 44]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2170] * kernel_shared_1[threadIdx_x // 7 * 128 + 44]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2177] * kernel_shared_1[threadIdx_x // 7 * 128 + 44]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2184] * kernel_shared_1[threadIdx_x // 7 * 128 + 44]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2191] * kernel_shared_1[threadIdx_x // 7 * 128 + 44]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2198] * kernel_shared_1[threadIdx_x // 7 * 128 + 44]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2205] * kernel_shared_1[threadIdx_x // 7 * 128 + 45]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2212] * kernel_shared_1[threadIdx_x // 7 * 128 + 45]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2219] * kernel_shared_1[threadIdx_x // 7 * 128 + 45]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2226] * kernel_shared_1[threadIdx_x // 7 * 128 + 45]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2233] * kernel_shared_1[threadIdx_x // 7 * 128 + 45]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2240] * kernel_shared_1[threadIdx_x // 7 * 128 + 45]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2247] * kernel_shared_1[threadIdx_x // 7 * 128 + 45]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2254] * kernel_shared_1[threadIdx_x // 7 * 128 + 46]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2261] * kernel_shared_1[threadIdx_x // 7 * 128 + 46]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2268] * kernel_shared_1[threadIdx_x // 7 * 128 + 46]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2275] * kernel_shared_1[threadIdx_x // 7 * 128 + 46]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2282] * kernel_shared_1[threadIdx_x // 7 * 128 + 46]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2289] * kernel_shared_1[threadIdx_x // 7 * 128 + 46]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2296] * kernel_shared_1[threadIdx_x // 7 * 128 + 46]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2303] * kernel_shared_1[threadIdx_x // 7 * 128 + 47]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2310] * kernel_shared_1[threadIdx_x // 7 * 128 + 47]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2317] * kernel_shared_1[threadIdx_x // 7 * 128 + 47]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2324] * kernel_shared_1[threadIdx_x // 7 * 128 + 47]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2331] * kernel_shared_1[threadIdx_x // 7 * 128 + 47]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2338] * kernel_shared_1[threadIdx_x // 7 * 128 + 47]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2345] * kernel_shared_1[threadIdx_x // 7 * 128 + 47]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2352] * kernel_shared_1[threadIdx_x // 7 * 128 + 48]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2359] * kernel_shared_1[threadIdx_x // 7 * 128 + 48]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2366] * kernel_shared_1[threadIdx_x // 7 * 128 + 48]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2373] * kernel_shared_1[threadIdx_x // 7 * 128 + 48]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2380] * kernel_shared_1[threadIdx_x // 7 * 128 + 48]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2387] * kernel_shared_1[threadIdx_x // 7 * 128 + 48]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2394] * kernel_shared_1[threadIdx_x // 7 * 128 + 48]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2401] * kernel_shared_1[threadIdx_x // 7 * 128 + 49]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2408] * kernel_shared_1[threadIdx_x // 7 * 128 + 49]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2415] * kernel_shared_1[threadIdx_x // 7 * 128 + 49]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2422] * kernel_shared_1[threadIdx_x // 7 * 128 + 49]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2429] * kernel_shared_1[threadIdx_x // 7 * 128 + 49]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2436] * kernel_shared_1[threadIdx_x // 7 * 128 + 49]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2443] * kernel_shared_1[threadIdx_x // 7 * 128 + 49]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2450] * kernel_shared_1[threadIdx_x // 7 * 128 + 50]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2457] * kernel_shared_1[threadIdx_x // 7 * 128 + 50]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2464] * kernel_shared_1[threadIdx_x // 7 * 128 + 50]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2471] * kernel_shared_1[threadIdx_x // 7 * 128 + 50]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2478] * kernel_shared_1[threadIdx_x // 7 * 128 + 50]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2485] * kernel_shared_1[threadIdx_x // 7 * 128 + 50]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2492] * kernel_shared_1[threadIdx_x // 7 * 128 + 50]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2499] * kernel_shared_1[threadIdx_x // 7 * 128 + 51]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2506] * kernel_shared_1[threadIdx_x // 7 * 128 + 51]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2513] * kernel_shared_1[threadIdx_x // 7 * 128 + 51]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2520] * kernel_shared_1[threadIdx_x // 7 * 128 + 51]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2527] * kernel_shared_1[threadIdx_x // 7 * 128 + 51]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2534] * kernel_shared_1[threadIdx_x // 7 * 128 + 51]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2541] * kernel_shared_1[threadIdx_x // 7 * 128 + 51]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2548] * kernel_shared_1[threadIdx_x // 7 * 128 + 52]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2555] * kernel_shared_1[threadIdx_x // 7 * 128 + 52]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2562] * kernel_shared_1[threadIdx_x // 7 * 128 + 52]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2569] * kernel_shared_1[threadIdx_x // 7 * 128 + 52]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2576] * kernel_shared_1[threadIdx_x // 7 * 128 + 52]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2583] * kernel_shared_1[threadIdx_x // 7 * 128 + 52]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2590] * kernel_shared_1[threadIdx_x // 7 * 128 + 52]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2597] * kernel_shared_1[threadIdx_x // 7 * 128 + 53]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2604] * kernel_shared_1[threadIdx_x // 7 * 128 + 53]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2611] * kernel_shared_1[threadIdx_x // 7 * 128 + 53]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2618] * kernel_shared_1[threadIdx_x // 7 * 128 + 53]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2625] * kernel_shared_1[threadIdx_x // 7 * 128 + 53]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2632] * kernel_shared_1[threadIdx_x // 7 * 128 + 53]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2639] * kernel_shared_1[threadIdx_x // 7 * 128 + 53]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2646] * kernel_shared_1[threadIdx_x // 7 * 128 + 54]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2653] * kernel_shared_1[threadIdx_x // 7 * 128 + 54]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2660] * kernel_shared_1[threadIdx_x // 7 * 128 + 54]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2667] * kernel_shared_1[threadIdx_x // 7 * 128 + 54]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2674] * kernel_shared_1[threadIdx_x // 7 * 128 + 54]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2681] * kernel_shared_1[threadIdx_x // 7 * 128 + 54]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2688] * kernel_shared_1[threadIdx_x // 7 * 128 + 54]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2695] * kernel_shared_1[threadIdx_x // 7 * 128 + 55]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2702] * kernel_shared_1[threadIdx_x // 7 * 128 + 55]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2709] * kernel_shared_1[threadIdx_x // 7 * 128 + 55]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2716] * kernel_shared_1[threadIdx_x // 7 * 128 + 55]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2723] * kernel_shared_1[threadIdx_x // 7 * 128 + 55]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2730] * kernel_shared_1[threadIdx_x // 7 * 128 + 55]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2737] * kernel_shared_1[threadIdx_x // 7 * 128 + 55]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2744] * kernel_shared_1[threadIdx_x // 7 * 128 + 56]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2751] * kernel_shared_1[threadIdx_x // 7 * 128 + 56]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2758] * kernel_shared_1[threadIdx_x // 7 * 128 + 56]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2765] * kernel_shared_1[threadIdx_x // 7 * 128 + 56]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2772] * kernel_shared_1[threadIdx_x // 7 * 128 + 56]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2779] * kernel_shared_1[threadIdx_x // 7 * 128 + 56]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2786] * kernel_shared_1[threadIdx_x // 7 * 128 + 56]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2793] * kernel_shared_1[threadIdx_x // 7 * 128 + 57]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2800] * kernel_shared_1[threadIdx_x // 7 * 128 + 57]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2807] * kernel_shared_1[threadIdx_x // 7 * 128 + 57]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2814] * kernel_shared_1[threadIdx_x // 7 * 128 + 57]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2821] * kernel_shared_1[threadIdx_x // 7 * 128 + 57]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2828] * kernel_shared_1[threadIdx_x // 7 * 128 + 57]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2835] * kernel_shared_1[threadIdx_x // 7 * 128 + 57]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2842] * kernel_shared_1[threadIdx_x // 7 * 128 + 58]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2849] * kernel_shared_1[threadIdx_x // 7 * 128 + 58]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2856] * kernel_shared_1[threadIdx_x // 7 * 128 + 58]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2863] * kernel_shared_1[threadIdx_x // 7 * 128 + 58]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2870] * kernel_shared_1[threadIdx_x // 7 * 128 + 58]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2877] * kernel_shared_1[threadIdx_x // 7 * 128 + 58]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2884] * kernel_shared_1[threadIdx_x // 7 * 128 + 58]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2891] * kernel_shared_1[threadIdx_x // 7 * 128 + 59]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2898] * kernel_shared_1[threadIdx_x // 7 * 128 + 59]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2905] * kernel_shared_1[threadIdx_x // 7 * 128 + 59]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2912] * kernel_shared_1[threadIdx_x // 7 * 128 + 59]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2919] * kernel_shared_1[threadIdx_x // 7 * 128 + 59]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2926] * kernel_shared_1[threadIdx_x // 7 * 128 + 59]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2933] * kernel_shared_1[threadIdx_x // 7 * 128 + 59]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2940] * kernel_shared_1[threadIdx_x // 7 * 128 + 60]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2947] * kernel_shared_1[threadIdx_x // 7 * 128 + 60]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2954] * kernel_shared_1[threadIdx_x // 7 * 128 + 60]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2961] * kernel_shared_1[threadIdx_x // 7 * 128 + 60]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2968] * kernel_shared_1[threadIdx_x // 7 * 128 + 60]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2975] * kernel_shared_1[threadIdx_x // 7 * 128 + 60]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2982] * kernel_shared_1[threadIdx_x // 7 * 128 + 60]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2989] * kernel_shared_1[threadIdx_x // 7 * 128 + 61]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2996] * kernel_shared_1[threadIdx_x // 7 * 128 + 61]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3003] * kernel_shared_1[threadIdx_x // 7 * 128 + 61]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3010] * kernel_shared_1[threadIdx_x // 7 * 128 + 61]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3017] * kernel_shared_1[threadIdx_x // 7 * 128 + 61]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3024] * kernel_shared_1[threadIdx_x // 7 * 128 + 61]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3031] * kernel_shared_1[threadIdx_x // 7 * 128 + 61]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3038] * kernel_shared_1[threadIdx_x // 7 * 128 + 62]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3045] * kernel_shared_1[threadIdx_x // 7 * 128 + 62]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3052] * kernel_shared_1[threadIdx_x // 7 * 128 + 62]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3059] * kernel_shared_1[threadIdx_x // 7 * 128 + 62]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3066] * kernel_shared_1[threadIdx_x // 7 * 128 + 62]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3073] * kernel_shared_1[threadIdx_x // 7 * 128 + 62]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3080] * kernel_shared_1[threadIdx_x // 7 * 128 + 62]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3087] * kernel_shared_1[threadIdx_x // 7 * 128 + 63]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3094] * kernel_shared_1[threadIdx_x // 7 * 128 + 63]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3101] * kernel_shared_1[threadIdx_x // 7 * 128 + 63]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3108] * kernel_shared_1[threadIdx_x // 7 * 128 + 63]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3115] * kernel_shared_1[threadIdx_x // 7 * 128 + 63]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3122] * kernel_shared_1[threadIdx_x // 7 * 128 + 63]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3129] * kernel_shared_1[threadIdx_x // 7 * 128 + 63]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3136] * kernel_shared_1[threadIdx_x // 7 * 128 + 64]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3143] * kernel_shared_1[threadIdx_x // 7 * 128 + 64]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3150] * kernel_shared_1[threadIdx_x // 7 * 128 + 64]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3157] * kernel_shared_1[threadIdx_x // 7 * 128 + 64]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3164] * kernel_shared_1[threadIdx_x // 7 * 128 + 64]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3171] * kernel_shared_1[threadIdx_x // 7 * 128 + 64]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3178] * kernel_shared_1[threadIdx_x // 7 * 128 + 64]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3185] * kernel_shared_1[threadIdx_x // 7 * 128 + 65]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3192] * kernel_shared_1[threadIdx_x // 7 * 128 + 65]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3199] * kernel_shared_1[threadIdx_x // 7 * 128 + 65]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3206] * kernel_shared_1[threadIdx_x // 7 * 128 + 65]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3213] * kernel_shared_1[threadIdx_x // 7 * 128 + 65]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3220] * kernel_shared_1[threadIdx_x // 7 * 128 + 65]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3227] * kernel_shared_1[threadIdx_x // 7 * 128 + 65]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3234] * kernel_shared_1[threadIdx_x // 7 * 128 + 66]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3241] * kernel_shared_1[threadIdx_x // 7 * 128 + 66]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3248] * kernel_shared_1[threadIdx_x // 7 * 128 + 66]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3255] * kernel_shared_1[threadIdx_x // 7 * 128 + 66]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3262] * kernel_shared_1[threadIdx_x // 7 * 128 + 66]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3269] * kernel_shared_1[threadIdx_x // 7 * 128 + 66]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3276] * kernel_shared_1[threadIdx_x // 7 * 128 + 66]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3283] * kernel_shared_1[threadIdx_x // 7 * 128 + 67]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3290] * kernel_shared_1[threadIdx_x // 7 * 128 + 67]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3297] * kernel_shared_1[threadIdx_x // 7 * 128 + 67]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3304] * kernel_shared_1[threadIdx_x // 7 * 128 + 67]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3311] * kernel_shared_1[threadIdx_x // 7 * 128 + 67]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3318] * kernel_shared_1[threadIdx_x // 7 * 128 + 67]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3325] * kernel_shared_1[threadIdx_x // 7 * 128 + 67]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3332] * kernel_shared_1[threadIdx_x // 7 * 128 + 68]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3339] * kernel_shared_1[threadIdx_x // 7 * 128 + 68]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3346] * kernel_shared_1[threadIdx_x // 7 * 128 + 68]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3353] * kernel_shared_1[threadIdx_x // 7 * 128 + 68]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3360] * kernel_shared_1[threadIdx_x // 7 * 128 + 68]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3367] * kernel_shared_1[threadIdx_x // 7 * 128 + 68]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3374] * kernel_shared_1[threadIdx_x // 7 * 128 + 68]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3381] * kernel_shared_1[threadIdx_x // 7 * 128 + 69]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3388] * kernel_shared_1[threadIdx_x // 7 * 128 + 69]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3395] * kernel_shared_1[threadIdx_x // 7 * 128 + 69]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3402] * kernel_shared_1[threadIdx_x // 7 * 128 + 69]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3409] * kernel_shared_1[threadIdx_x // 7 * 128 + 69]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3416] * kernel_shared_1[threadIdx_x // 7 * 128 + 69]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3423] * kernel_shared_1[threadIdx_x // 7 * 128 + 69]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3430] * kernel_shared_1[threadIdx_x // 7 * 128 + 70]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3437] * kernel_shared_1[threadIdx_x // 7 * 128 + 70]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3444] * kernel_shared_1[threadIdx_x // 7 * 128 + 70]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3451] * kernel_shared_1[threadIdx_x // 7 * 128 + 70]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3458] * kernel_shared_1[threadIdx_x // 7 * 128 + 70]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3465] * kernel_shared_1[threadIdx_x // 7 * 128 + 70]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3472] * kernel_shared_1[threadIdx_x // 7 * 128 + 70]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3479] * kernel_shared_1[threadIdx_x // 7 * 128 + 71]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3486] * kernel_shared_1[threadIdx_x // 7 * 128 + 71]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3493] * kernel_shared_1[threadIdx_x // 7 * 128 + 71]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3500] * kernel_shared_1[threadIdx_x // 7 * 128 + 71]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3507] * kernel_shared_1[threadIdx_x // 7 * 128 + 71]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3514] * kernel_shared_1[threadIdx_x // 7 * 128 + 71]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3521] * kernel_shared_1[threadIdx_x // 7 * 128 + 71]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3528] * kernel_shared_1[threadIdx_x // 7 * 128 + 72]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3535] * kernel_shared_1[threadIdx_x // 7 * 128 + 72]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3542] * kernel_shared_1[threadIdx_x // 7 * 128 + 72]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3549] * kernel_shared_1[threadIdx_x // 7 * 128 + 72]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3556] * kernel_shared_1[threadIdx_x // 7 * 128 + 72]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3563] * kernel_shared_1[threadIdx_x // 7 * 128 + 72]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3570] * kernel_shared_1[threadIdx_x // 7 * 128 + 72]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3577] * kernel_shared_1[threadIdx_x // 7 * 128 + 73]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3584] * kernel_shared_1[threadIdx_x // 7 * 128 + 73]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3591] * kernel_shared_1[threadIdx_x // 7 * 128 + 73]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3598] * kernel_shared_1[threadIdx_x // 7 * 128 + 73]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3605] * kernel_shared_1[threadIdx_x // 7 * 128 + 73]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3612] * kernel_shared_1[threadIdx_x // 7 * 128 + 73]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3619] * kernel_shared_1[threadIdx_x // 7 * 128 + 73]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3626] * kernel_shared_1[threadIdx_x // 7 * 128 + 74]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3633] * kernel_shared_1[threadIdx_x // 7 * 128 + 74]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3640] * kernel_shared_1[threadIdx_x // 7 * 128 + 74]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3647] * kernel_shared_1[threadIdx_x // 7 * 128 + 74]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3654] * kernel_shared_1[threadIdx_x // 7 * 128 + 74]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3661] * kernel_shared_1[threadIdx_x // 7 * 128 + 74]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3668] * kernel_shared_1[threadIdx_x // 7 * 128 + 74]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3675] * kernel_shared_1[threadIdx_x // 7 * 128 + 75]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3682] * kernel_shared_1[threadIdx_x // 7 * 128 + 75]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3689] * kernel_shared_1[threadIdx_x // 7 * 128 + 75]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3696] * kernel_shared_1[threadIdx_x // 7 * 128 + 75]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3703] * kernel_shared_1[threadIdx_x // 7 * 128 + 75]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3710] * kernel_shared_1[threadIdx_x // 7 * 128 + 75]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3717] * kernel_shared_1[threadIdx_x // 7 * 128 + 75]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3724] * kernel_shared_1[threadIdx_x // 7 * 128 + 76]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3731] * kernel_shared_1[threadIdx_x // 7 * 128 + 76]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3738] * kernel_shared_1[threadIdx_x // 7 * 128 + 76]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3745] * kernel_shared_1[threadIdx_x // 7 * 128 + 76]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3752] * kernel_shared_1[threadIdx_x // 7 * 128 + 76]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3759] * kernel_shared_1[threadIdx_x // 7 * 128 + 76]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3766] * kernel_shared_1[threadIdx_x // 7 * 128 + 76]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3773] * kernel_shared_1[threadIdx_x // 7 * 128 + 77]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3780] * kernel_shared_1[threadIdx_x // 7 * 128 + 77]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3787] * kernel_shared_1[threadIdx_x // 7 * 128 + 77]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3794] * kernel_shared_1[threadIdx_x // 7 * 128 + 77]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3801] * kernel_shared_1[threadIdx_x // 7 * 128 + 77]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3808] * kernel_shared_1[threadIdx_x // 7 * 128 + 77]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3815] * kernel_shared_1[threadIdx_x // 7 * 128 + 77]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3822] * kernel_shared_1[threadIdx_x // 7 * 128 + 78]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3829] * kernel_shared_1[threadIdx_x // 7 * 128 + 78]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3836] * kernel_shared_1[threadIdx_x // 7 * 128 + 78]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3843] * kernel_shared_1[threadIdx_x // 7 * 128 + 78]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3850] * kernel_shared_1[threadIdx_x // 7 * 128 + 78]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3857] * kernel_shared_1[threadIdx_x // 7 * 128 + 78]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3864] * kernel_shared_1[threadIdx_x // 7 * 128 + 78]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3871] * kernel_shared_1[threadIdx_x // 7 * 128 + 79]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3878] * kernel_shared_1[threadIdx_x // 7 * 128 + 79]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3885] * kernel_shared_1[threadIdx_x // 7 * 128 + 79]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3892] * kernel_shared_1[threadIdx_x // 7 * 128 + 79]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3899] * kernel_shared_1[threadIdx_x // 7 * 128 + 79]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3906] * kernel_shared_1[threadIdx_x // 7 * 128 + 79]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3913] * kernel_shared_1[threadIdx_x // 7 * 128 + 79]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3920] * kernel_shared_1[threadIdx_x // 7 * 128 + 80]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3927] * kernel_shared_1[threadIdx_x // 7 * 128 + 80]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3934] * kernel_shared_1[threadIdx_x // 7 * 128 + 80]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3941] * kernel_shared_1[threadIdx_x // 7 * 128 + 80]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3948] * kernel_shared_1[threadIdx_x // 7 * 128 + 80]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3955] * kernel_shared_1[threadIdx_x // 7 * 128 + 80]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3962] * kernel_shared_1[threadIdx_x // 7 * 128 + 80]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3969] * kernel_shared_1[threadIdx_x // 7 * 128 + 81]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3976] * kernel_shared_1[threadIdx_x // 7 * 128 + 81]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3983] * kernel_shared_1[threadIdx_x // 7 * 128 + 81]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3990] * kernel_shared_1[threadIdx_x // 7 * 128 + 81]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3997] * kernel_shared_1[threadIdx_x // 7 * 128 + 81]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4004] * kernel_shared_1[threadIdx_x // 7 * 128 + 81]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4011] * kernel_shared_1[threadIdx_x // 7 * 128 + 81]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4018] * kernel_shared_1[threadIdx_x // 7 * 128 + 82]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4025] * kernel_shared_1[threadIdx_x // 7 * 128 + 82]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4032] * kernel_shared_1[threadIdx_x // 7 * 128 + 82]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4039] * kernel_shared_1[threadIdx_x // 7 * 128 + 82]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4046] * kernel_shared_1[threadIdx_x // 7 * 128 + 82]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4053] * kernel_shared_1[threadIdx_x // 7 * 128 + 82]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4060] * kernel_shared_1[threadIdx_x // 7 * 128 + 82]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4067] * kernel_shared_1[threadIdx_x // 7 * 128 + 83]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4074] * kernel_shared_1[threadIdx_x // 7 * 128 + 83]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4081] * kernel_shared_1[threadIdx_x // 7 * 128 + 83]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4088] * kernel_shared_1[threadIdx_x // 7 * 128 + 83]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4095] * kernel_shared_1[threadIdx_x // 7 * 128 + 83]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4102] * kernel_shared_1[threadIdx_x // 7 * 128 + 83]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4109] * kernel_shared_1[threadIdx_x // 7 * 128 + 83]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4116] * kernel_shared_1[threadIdx_x // 7 * 128 + 84]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4123] * kernel_shared_1[threadIdx_x // 7 * 128 + 84]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4130] * kernel_shared_1[threadIdx_x // 7 * 128 + 84]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4137] * kernel_shared_1[threadIdx_x // 7 * 128 + 84]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4144] * kernel_shared_1[threadIdx_x // 7 * 128 + 84]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4151] * kernel_shared_1[threadIdx_x // 7 * 128 + 84]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4158] * kernel_shared_1[threadIdx_x // 7 * 128 + 84]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4165] * kernel_shared_1[threadIdx_x // 7 * 128 + 85]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4172] * kernel_shared_1[threadIdx_x // 7 * 128 + 85]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4179] * kernel_shared_1[threadIdx_x // 7 * 128 + 85]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4186] * kernel_shared_1[threadIdx_x // 7 * 128 + 85]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4193] * kernel_shared_1[threadIdx_x // 7 * 128 + 85]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4200] * kernel_shared_1[threadIdx_x // 7 * 128 + 85]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4207] * kernel_shared_1[threadIdx_x // 7 * 128 + 85]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4214] * kernel_shared_1[threadIdx_x // 7 * 128 + 86]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4221] * kernel_shared_1[threadIdx_x // 7 * 128 + 86]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4228] * kernel_shared_1[threadIdx_x // 7 * 128 + 86]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4235] * kernel_shared_1[threadIdx_x // 7 * 128 + 86]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4242] * kernel_shared_1[threadIdx_x // 7 * 128 + 86]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4249] * kernel_shared_1[threadIdx_x // 7 * 128 + 86]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4256] * kernel_shared_1[threadIdx_x // 7 * 128 + 86]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4263] * kernel_shared_1[threadIdx_x // 7 * 128 + 87]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4270] * kernel_shared_1[threadIdx_x // 7 * 128 + 87]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4277] * kernel_shared_1[threadIdx_x // 7 * 128 + 87]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4284] * kernel_shared_1[threadIdx_x // 7 * 128 + 87]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4291] * kernel_shared_1[threadIdx_x // 7 * 128 + 87]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4298] * kernel_shared_1[threadIdx_x // 7 * 128 + 87]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4305] * kernel_shared_1[threadIdx_x // 7 * 128 + 87]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4312] * kernel_shared_1[threadIdx_x // 7 * 128 + 88]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4319] * kernel_shared_1[threadIdx_x // 7 * 128 + 88]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4326] * kernel_shared_1[threadIdx_x // 7 * 128 + 88]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4333] * kernel_shared_1[threadIdx_x // 7 * 128 + 88]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4340] * kernel_shared_1[threadIdx_x // 7 * 128 + 88]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4347] * kernel_shared_1[threadIdx_x // 7 * 128 + 88]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4354] * kernel_shared_1[threadIdx_x // 7 * 128 + 88]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4361] * kernel_shared_1[threadIdx_x // 7 * 128 + 89]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4368] * kernel_shared_1[threadIdx_x // 7 * 128 + 89]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4375] * kernel_shared_1[threadIdx_x // 7 * 128 + 89]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4382] * kernel_shared_1[threadIdx_x // 7 * 128 + 89]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4389] * kernel_shared_1[threadIdx_x // 7 * 128 + 89]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4396] * kernel_shared_1[threadIdx_x // 7 * 128 + 89]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4403] * kernel_shared_1[threadIdx_x // 7 * 128 + 89]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4410] * kernel_shared_1[threadIdx_x // 7 * 128 + 90]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4417] * kernel_shared_1[threadIdx_x // 7 * 128 + 90]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4424] * kernel_shared_1[threadIdx_x // 7 * 128 + 90]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4431] * kernel_shared_1[threadIdx_x // 7 * 128 + 90]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4438] * kernel_shared_1[threadIdx_x // 7 * 128 + 90]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4445] * kernel_shared_1[threadIdx_x // 7 * 128 + 90]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4452] * kernel_shared_1[threadIdx_x // 7 * 128 + 90]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4459] * kernel_shared_1[threadIdx_x // 7 * 128 + 91]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4466] * kernel_shared_1[threadIdx_x // 7 * 128 + 91]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4473] * kernel_shared_1[threadIdx_x // 7 * 128 + 91]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4480] * kernel_shared_1[threadIdx_x // 7 * 128 + 91]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4487] * kernel_shared_1[threadIdx_x // 7 * 128 + 91]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4494] * kernel_shared_1[threadIdx_x // 7 * 128 + 91]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4501] * kernel_shared_1[threadIdx_x // 7 * 128 + 91]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4508] * kernel_shared_1[threadIdx_x // 7 * 128 + 92]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4515] * kernel_shared_1[threadIdx_x // 7 * 128 + 92]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4522] * kernel_shared_1[threadIdx_x // 7 * 128 + 92]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4529] * kernel_shared_1[threadIdx_x // 7 * 128 + 92]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4536] * kernel_shared_1[threadIdx_x // 7 * 128 + 92]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4543] * kernel_shared_1[threadIdx_x // 7 * 128 + 92]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4550] * kernel_shared_1[threadIdx_x // 7 * 128 + 92]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4557] * kernel_shared_1[threadIdx_x // 7 * 128 + 93]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4564] * kernel_shared_1[threadIdx_x // 7 * 128 + 93]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4571] * kernel_shared_1[threadIdx_x // 7 * 128 + 93]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4578] * kernel_shared_1[threadIdx_x // 7 * 128 + 93]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4585] * kernel_shared_1[threadIdx_x // 7 * 128 + 93]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4592] * kernel_shared_1[threadIdx_x // 7 * 128 + 93]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4599] * kernel_shared_1[threadIdx_x // 7 * 128 + 93]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4606] * kernel_shared_1[threadIdx_x // 7 * 128 + 94]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4613] * kernel_shared_1[threadIdx_x // 7 * 128 + 94]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4620] * kernel_shared_1[threadIdx_x // 7 * 128 + 94]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4627] * kernel_shared_1[threadIdx_x // 7 * 128 + 94]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4634] * kernel_shared_1[threadIdx_x // 7 * 128 + 94]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4641] * kernel_shared_1[threadIdx_x // 7 * 128 + 94]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4648] * kernel_shared_1[threadIdx_x // 7 * 128 + 94]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4655] * kernel_shared_1[threadIdx_x // 7 * 128 + 95]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4662] * kernel_shared_1[threadIdx_x // 7 * 128 + 95]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4669] * kernel_shared_1[threadIdx_x // 7 * 128 + 95]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4676] * kernel_shared_1[threadIdx_x // 7 * 128 + 95]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4683] * kernel_shared_1[threadIdx_x // 7 * 128 + 95]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4690] * kernel_shared_1[threadIdx_x // 7 * 128 + 95]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4697] * kernel_shared_1[threadIdx_x // 7 * 128 + 95]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4704] * kernel_shared_1[threadIdx_x // 7 * 128 + 96]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4711] * kernel_shared_1[threadIdx_x // 7 * 128 + 96]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4718] * kernel_shared_1[threadIdx_x // 7 * 128 + 96]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4725] * kernel_shared_1[threadIdx_x // 7 * 128 + 96]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4732] * kernel_shared_1[threadIdx_x // 7 * 128 + 96]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4739] * kernel_shared_1[threadIdx_x // 7 * 128 + 96]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4746] * kernel_shared_1[threadIdx_x // 7 * 128 + 96]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4753] * kernel_shared_1[threadIdx_x // 7 * 128 + 97]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4760] * kernel_shared_1[threadIdx_x // 7 * 128 + 97]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4767] * kernel_shared_1[threadIdx_x // 7 * 128 + 97]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4774] * kernel_shared_1[threadIdx_x // 7 * 128 + 97]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4781] * kernel_shared_1[threadIdx_x // 7 * 128 + 97]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4788] * kernel_shared_1[threadIdx_x // 7 * 128 + 97]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4795] * kernel_shared_1[threadIdx_x // 7 * 128 + 97]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4802] * kernel_shared_1[threadIdx_x // 7 * 128 + 98]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4809] * kernel_shared_1[threadIdx_x // 7 * 128 + 98]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4816] * kernel_shared_1[threadIdx_x // 7 * 128 + 98]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4823] * kernel_shared_1[threadIdx_x // 7 * 128 + 98]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4830] * kernel_shared_1[threadIdx_x // 7 * 128 + 98]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4837] * kernel_shared_1[threadIdx_x // 7 * 128 + 98]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4844] * kernel_shared_1[threadIdx_x // 7 * 128 + 98]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4851] * kernel_shared_1[threadIdx_x // 7 * 128 + 99]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4858] * kernel_shared_1[threadIdx_x // 7 * 128 + 99]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4865] * kernel_shared_1[threadIdx_x // 7 * 128 + 99]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4872] * kernel_shared_1[threadIdx_x // 7 * 128 + 99]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4879] * kernel_shared_1[threadIdx_x // 7 * 128 + 99]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4886] * kernel_shared_1[threadIdx_x // 7 * 128 + 99]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4893] * kernel_shared_1[threadIdx_x // 7 * 128 + 99]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4900] * kernel_shared_1[threadIdx_x // 7 * 128 + 100]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4907] * kernel_shared_1[threadIdx_x // 7 * 128 + 100]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4914] * kernel_shared_1[threadIdx_x // 7 * 128 + 100]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4921] * kernel_shared_1[threadIdx_x // 7 * 128 + 100]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4928] * kernel_shared_1[threadIdx_x // 7 * 128 + 100]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4935] * kernel_shared_1[threadIdx_x // 7 * 128 + 100]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4942] * kernel_shared_1[threadIdx_x // 7 * 128 + 100]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4949] * kernel_shared_1[threadIdx_x // 7 * 128 + 101]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4956] * kernel_shared_1[threadIdx_x // 7 * 128 + 101]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4963] * kernel_shared_1[threadIdx_x // 7 * 128 + 101]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4970] * kernel_shared_1[threadIdx_x // 7 * 128 + 101]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4977] * kernel_shared_1[threadIdx_x // 7 * 128 + 101]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4984] * kernel_shared_1[threadIdx_x // 7 * 128 + 101]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4991] * kernel_shared_1[threadIdx_x // 7 * 128 + 101]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4998] * kernel_shared_1[threadIdx_x // 7 * 128 + 102]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5005] * kernel_shared_1[threadIdx_x // 7 * 128 + 102]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5012] * kernel_shared_1[threadIdx_x // 7 * 128 + 102]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5019] * kernel_shared_1[threadIdx_x // 7 * 128 + 102]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5026] * kernel_shared_1[threadIdx_x // 7 * 128 + 102]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5033] * kernel_shared_1[threadIdx_x // 7 * 128 + 102]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5040] * kernel_shared_1[threadIdx_x // 7 * 128 + 102]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5047] * kernel_shared_1[threadIdx_x // 7 * 128 + 103]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5054] * kernel_shared_1[threadIdx_x // 7 * 128 + 103]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5061] * kernel_shared_1[threadIdx_x // 7 * 128 + 103]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5068] * kernel_shared_1[threadIdx_x // 7 * 128 + 103]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5075] * kernel_shared_1[threadIdx_x // 7 * 128 + 103]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5082] * kernel_shared_1[threadIdx_x // 7 * 128 + 103]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5089] * kernel_shared_1[threadIdx_x // 7 * 128 + 103]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5096] * kernel_shared_1[threadIdx_x // 7 * 128 + 104]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5103] * kernel_shared_1[threadIdx_x // 7 * 128 + 104]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5110] * kernel_shared_1[threadIdx_x // 7 * 128 + 104]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5117] * kernel_shared_1[threadIdx_x // 7 * 128 + 104]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5124] * kernel_shared_1[threadIdx_x // 7 * 128 + 104]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5131] * kernel_shared_1[threadIdx_x // 7 * 128 + 104]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5138] * kernel_shared_1[threadIdx_x // 7 * 128 + 104]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5145] * kernel_shared_1[threadIdx_x // 7 * 128 + 105]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5152] * kernel_shared_1[threadIdx_x // 7 * 128 + 105]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5159] * kernel_shared_1[threadIdx_x // 7 * 128 + 105]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5166] * kernel_shared_1[threadIdx_x // 7 * 128 + 105]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5173] * kernel_shared_1[threadIdx_x // 7 * 128 + 105]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5180] * kernel_shared_1[threadIdx_x // 7 * 128 + 105]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5187] * kernel_shared_1[threadIdx_x // 7 * 128 + 105]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5194] * kernel_shared_1[threadIdx_x // 7 * 128 + 106]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5201] * kernel_shared_1[threadIdx_x // 7 * 128 + 106]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5208] * kernel_shared_1[threadIdx_x // 7 * 128 + 106]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5215] * kernel_shared_1[threadIdx_x // 7 * 128 + 106]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5222] * kernel_shared_1[threadIdx_x // 7 * 128 + 106]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5229] * kernel_shared_1[threadIdx_x // 7 * 128 + 106]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5236] * kernel_shared_1[threadIdx_x // 7 * 128 + 106]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5243] * kernel_shared_1[threadIdx_x // 7 * 128 + 107]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5250] * kernel_shared_1[threadIdx_x // 7 * 128 + 107]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5257] * kernel_shared_1[threadIdx_x // 7 * 128 + 107]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5264] * kernel_shared_1[threadIdx_x // 7 * 128 + 107]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5271] * kernel_shared_1[threadIdx_x // 7 * 128 + 107]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5278] * kernel_shared_1[threadIdx_x // 7 * 128 + 107]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5285] * kernel_shared_1[threadIdx_x // 7 * 128 + 107]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5292] * kernel_shared_1[threadIdx_x // 7 * 128 + 108]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5299] * kernel_shared_1[threadIdx_x // 7 * 128 + 108]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5306] * kernel_shared_1[threadIdx_x // 7 * 128 + 108]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5313] * kernel_shared_1[threadIdx_x // 7 * 128 + 108]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5320] * kernel_shared_1[threadIdx_x // 7 * 128 + 108]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5327] * kernel_shared_1[threadIdx_x // 7 * 128 + 108]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5334] * kernel_shared_1[threadIdx_x // 7 * 128 + 108]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5341] * kernel_shared_1[threadIdx_x // 7 * 128 + 109]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5348] * kernel_shared_1[threadIdx_x // 7 * 128 + 109]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5355] * kernel_shared_1[threadIdx_x // 7 * 128 + 109]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5362] * kernel_shared_1[threadIdx_x // 7 * 128 + 109]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5369] * kernel_shared_1[threadIdx_x // 7 * 128 + 109]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5376] * kernel_shared_1[threadIdx_x // 7 * 128 + 109]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5383] * kernel_shared_1[threadIdx_x // 7 * 128 + 109]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5390] * kernel_shared_1[threadIdx_x // 7 * 128 + 110]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5397] * kernel_shared_1[threadIdx_x // 7 * 128 + 110]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5404] * kernel_shared_1[threadIdx_x // 7 * 128 + 110]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5411] * kernel_shared_1[threadIdx_x // 7 * 128 + 110]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5418] * kernel_shared_1[threadIdx_x // 7 * 128 + 110]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5425] * kernel_shared_1[threadIdx_x // 7 * 128 + 110]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5432] * kernel_shared_1[threadIdx_x // 7 * 128 + 110]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5439] * kernel_shared_1[threadIdx_x // 7 * 128 + 111]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5446] * kernel_shared_1[threadIdx_x // 7 * 128 + 111]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5453] * kernel_shared_1[threadIdx_x // 7 * 128 + 111]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5460] * kernel_shared_1[threadIdx_x // 7 * 128 + 111]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5467] * kernel_shared_1[threadIdx_x // 7 * 128 + 111]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5474] * kernel_shared_1[threadIdx_x // 7 * 128 + 111]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5481] * kernel_shared_1[threadIdx_x // 7 * 128 + 111]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5488] * kernel_shared_1[threadIdx_x // 7 * 128 + 112]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5495] * kernel_shared_1[threadIdx_x // 7 * 128 + 112]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5502] * kernel_shared_1[threadIdx_x // 7 * 128 + 112]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5509] * kernel_shared_1[threadIdx_x // 7 * 128 + 112]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5516] * kernel_shared_1[threadIdx_x // 7 * 128 + 112]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5523] * kernel_shared_1[threadIdx_x // 7 * 128 + 112]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5530] * kernel_shared_1[threadIdx_x // 7 * 128 + 112]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5537] * kernel_shared_1[threadIdx_x // 7 * 128 + 113]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5544] * kernel_shared_1[threadIdx_x // 7 * 128 + 113]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5551] * kernel_shared_1[threadIdx_x // 7 * 128 + 113]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5558] * kernel_shared_1[threadIdx_x // 7 * 128 + 113]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5565] * kernel_shared_1[threadIdx_x // 7 * 128 + 113]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5572] * kernel_shared_1[threadIdx_x // 7 * 128 + 113]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5579] * kernel_shared_1[threadIdx_x // 7 * 128 + 113]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5586] * kernel_shared_1[threadIdx_x // 7 * 128 + 114]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5593] * kernel_shared_1[threadIdx_x // 7 * 128 + 114]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5600] * kernel_shared_1[threadIdx_x // 7 * 128 + 114]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5607] * kernel_shared_1[threadIdx_x // 7 * 128 + 114]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5614] * kernel_shared_1[threadIdx_x // 7 * 128 + 114]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5621] * kernel_shared_1[threadIdx_x // 7 * 128 + 114]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5628] * kernel_shared_1[threadIdx_x // 7 * 128 + 114]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5635] * kernel_shared_1[threadIdx_x // 7 * 128 + 115]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5642] * kernel_shared_1[threadIdx_x // 7 * 128 + 115]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5649] * kernel_shared_1[threadIdx_x // 7 * 128 + 115]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5656] * kernel_shared_1[threadIdx_x // 7 * 128 + 115]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5663] * kernel_shared_1[threadIdx_x // 7 * 128 + 115]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5670] * kernel_shared_1[threadIdx_x // 7 * 128 + 115]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5677] * kernel_shared_1[threadIdx_x // 7 * 128 + 115]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5684] * kernel_shared_1[threadIdx_x // 7 * 128 + 116]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5691] * kernel_shared_1[threadIdx_x // 7 * 128 + 116]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5698] * kernel_shared_1[threadIdx_x // 7 * 128 + 116]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5705] * kernel_shared_1[threadIdx_x // 7 * 128 + 116]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5712] * kernel_shared_1[threadIdx_x // 7 * 128 + 116]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5719] * kernel_shared_1[threadIdx_x // 7 * 128 + 116]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5726] * kernel_shared_1[threadIdx_x // 7 * 128 + 116]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5733] * kernel_shared_1[threadIdx_x // 7 * 128 + 117]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5740] * kernel_shared_1[threadIdx_x // 7 * 128 + 117]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5747] * kernel_shared_1[threadIdx_x // 7 * 128 + 117]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5754] * kernel_shared_1[threadIdx_x // 7 * 128 + 117]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5761] * kernel_shared_1[threadIdx_x // 7 * 128 + 117]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5768] * kernel_shared_1[threadIdx_x // 7 * 128 + 117]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5775] * kernel_shared_1[threadIdx_x // 7 * 128 + 117]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5782] * kernel_shared_1[threadIdx_x // 7 * 128 + 118]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5789] * kernel_shared_1[threadIdx_x // 7 * 128 + 118]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5796] * kernel_shared_1[threadIdx_x // 7 * 128 + 118]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5803] * kernel_shared_1[threadIdx_x // 7 * 128 + 118]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5810] * kernel_shared_1[threadIdx_x // 7 * 128 + 118]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5817] * kernel_shared_1[threadIdx_x // 7 * 128 + 118]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5824] * kernel_shared_1[threadIdx_x // 7 * 128 + 118]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5831] * kernel_shared_1[threadIdx_x // 7 * 128 + 119]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5838] * kernel_shared_1[threadIdx_x // 7 * 128 + 119]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5845] * kernel_shared_1[threadIdx_x // 7 * 128 + 119]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5852] * kernel_shared_1[threadIdx_x // 7 * 128 + 119]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5859] * kernel_shared_1[threadIdx_x // 7 * 128 + 119]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5866] * kernel_shared_1[threadIdx_x // 7 * 128 + 119]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5873] * kernel_shared_1[threadIdx_x // 7 * 128 + 119]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5880] * kernel_shared_1[threadIdx_x // 7 * 128 + 120]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5887] * kernel_shared_1[threadIdx_x // 7 * 128 + 120]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5894] * kernel_shared_1[threadIdx_x // 7 * 128 + 120]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5901] * kernel_shared_1[threadIdx_x // 7 * 128 + 120]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5908] * kernel_shared_1[threadIdx_x // 7 * 128 + 120]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5915] * kernel_shared_1[threadIdx_x // 7 * 128 + 120]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5922] * kernel_shared_1[threadIdx_x // 7 * 128 + 120]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5929] * kernel_shared_1[threadIdx_x // 7 * 128 + 121]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5936] * kernel_shared_1[threadIdx_x // 7 * 128 + 121]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5943] * kernel_shared_1[threadIdx_x // 7 * 128 + 121]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5950] * kernel_shared_1[threadIdx_x // 7 * 128 + 121]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5957] * kernel_shared_1[threadIdx_x // 7 * 128 + 121]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5964] * kernel_shared_1[threadIdx_x // 7 * 128 + 121]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5971] * kernel_shared_1[threadIdx_x // 7 * 128 + 121]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5978] * kernel_shared_1[threadIdx_x // 7 * 128 + 122]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5985] * kernel_shared_1[threadIdx_x // 7 * 128 + 122]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5992] * kernel_shared_1[threadIdx_x // 7 * 128 + 122]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5999] * kernel_shared_1[threadIdx_x // 7 * 128 + 122]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 6006] * kernel_shared_1[threadIdx_x // 7 * 128 + 122]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 6013] * kernel_shared_1[threadIdx_x // 7 * 128 + 122]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 6020] * kernel_shared_1[threadIdx_x // 7 * 128 + 122]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 6027] * kernel_shared_1[threadIdx_x // 7 * 128 + 123]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 6034] * kernel_shared_1[threadIdx_x // 7 * 128 + 123]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 6041] * kernel_shared_1[threadIdx_x // 7 * 128 + 123]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 6048] * kernel_shared_1[threadIdx_x // 7 * 128 + 123]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 6055] * kernel_shared_1[threadIdx_x // 7 * 128 + 123]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 6062] * kernel_shared_1[threadIdx_x // 7 * 128 + 123]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 6069] * kernel_shared_1[threadIdx_x // 7 * 128 + 123]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 6076] * kernel_shared_1[threadIdx_x // 7 * 128 + 124]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 6083] * kernel_shared_1[threadIdx_x // 7 * 128 + 124]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 6090] * kernel_shared_1[threadIdx_x // 7 * 128 + 124]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 6097] * kernel_shared_1[threadIdx_x // 7 * 128 + 124]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 6104] * kernel_shared_1[threadIdx_x // 7 * 128 + 124]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 6111] * kernel_shared_1[threadIdx_x // 7 * 128 + 124]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 6118] * kernel_shared_1[threadIdx_x // 7 * 128 + 124]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 6125] * kernel_shared_1[threadIdx_x // 7 * 128 + 125]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 6132] * kernel_shared_1[threadIdx_x // 7 * 128 + 125]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 6139] * kernel_shared_1[threadIdx_x // 7 * 128 + 125]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 6146] * kernel_shared_1[threadIdx_x // 7 * 128 + 125]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 6153] * kernel_shared_1[threadIdx_x // 7 * 128 + 125]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 6160] * kernel_shared_1[threadIdx_x // 7 * 128 + 125]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 6167] * kernel_shared_1[threadIdx_x // 7 * 128 + 125]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 6174] * kernel_shared_1[threadIdx_x // 7 * 128 + 126]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 6181] * kernel_shared_1[threadIdx_x // 7 * 128 + 126]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 6188] * kernel_shared_1[threadIdx_x // 7 * 128 + 126]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 6195] * kernel_shared_1[threadIdx_x // 7 * 128 + 126]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 6202] * kernel_shared_1[threadIdx_x // 7 * 128 + 126]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 6209] * kernel_shared_1[threadIdx_x // 7 * 128 + 126]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 6216] * kernel_shared_1[threadIdx_x // 7 * 128 + 126]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 6223] * kernel_shared_1[threadIdx_x // 7 * 128 + 127]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 6230] * kernel_shared_1[threadIdx_x // 7 * 128 + 127]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 6237] * kernel_shared_1[threadIdx_x // 7 * 128 + 127]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 6244] * kernel_shared_1[threadIdx_x // 7 * 128 + 127]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 6251] * kernel_shared_1[threadIdx_x // 7 * 128 + 127]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 6258] * kernel_shared_1[threadIdx_x // 7 * 128 + 127]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 6265] * kernel_shared_1[threadIdx_x // 7 * 128 + 127]
- compute_1 = T.Buffer((25088,), data=compute.data)
- bias_1 = T.Buffer((512,), data=bias.data)
- compute_1[blockIdx_x * 784 + threadIdx_x // 7 * 49 + threadIdx_x % 7] = T.max(conv2d_nchw_1[0] + bias_1[blockIdx_x * 16 + threadIdx_x // 7], T.float32(0))
- compute_1[blockIdx_x * 784 + threadIdx_x // 7 * 49 + threadIdx_x % 7 + 7] = T.max(conv2d_nchw_1[1] + bias_1[blockIdx_x * 16 + threadIdx_x // 7], T.float32(0))
- compute_1[blockIdx_x * 784 + threadIdx_x // 7 * 49 + threadIdx_x % 7 + 14] = T.max(conv2d_nchw_1[2] + bias_1[blockIdx_x * 16 + threadIdx_x // 7], T.float32(0))
- compute_1[blockIdx_x * 784 + threadIdx_x // 7 * 49 + threadIdx_x % 7 + 21] = T.max(conv2d_nchw_1[3] + bias_1[blockIdx_x * 16 + threadIdx_x // 7], T.float32(0))
- compute_1[blockIdx_x * 784 + threadIdx_x // 7 * 49 + threadIdx_x % 7 + 28] = T.max(conv2d_nchw_1[4] + bias_1[blockIdx_x * 16 + threadIdx_x // 7], T.float32(0))
- compute_1[blockIdx_x * 784 + threadIdx_x // 7 * 49 + threadIdx_x % 7 + 35] = T.max(conv2d_nchw_1[5] + bias_1[blockIdx_x * 16 + threadIdx_x // 7], T.float32(0))
- compute_1[blockIdx_x * 784 + threadIdx_x // 7 * 49 + threadIdx_x % 7 + 42] = T.max(conv2d_nchw_1[6] + bias_1[blockIdx_x * 16 + threadIdx_x // 7], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1] = T.if_then_else(1 <= threadIdx_x_1 // 9 + ry_outer_outer and threadIdx_x_1 // 9 + ry_outer_outer < 8 and 1 <= threadIdx_x_1 % 9 and threadIdx_x_1 % 9 < 8, data_1[cse_var_4 + threadIdx_x_1 // 9 * 7 + cse_var_3 + 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(1 <= (threadIdx_x_1 + 56) % 63 // 9 + ry_outer_outer and (threadIdx_x_1 + 56) % 63 // 9 + ry_outer_outer < 8 and 1 <= (threadIdx_x_1 + 2) % 9 and (threadIdx_x_1 + 2) % 9 < 8, data_1[cse_var_4 + (threadIdx_x_1 + 56) // 9 * 7 + cse_var_3 + (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(1 <= (threadIdx_x_1 + 49) % 63 // 9 + ry_outer_outer and (threadIdx_x_1 + 49) % 63 // 9 + ry_outer_outer < 8 and 1 <= (threadIdx_x_1 + 4) % 9 and (threadIdx_x_1 + 4) % 9 < 8, data_1[cse_var_4 + (threadIdx_x_1 + 112) // 9 * 7 + cse_var_3 + (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(1 <= (threadIdx_x_1 + 42) % 63 // 9 + ry_outer_outer and (threadIdx_x_1 + 42) % 63 // 9 + ry_outer_outer < 8 and 1 <= (threadIdx_x_1 + 6) % 9 and (threadIdx_x_1 + 6) % 9 < 8, data_1[cse_var_4 + (threadIdx_x_1 + 168) // 9 * 7 + cse_var_3 + (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(1 <= (threadIdx_x_1 + 35) % 63 // 9 + ry_outer_outer and (threadIdx_x_1 + 35) % 63 // 9 + ry_outer_outer < 8 and 1 <= (threadIdx_x_1 + 8) % 9 and (threadIdx_x_1 + 8) % 9 < 8, data_1[cse_var_4 + (threadIdx_x_1 + 224) // 9 * 7 + cse_var_3 + (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(1 <= (threadIdx_x_1 + 28) % 63 // 9 + ry_outer_outer and (threadIdx_x_1 + 28) % 63 // 9 + ry_outer_outer < 8 and 1 <= (threadIdx_x_1 + 1) % 9 and (threadIdx_x_1 + 1) % 9 < 8, data_1[cse_var_4 + (threadIdx_x_1 + 280) // 9 * 7 + cse_var_3 + (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 + 21) % 63 // 9 + ry_outer_outer and (threadIdx_x_1 + 21) % 63 // 9 + ry_outer_outer < 8 and 1 <= (threadIdx_x_1 + 3) % 9 and (threadIdx_x_1 + 3) % 9 < 8, data_1[cse_var_4 + (threadIdx_x_1 + 336) // 9 * 7 + cse_var_3 + (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(1 <= (threadIdx_x_1 + 14) % 63 // 9 + ry_outer_outer and (threadIdx_x_1 + 14) % 63 // 9 + ry_outer_outer < 8 and 1 <= (threadIdx_x_1 + 5) % 9 and (threadIdx_x_1 + 5) % 9 < 8, data_1[cse_var_4 + (threadIdx_x_1 + 392) // 9 * 7 + cse_var_3 + (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(1 <= (threadIdx_x_1 + 7) // 9 + ry_outer_outer and (threadIdx_x_1 + 7) // 9 + ry_outer_outer < 8 and 1 <= (threadIdx_x_1 + 7) % 9 and (threadIdx_x_1 + 7) % 9 < 8, data_1[cse_var_4 + (threadIdx_x_1 + 448) // 9 * 7 + cse_var_3 + (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(1 <= threadIdx_x_1 // 9 + ry_outer_outer and threadIdx_x_1 // 9 + ry_outer_outer < 8 and 1 <= threadIdx_x_1 % 9 and threadIdx_x_1 % 9 < 8, data_1[cse_var_4 + threadIdx_x_1 // 9 * 7 + cse_var_3 + threadIdx_x_1 % 9 + 384], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 560] = T.if_then_else(1 <= (threadIdx_x_1 + 56) % 63 // 9 + ry_outer_outer and (threadIdx_x_1 + 56) % 63 // 9 + ry_outer_outer < 8 and 1 <= (threadIdx_x_1 + 2) % 9 and (threadIdx_x_1 + 2) % 9 < 8, data_1[cse_var_4 + (threadIdx_x_1 + 560) // 9 * 7 + cse_var_3 + (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(1 <= (threadIdx_x_1 + 49) % 63 // 9 + ry_outer_outer and (threadIdx_x_1 + 49) % 63 // 9 + ry_outer_outer < 8 and 1 <= (threadIdx_x_1 + 4) % 9 and (threadIdx_x_1 + 4) % 9 < 8, data_1[cse_var_4 + (threadIdx_x_1 + 616) // 9 * 7 + cse_var_3 + (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(1 <= (threadIdx_x_1 + 42) % 63 // 9 + ry_outer_outer and (threadIdx_x_1 + 42) % 63 // 9 + ry_outer_outer < 8 and 1 <= (threadIdx_x_1 + 6) % 9 and (threadIdx_x_1 + 6) % 9 < 8, data_1[cse_var_4 + (threadIdx_x_1 + 672) // 9 * 7 + cse_var_3 + (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(1 <= (threadIdx_x_1 + 35) % 63 // 9 + ry_outer_outer and (threadIdx_x_1 + 35) % 63 // 9 + ry_outer_outer < 8 and 1 <= (threadIdx_x_1 + 8) % 9 and (threadIdx_x_1 + 8) % 9 < 8, data_1[cse_var_4 + (threadIdx_x_1 + 728) // 9 * 7 + cse_var_3 + (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(1 <= (threadIdx_x_1 + 28) % 63 // 9 + ry_outer_outer and (threadIdx_x_1 + 28) % 63 // 9 + ry_outer_outer < 8 and 1 <= (threadIdx_x_1 + 1) % 9 and (threadIdx_x_1 + 1) % 9 < 8, data_1[cse_var_4 + (threadIdx_x_1 + 784) // 9 * 7 + cse_var_3 + (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(1 <= (threadIdx_x_1 + 21) % 63 // 9 + ry_outer_outer and (threadIdx_x_1 + 21) % 63 // 9 + ry_outer_outer < 8 and 1 <= (threadIdx_x_1 + 3) % 9 and (threadIdx_x_1 + 3) % 9 < 8, data_1[cse_var_4 + (threadIdx_x_1 + 840) // 9 * 7 + cse_var_3 + (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(1 <= (threadIdx_x_1 + 14) % 63 // 9 + ry_outer_outer and (threadIdx_x_1 + 14) % 63 // 9 + ry_outer_outer < 8 and 1 <= (threadIdx_x_1 + 5) % 9 and (threadIdx_x_1 + 5) % 9 < 8, data_1[cse_var_4 + (threadIdx_x_1 + 896) // 9 * 7 + cse_var_3 + (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(1 <= (threadIdx_x_1 + 7) // 9 + ry_outer_outer and (threadIdx_x_1 + 7) // 9 + ry_outer_outer < 8 and 1 <= (threadIdx_x_1 + 7) % 9 and (threadIdx_x_1 + 7) % 9 < 8, data_1[cse_var_4 + (threadIdx_x_1 + 952) // 9 * 7 + cse_var_3 + (threadIdx_x_1 + 7) % 9 - 8], T.float32(0))
+ threadIdx_x_2 = T.env_thread("threadIdx.x")
+ kernel_shared_1 = T.Buffer((1536,), 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 * 147456 + threadIdx_x_2 // 48 * 4608 + cse_var_2 + threadIdx_x_2 % 48 // 3 * 9 + cse_var_1 + threadIdx_x_2 % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 56) // 48 * 48 + (threadIdx_x_2 + 8) % 48 // 3 * 3 + (threadIdx_x_2 + 2) % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 56) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 + 8) % 48 // 3 * 9 + cse_var_1 + (threadIdx_x_2 + 2) % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 112) // 48 * 48 + (threadIdx_x_2 + 16) % 48 // 3 * 3 + (threadIdx_x_2 + 1) % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 112) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 + 16) % 48 // 3 * 9 + cse_var_1 + (threadIdx_x_2 + 1) % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 168) // 48 * 48 + (threadIdx_x_2 // 3 + 8) % 16 * 3 + threadIdx_x_2 % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 168) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 // 3 + 8) % 16 * 9 + cse_var_1 + threadIdx_x_2 % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 224) // 48 * 48 + (threadIdx_x_2 + 32) % 48 // 3 * 3 + (threadIdx_x_2 + 2) % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 224) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 + 32) % 48 // 3 * 9 + cse_var_1 + (threadIdx_x_2 + 2) % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 280) // 48 * 48 + (threadIdx_x_2 + 40) % 48 // 3 * 3 + (threadIdx_x_2 + 1) % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 280) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 + 40) % 48 // 3 * 9 + cse_var_1 + (threadIdx_x_2 + 1) % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 336] = kernel_1[blockIdx_x * 147456 + threadIdx_x_2 // 48 * 4608 + cse_var_2 + threadIdx_x_2 % 48 // 3 * 9 + cse_var_1 + threadIdx_x_2 % 3 + 32256]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 392) // 48 * 48 + (threadIdx_x_2 + 8) % 48 // 3 * 3 + (threadIdx_x_2 + 2) % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 392) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 + 8) % 48 // 3 * 9 + cse_var_1 + (threadIdx_x_2 + 2) % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 448) // 48 * 48 + (threadIdx_x_2 + 16) % 48 // 3 * 3 + (threadIdx_x_2 + 1) % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 448) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 + 16) % 48 // 3 * 9 + cse_var_1 + (threadIdx_x_2 + 1) % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 504) // 48 * 48 + (threadIdx_x_2 // 3 + 8) % 16 * 3 + threadIdx_x_2 % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 504) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 // 3 + 8) % 16 * 9 + cse_var_1 + threadIdx_x_2 % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 560) // 48 * 48 + (threadIdx_x_2 + 32) % 48 // 3 * 3 + (threadIdx_x_2 + 2) % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 560) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 + 32) % 48 // 3 * 9 + cse_var_1 + (threadIdx_x_2 + 2) % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 616) // 48 * 48 + (threadIdx_x_2 + 40) % 48 // 3 * 3 + (threadIdx_x_2 + 1) % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 616) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 + 40) % 48 // 3 * 9 + cse_var_1 + (threadIdx_x_2 + 1) % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 672] = kernel_1[blockIdx_x * 147456 + threadIdx_x_2 // 48 * 4608 + cse_var_2 + threadIdx_x_2 % 48 // 3 * 9 + cse_var_1 + threadIdx_x_2 % 3 + 64512]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 728) // 48 * 48 + (threadIdx_x_2 + 8) % 48 // 3 * 3 + (threadIdx_x_2 + 2) % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 728) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 + 8) % 48 // 3 * 9 + cse_var_1 + (threadIdx_x_2 + 2) % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 784) // 48 * 48 + (threadIdx_x_2 + 16) % 48 // 3 * 3 + (threadIdx_x_2 + 1) % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 784) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 + 16) % 48 // 3 * 9 + cse_var_1 + (threadIdx_x_2 + 1) % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 840) // 48 * 48 + (threadIdx_x_2 // 3 + 8) % 16 * 3 + threadIdx_x_2 % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 840) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 // 3 + 8) % 16 * 9 + cse_var_1 + threadIdx_x_2 % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 896) // 48 * 48 + (threadIdx_x_2 + 32) % 48 // 3 * 3 + (threadIdx_x_2 + 2) % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 896) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 + 32) % 48 // 3 * 9 + cse_var_1 + (threadIdx_x_2 + 2) % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 952) // 48 * 48 + (threadIdx_x_2 + 40) % 48 // 3 * 3 + (threadIdx_x_2 + 1) % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 952) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 + 40) % 48 // 3 * 9 + cse_var_1 + (threadIdx_x_2 + 1) % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1008] = kernel_1[blockIdx_x * 147456 + threadIdx_x_2 // 48 * 4608 + cse_var_2 + threadIdx_x_2 % 48 // 3 * 9 + cse_var_1 + threadIdx_x_2 % 3 + 96768]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 1064) // 48 * 48 + (threadIdx_x_2 + 8) % 48 // 3 * 3 + (threadIdx_x_2 + 2) % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 1064) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 + 8) % 48 // 3 * 9 + cse_var_1 + (threadIdx_x_2 + 2) % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 1120) // 48 * 48 + (threadIdx_x_2 + 16) % 48 // 3 * 3 + (threadIdx_x_2 + 1) % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 1120) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 + 16) % 48 // 3 * 9 + cse_var_1 + (threadIdx_x_2 + 1) % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 1176) // 48 * 48 + (threadIdx_x_2 // 3 + 8) % 16 * 3 + threadIdx_x_2 % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 1176) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 // 3 + 8) % 16 * 9 + cse_var_1 + threadIdx_x_2 % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 1232) // 48 * 48 + (threadIdx_x_2 + 32) % 48 // 3 * 3 + (threadIdx_x_2 + 2) % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 1232) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 + 32) % 48 // 3 * 9 + cse_var_1 + (threadIdx_x_2 + 2) % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 1288) // 48 * 48 + (threadIdx_x_2 + 40) % 48 // 3 * 3 + (threadIdx_x_2 + 1) % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 1288) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 + 40) % 48 // 3 * 9 + cse_var_1 + (threadIdx_x_2 + 1) % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1344] = kernel_1[blockIdx_x * 147456 + threadIdx_x_2 // 48 * 4608 + cse_var_2 + threadIdx_x_2 % 48 // 3 * 9 + cse_var_1 + threadIdx_x_2 % 3 + 129024]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 1400) // 48 * 48 + (threadIdx_x_2 + 8) % 48 // 3 * 3 + (threadIdx_x_2 + 2) % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 1400) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 + 8) % 48 // 3 * 9 + cse_var_1 + (threadIdx_x_2 + 2) % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 1456) // 48 * 48 + (threadIdx_x_2 + 16) % 48 // 3 * 3 + (threadIdx_x_2 + 1) % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 1456) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 + 16) % 48 // 3 * 9 + cse_var_1 + (threadIdx_x_2 + 1) % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ if T.likely(threadIdx_x_2 < 24):
+ kernel_shared_1[(threadIdx_x_2 + 1512) // 48 * 48 + threadIdx_x_2 + 24] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 1512) // 48 * 4608 + cse_var_2 + threadIdx_x_2 // 3 * 9 + cse_var_1 + threadIdx_x_2 % 3 + 72]
+ for rc_outer_inner, rx_outer_inner in T.grid(2, 3):
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 9] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 18] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 27] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 36] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 45] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 54] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 63] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 3]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 72] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 3]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 81] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 3]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 90] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 3]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 99] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 3]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 108] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 3]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 117] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 3]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 126] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 6]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 135] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 6]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 144] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 6]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 153] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 6]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 162] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 6]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 171] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 6]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 180] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 6]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 189] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 9]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 198] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 9]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 207] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 9]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 216] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 9]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 225] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 9]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 234] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 9]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 243] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 9]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 252] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 12]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 261] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 12]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 270] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 12]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 279] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 12]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 288] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 12]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 297] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 12]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 306] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 12]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 315] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 15]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 324] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 15]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 333] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 15]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 342] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 15]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 351] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 15]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 360] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 15]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 369] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 15]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 378] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 18]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 387] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 18]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 396] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 18]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 405] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 18]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 414] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 18]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 423] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 18]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 432] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 18]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 441] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 21]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 450] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 21]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 459] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 21]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 468] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 21]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 477] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 21]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 486] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 21]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 495] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 21]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 48]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 9] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 48]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 18] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 48]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 27] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 48]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 36] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 48]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 45] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 48]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 54] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 48]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 63] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 51]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 72] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 51]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 81] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 51]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 90] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 51]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 99] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 51]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 108] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 51]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 117] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 51]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 126] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 54]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 135] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 54]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 144] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 54]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 153] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 54]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 162] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 54]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 171] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 54]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 180] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 54]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 189] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 57]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 198] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 57]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 207] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 57]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 216] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 57]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 225] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 57]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 234] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 57]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 243] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 57]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 252] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 60]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 261] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 60]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 270] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 60]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 279] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 60]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 288] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 60]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 297] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 60]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 306] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 60]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 315] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 63]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 324] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 63]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 333] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 63]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 342] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 63]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 351] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 63]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 360] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 63]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 369] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 63]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 378] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 66]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 387] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 66]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 396] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 66]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 405] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 66]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 414] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 66]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 423] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 66]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 432] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 66]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 441] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 69]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 450] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 69]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 459] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 69]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 468] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 69]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 477] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 69]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 486] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 69]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 495] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 69]
+ conv2d_nchw_1[14] = conv2d_nchw_1[14] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 96]
+ conv2d_nchw_1[15] = conv2d_nchw_1[15] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 9] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 96]
+ conv2d_nchw_1[16] = conv2d_nchw_1[16] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 18] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 96]
+ conv2d_nchw_1[17] = conv2d_nchw_1[17] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 27] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 96]
+ conv2d_nchw_1[18] = conv2d_nchw_1[18] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 36] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 96]
+ conv2d_nchw_1[19] = conv2d_nchw_1[19] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 45] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 96]
+ conv2d_nchw_1[20] = conv2d_nchw_1[20] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 54] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 96]
+ conv2d_nchw_1[14] = conv2d_nchw_1[14] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 63] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 99]
+ conv2d_nchw_1[15] = conv2d_nchw_1[15] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 72] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 99]
+ conv2d_nchw_1[16] = conv2d_nchw_1[16] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 81] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 99]
+ conv2d_nchw_1[17] = conv2d_nchw_1[17] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 90] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 99]
+ conv2d_nchw_1[18] = conv2d_nchw_1[18] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 99] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 99]
+ conv2d_nchw_1[19] = conv2d_nchw_1[19] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 108] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 99]
+ conv2d_nchw_1[20] = conv2d_nchw_1[20] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 117] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 99]
+ conv2d_nchw_1[14] = conv2d_nchw_1[14] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 126] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 102]
+ conv2d_nchw_1[15] = conv2d_nchw_1[15] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 135] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 102]
+ conv2d_nchw_1[16] = conv2d_nchw_1[16] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 144] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 102]
+ conv2d_nchw_1[17] = conv2d_nchw_1[17] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 153] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 102]
+ conv2d_nchw_1[18] = conv2d_nchw_1[18] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 162] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 102]
+ conv2d_nchw_1[19] = conv2d_nchw_1[19] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 171] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 102]
+ conv2d_nchw_1[20] = conv2d_nchw_1[20] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 180] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 102]
+ conv2d_nchw_1[14] = conv2d_nchw_1[14] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 189] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 105]
+ conv2d_nchw_1[15] = conv2d_nchw_1[15] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 198] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 105]
+ conv2d_nchw_1[16] = conv2d_nchw_1[16] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 207] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 105]
+ conv2d_nchw_1[17] = conv2d_nchw_1[17] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 216] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 105]
+ conv2d_nchw_1[18] = conv2d_nchw_1[18] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 225] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 105]
+ conv2d_nchw_1[19] = conv2d_nchw_1[19] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 234] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 105]
+ conv2d_nchw_1[20] = conv2d_nchw_1[20] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 243] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 105]
+ conv2d_nchw_1[14] = conv2d_nchw_1[14] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 252] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 108]
+ conv2d_nchw_1[15] = conv2d_nchw_1[15] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 261] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 108]
+ conv2d_nchw_1[16] = conv2d_nchw_1[16] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 270] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 108]
+ conv2d_nchw_1[17] = conv2d_nchw_1[17] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 279] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 108]
+ conv2d_nchw_1[18] = conv2d_nchw_1[18] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 288] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 108]
+ conv2d_nchw_1[19] = conv2d_nchw_1[19] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 297] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 108]
+ conv2d_nchw_1[20] = conv2d_nchw_1[20] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 306] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 108]
+ conv2d_nchw_1[14] = conv2d_nchw_1[14] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 315] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 111]
+ conv2d_nchw_1[15] = conv2d_nchw_1[15] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 324] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 111]
+ conv2d_nchw_1[16] = conv2d_nchw_1[16] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 333] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 111]
+ conv2d_nchw_1[17] = conv2d_nchw_1[17] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 342] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 111]
+ conv2d_nchw_1[18] = conv2d_nchw_1[18] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 351] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 111]
+ conv2d_nchw_1[19] = conv2d_nchw_1[19] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 360] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 111]
+ conv2d_nchw_1[20] = conv2d_nchw_1[20] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 369] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 111]
+ conv2d_nchw_1[14] = conv2d_nchw_1[14] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 378] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 114]
+ conv2d_nchw_1[15] = conv2d_nchw_1[15] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 387] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 114]
+ conv2d_nchw_1[16] = conv2d_nchw_1[16] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 396] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 114]
+ conv2d_nchw_1[17] = conv2d_nchw_1[17] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 405] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 114]
+ conv2d_nchw_1[18] = conv2d_nchw_1[18] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 414] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 114]
+ conv2d_nchw_1[19] = conv2d_nchw_1[19] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 423] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 114]
+ conv2d_nchw_1[20] = conv2d_nchw_1[20] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 432] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 114]
+ conv2d_nchw_1[14] = conv2d_nchw_1[14] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 441] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 117]
+ conv2d_nchw_1[15] = conv2d_nchw_1[15] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 450] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 117]
+ conv2d_nchw_1[16] = conv2d_nchw_1[16] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 459] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 117]
+ conv2d_nchw_1[17] = conv2d_nchw_1[17] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 468] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 117]
+ conv2d_nchw_1[18] = conv2d_nchw_1[18] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 477] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 117]
+ conv2d_nchw_1[19] = conv2d_nchw_1[19] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 486] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 117]
+ conv2d_nchw_1[20] = conv2d_nchw_1[20] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 495] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 117]
+ conv2d_nchw_1[21] = conv2d_nchw_1[21] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 144]
+ conv2d_nchw_1[22] = conv2d_nchw_1[22] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 9] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 144]
+ conv2d_nchw_1[23] = conv2d_nchw_1[23] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 18] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 144]
+ conv2d_nchw_1[24] = conv2d_nchw_1[24] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 27] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 144]
+ conv2d_nchw_1[25] = conv2d_nchw_1[25] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 36] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 144]
+ conv2d_nchw_1[26] = conv2d_nchw_1[26] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 45] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 144]
+ conv2d_nchw_1[27] = conv2d_nchw_1[27] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 54] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 144]
+ conv2d_nchw_1[21] = conv2d_nchw_1[21] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 63] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 147]
+ conv2d_nchw_1[22] = conv2d_nchw_1[22] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 72] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 147]
+ conv2d_nchw_1[23] = conv2d_nchw_1[23] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 81] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 147]
+ conv2d_nchw_1[24] = conv2d_nchw_1[24] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 90] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 147]
+ conv2d_nchw_1[25] = conv2d_nchw_1[25] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 99] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 147]
+ conv2d_nchw_1[26] = conv2d_nchw_1[26] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 108] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 147]
+ conv2d_nchw_1[27] = conv2d_nchw_1[27] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 117] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 147]
+ conv2d_nchw_1[21] = conv2d_nchw_1[21] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 126] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 150]
+ conv2d_nchw_1[22] = conv2d_nchw_1[22] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 135] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 150]
+ conv2d_nchw_1[23] = conv2d_nchw_1[23] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 144] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 150]
+ conv2d_nchw_1[24] = conv2d_nchw_1[24] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 153] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 150]
+ conv2d_nchw_1[25] = conv2d_nchw_1[25] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 162] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 150]
+ conv2d_nchw_1[26] = conv2d_nchw_1[26] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 171] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 150]
+ conv2d_nchw_1[27] = conv2d_nchw_1[27] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 180] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 150]
+ conv2d_nchw_1[21] = conv2d_nchw_1[21] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 189] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 153]
+ conv2d_nchw_1[22] = conv2d_nchw_1[22] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 198] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 153]
+ conv2d_nchw_1[23] = conv2d_nchw_1[23] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 207] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 153]
+ conv2d_nchw_1[24] = conv2d_nchw_1[24] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 216] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 153]
+ conv2d_nchw_1[25] = conv2d_nchw_1[25] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 225] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 153]
+ conv2d_nchw_1[26] = conv2d_nchw_1[26] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 234] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 153]
+ conv2d_nchw_1[27] = conv2d_nchw_1[27] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 243] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 153]
+ conv2d_nchw_1[21] = conv2d_nchw_1[21] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 252] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 156]
+ conv2d_nchw_1[22] = conv2d_nchw_1[22] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 261] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 156]
+ conv2d_nchw_1[23] = conv2d_nchw_1[23] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 270] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 156]
+ conv2d_nchw_1[24] = conv2d_nchw_1[24] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 279] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 156]
+ conv2d_nchw_1[25] = conv2d_nchw_1[25] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 288] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 156]
+ conv2d_nchw_1[26] = conv2d_nchw_1[26] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 297] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 156]
+ conv2d_nchw_1[27] = conv2d_nchw_1[27] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 306] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 156]
+ conv2d_nchw_1[21] = conv2d_nchw_1[21] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 315] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 159]
+ conv2d_nchw_1[22] = conv2d_nchw_1[22] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 324] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 159]
+ conv2d_nchw_1[23] = conv2d_nchw_1[23] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 333] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 159]
+ conv2d_nchw_1[24] = conv2d_nchw_1[24] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 342] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 159]
+ conv2d_nchw_1[25] = conv2d_nchw_1[25] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 351] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 159]
+ conv2d_nchw_1[26] = conv2d_nchw_1[26] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 360] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 159]
+ conv2d_nchw_1[27] = conv2d_nchw_1[27] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 369] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 159]
+ conv2d_nchw_1[21] = conv2d_nchw_1[21] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 378] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 162]
+ conv2d_nchw_1[22] = conv2d_nchw_1[22] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 387] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 162]
+ conv2d_nchw_1[23] = conv2d_nchw_1[23] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 396] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 162]
+ conv2d_nchw_1[24] = conv2d_nchw_1[24] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 405] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 162]
+ conv2d_nchw_1[25] = conv2d_nchw_1[25] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 414] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 162]
+ conv2d_nchw_1[26] = conv2d_nchw_1[26] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 423] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 162]
+ conv2d_nchw_1[27] = conv2d_nchw_1[27] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 432] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 162]
+ conv2d_nchw_1[21] = conv2d_nchw_1[21] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 441] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 165]
+ conv2d_nchw_1[22] = conv2d_nchw_1[22] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 450] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 165]
+ conv2d_nchw_1[23] = conv2d_nchw_1[23] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 459] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 165]
+ conv2d_nchw_1[24] = conv2d_nchw_1[24] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 468] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 165]
+ conv2d_nchw_1[25] = conv2d_nchw_1[25] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 477] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 165]
+ conv2d_nchw_1[26] = conv2d_nchw_1[26] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 486] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 165]
+ conv2d_nchw_1[27] = conv2d_nchw_1[27] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 495] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 165]
+ for i1_inner, i2_inner in T.grid(4, 7):
+ compute_1 = T.Buffer((25088,), data=compute.data)
+ bias_1 = T.Buffer((512,), data=bias.data)
+ compute_1[blockIdx_x * 1568 + threadIdx_x // 7 * 196 + i1_inner * 49 + i2_inner * 7 + threadIdx_x % 7] = T.max(conv2d_nchw_1[i1_inner * 7 + i2_inner] + bias_1[blockIdx_x * 32 + threadIdx_x // 7 * 4 + i1_inner], T.float32(0))
@@ -1393,7 +659,7 @@ We build the binary and check its correctness and performance.
.. code-block:: none
- Execution time of this operator: 0.388 ms
+ Execution time of this operator: 0.281 ms
@@ -1442,33 +708,33 @@ They can be used for debugging and learning the behavior of the auto-scheduler.
conv2d_nchw_nn_o_o_o_i, conv2d_nchw_nn_o_o_i = s[conv2d_nchw].split(conv2d_nchw_nn_o_o_i, factor=1)
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=1)
- 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_i, conv2d_nchw_ff_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_i, factor=4)
+ 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_i, conv2d_nchw_yy_i = s[conv2d_nchw].split(conv2d_nchw_yy, factor=7)
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=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_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_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=4)
- conv2d_nchw_rc_o_o, conv2d_nchw_rc_o_i = s[conv2d_nchw].split(conv2d_nchw_rc_o_i, factor=32)
+ 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=2)
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_rx_o_o, conv2d_nchw_rx_o_i = s[conv2d_nchw].split(conv2d_nchw_rx_o_i, factor=1)
+ conv2d_nchw_rx_o_o, conv2d_nchw_rx_o_i = s[conv2d_nchw].split(conv2d_nchw_rx_o_i, factor=3)
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=1)
- compute_i1_o_o_i, compute_i1_o_i = s[compute].split(compute_i1_o_i, factor=16)
+ compute_i1_o_i, compute_i1_i = s[compute].split(compute_i1, factor=4)
+ 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_i, compute_i2_i = s[compute].split(compute_i2, factor=7)
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_i2_o_o_o, compute_i2_o_o_i = s[compute].split(compute_i2_o_o_i, factor=1)
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)
@@ -1488,16 +754,16 @@ They can be used for debugging and learning the behavior of the auto-scheduler.
compute_i0_o_i_i1_o_i_fused_i2_o_i_fused_i3_o_i_fused = s[compute].fuse(compute_i0_o_i, compute_i1_o_i, compute_i2_o_i, compute_i3_o_i)
s[compute].bind(compute_i0_o_i_i1_o_i_fused_i2_o_i_fused_i3_o_i_fused, te.thread_axis("threadIdx.x"))
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=32)
+ 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=112)
+ 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=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=112)
+ 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", 1024)
+ s[conv2d_nchw].pragma(conv2d_nchw_nn_o_o_o_o, "auto_unroll_max_step", 512)
s[conv2d_nchw].pragma(conv2d_nchw_nn_o_o_o_o, "unroll_explicit", True)
CUDA source code:
@@ -1522,10 +788,10 @@ 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__(112) default_function_kernel0(float* __restrict__ data, float* __restrict__ kernel, float* __restrict__ compute, float* __restrict__ bias) {
- float conv2d_nchw[7];
- __shared__ float pad_temp_shared[6272];
- __shared__ float kernel_shared[2048];
+ 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[28];
+ __shared__ float pad_temp_shared[1008];
+ __shared__ float kernel_shared[1536];
conv2d_nchw[0] = 0.000000e+00f;
conv2d_nchw[1] = 0.000000e+00f;
conv2d_nchw[2] = 0.000000e+00f;
@@ -1533,1069 +799,314 @@ They can be used for debugging and learning the behavior of the auto-scheduler.
conv2d_nchw[4] = 0.000000e+00f;
conv2d_nchw[5] = 0.000000e+00f;
conv2d_nchw[6] = 0.000000e+00f;
- for (int rc_outer_outer = 0; rc_outer_outer < 4; ++rc_outer_outer) {
+ conv2d_nchw[7] = 0.000000e+00f;
+ conv2d_nchw[8] = 0.000000e+00f;
+ conv2d_nchw[9] = 0.000000e+00f;
+ conv2d_nchw[10] = 0.000000e+00f;
+ conv2d_nchw[11] = 0.000000e+00f;
+ conv2d_nchw[12] = 0.000000e+00f;
+ conv2d_nchw[13] = 0.000000e+00f;
+ conv2d_nchw[14] = 0.000000e+00f;
+ conv2d_nchw[15] = 0.000000e+00f;
+ conv2d_nchw[16] = 0.000000e+00f;
+ conv2d_nchw[17] = 0.000000e+00f;
+ conv2d_nchw[18] = 0.000000e+00f;
+ conv2d_nchw[19] = 0.000000e+00f;
+ conv2d_nchw[20] = 0.000000e+00f;
+ conv2d_nchw[21] = 0.000000e+00f;
+ conv2d_nchw[22] = 0.000000e+00f;
+ conv2d_nchw[23] = 0.000000e+00f;
+ conv2d_nchw[24] = 0.000000e+00f;
+ conv2d_nchw[25] = 0.000000e+00f;
+ conv2d_nchw[26] = 0.000000e+00f;
+ conv2d_nchw[27] = 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();
- pad_temp_shared[((int)threadIdx.x)] = (((((1 <= (((((int)threadIdx.x) % 49) / 7) + ry_outer_outer)) && ((((((int)threadIdx.x) % 49) / 7) + ry_outer_outer) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) - 8)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 112)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 104)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 224)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 216)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 336)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 328)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 448)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 440)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 560)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 552)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 672)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 664)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 784)] = (((((1 <= (((((int)threadIdx.x) % 49) / 7) + ry_outer_outer)) && ((((((int)threadIdx.x) % 49) / 7) + ry_outer_outer) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 776)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 896)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 888)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 1008)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 1000)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 1120)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 1112)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 1232)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 1224)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 1344)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 1336)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 1456)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 1448)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 1568)] = (((((1 <= (((((int)threadIdx.x) % 49) / 7) + ry_outer_outer)) && ((((((int)threadIdx.x) % 49) / 7) + ry_outer_outer) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 1560)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 1680)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 1672)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 1792)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 1784)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 1904)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 1896)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 2016)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 2008)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 2128)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 2120)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 2240)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 2232)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 2352)] = (((((1 <= (((((int)threadIdx.x) % 49) / 7) + ry_outer_outer)) && ((((((int)threadIdx.x) % 49) / 7) + ry_outer_outer) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 2344)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 2464)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 2456)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 2576)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 2568)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 2688)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 2680)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 2800)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 2792)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 2912)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 2904)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 3024)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 3016)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 3136)] = (((((1 <= (((((int)threadIdx.x) % 49) / 7) + ry_outer_outer)) && ((((((int)threadIdx.x) % 49) / 7) + ry_outer_outer) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 3128)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 3248)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 3240)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 3360)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 3352)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 3472)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 3464)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 3584)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 3576)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 3696)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 3688)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 3808)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 3800)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 3920)] = (((((1 <= (((((int)threadIdx.x) % 49) / 7) + ry_outer_outer)) && ((((((int)threadIdx.x) % 49) / 7) + ry_outer_outer) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 3912)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 4032)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 4024)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 4144)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 4136)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 4256)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 4248)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 4368)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 4360)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 4480)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 4472)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 4592)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 4584)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 4704)] = (((((1 <= (((((int)threadIdx.x) % 49) / 7) + ry_outer_outer)) && ((((((int)threadIdx.x) % 49) / 7) + ry_outer_outer) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 4696)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 4816)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 4808)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 4928)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 4920)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 5040)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 5032)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 5152)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 5144)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 5264)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 5256)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 5376)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 5368)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 5488)] = (((((1 <= (((((int)threadIdx.x) % 49) / 7) + ry_outer_outer)) && ((((((int)threadIdx.x) % 49) / 7) + ry_outer_outer) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 5480)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 5600)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 5592)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 5712)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 5704)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 5824)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 5816)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 5936)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 5928)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 6048)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 6040)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 6160)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 6152)] : 0.000000e+00f);
- if (((int)threadIdx.x) < 64) {
- kernel_shared[(((int)threadIdx.x) * 32)] = kernel[((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 1)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 9)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 2)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 18)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 3)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 27)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 4)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 36)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 5)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 45)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 6)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 54)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 7)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 63)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 8)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 72)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 9)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 81)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 10)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 90)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 11)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 99)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 12)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 108)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 13)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 117)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 14)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 126)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 15)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 135)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 16)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 144)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 17)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 153)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 18)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 162)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 19)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 171)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 20)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 180)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 21)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 189)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 22)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 198)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 23)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 207)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 24)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 216)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 25)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 225)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 26)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 234)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 27)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 243)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 28)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 252)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 29)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 261)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 30)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 270)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 31)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 279)];
+ __syncthreads();
+ pad_temp_shared[((int)threadIdx.x)] = (((((1 <= ((((int)threadIdx.x) / 9) + ry_outer_outer)) && (((((int)threadIdx.x) / 9) + ry_outer_outer) < 8)) && (1 <= (((int)threadIdx.x) % 9))) && ((((int)threadIdx.x) % 9) < 8)) ? data[(((((rc_outer_outer * 784) + ((((int)threadIdx.x) / 9) * 7)) + (ry_outer_outer * 7)) + (((int)threadIdx.x) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 56)] = (((((1 <= ((((((int)threadIdx.x) + 56) % 63) / 9) + ry_outer_outer)) && (((((((int)threadIdx.x) + 56) % 63) / 9) + ry_outer_outer) < 8)) && (1 <= ((((int)threadIdx.x) + 2) % 9))) && (((((int)threadIdx.x) + 2) % 9) < 8)) ? data[(((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 56) / 9) * 7)) + (ry_outer_outer * 7)) + ((((int)threadIdx.x) + 2) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 112)] = (((((1 <= ((((((int)threadIdx.x) + 49) % 63) / 9) + ry_outer_outer)) && (((((((int)threadIdx.x) + 49) % 63) / 9) + ry_outer_outer) < 8)) && (1 <= ((((int)threadIdx.x) + 4) % 9))) && (((((int)threadIdx.x) + 4) % 9) < 8)) ? data[(((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 112) / 9) * 7)) + (ry_outer_outer * 7)) + ((((int)threadIdx.x) + 4) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 168)] = (((((1 <= ((((((int)threadIdx.x) + 42) % 63) / 9) + ry_outer_outer)) && (((((((int)threadIdx.x) + 42) % 63) / 9) + ry_outer_outer) < 8)) && (1 <= ((((int)threadIdx.x) + 6) % 9))) && (((((int)threadIdx.x) + 6) % 9) < 8)) ? data[(((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 168) / 9) * 7)) + (ry_outer_outer * 7)) + ((((int)threadIdx.x) + 6) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 224)] = (((((1 <= ((((((int)threadIdx.x) + 35) % 63) / 9) + ry_outer_outer)) && (((((((int)threadIdx.x) + 35) % 63) / 9) + ry_outer_outer) < 8)) && (1 <= ((((int)threadIdx.x) + 8) % 9))) && (((((int)threadIdx.x) + 8) % 9) < 8)) ? data[(((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 224) / 9) * 7)) + (ry_outer_outer * 7)) + ((((int)threadIdx.x) + 8) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 280)] = (((((1 <= ((((((int)threadIdx.x) + 28) % 63) / 9) + ry_outer_outer)) && (((((((int)threadIdx.x) + 28) % 63) / 9) + ry_outer_outer) < 8)) && (1 <= ((((int)threadIdx.x) + 1) % 9))) && (((((int)threadIdx.x) + 1) % 9) < 8)) ? data[(((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 280) / 9) * 7)) + (ry_outer_outer * 7)) + ((((int)threadIdx.x) + 1) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 336)] = (((((1 <= ((((((int)threadIdx.x) + 21) % 63) / 9) + ry_outer_outer)) && (((((((int)threadIdx.x) + 21) % 63) / 9) + ry_outer_outer) < 8)) && (1 <= ((((int)threadIdx.x) + 3) % 9))) && (((((int)threadIdx.x) + 3) % 9) < 8)) ? data[(((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 336) / 9) * 7)) + (ry_outer_outer * 7)) + ((((int)threadIdx.x) + 3) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 392)] = (((((1 <= ((((((int)threadIdx.x) + 14) % 63) / 9) + ry_outer_outer)) && (((((((int)threadIdx.x) + 14) % 63) / 9) + ry_outer_outer) < 8)) && (1 <= ((((int)threadIdx.x) + 5) % 9))) && (((((int)threadIdx.x) + 5) % 9) < 8)) ? data[(((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 392) / 9) * 7)) + (ry_outer_outer * 7)) + ((((int)threadIdx.x) + 5) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 448)] = (((((1 <= (((((int)threadIdx.x) + 7) / 9) + ry_outer_outer)) && ((((((int)threadIdx.x) + 7) / 9) + ry_outer_outer) < 8)) && (1 <= ((((int)threadIdx.x) + 7) % 9))) && (((((int)threadIdx.x) + 7) % 9) < 8)) ? data[(((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 448) / 9) * 7)) + (ry_outer_outer * 7)) + ((((int)threadIdx.x) + 7) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 504)] = (((((1 <= ((((int)threadIdx.x) / 9) + ry_outer_outer)) && (((((int)threadIdx.x) / 9) + ry_outer_outer) < 8)) && (1 <= (((int)threadIdx.x) % 9))) && ((((int)threadIdx.x) % 9) < 8)) ? data[(((((rc_outer_outer * 784) + ((((int)threadIdx.x) / 9) * 7)) + (ry_outer_outer * 7)) + (((int)threadIdx.x) % 9)) + 384)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 560)] = (((((1 <= ((((((int)threadIdx.x) + 56) % 63) / 9) + ry_outer_outer)) && (((((((int)threadIdx.x) + 56) % 63) / 9) + ry_outer_outer) < 8)) && (1 <= ((((int)threadIdx.x) + 2) % 9))) && (((((int)threadIdx.x) + 2) % 9) < 8)) ? data[(((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 560) / 9) * 7)) + (ry_outer_outer * 7)) + ((((int)threadIdx.x) + 2) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 616)] = (((((1 <= ((((((int)threadIdx.x) + 49) % 63) / 9) + ry_outer_outer)) && (((((((int)threadIdx.x) + 49) % 63) / 9) + ry_outer_outer) < 8)) && (1 <= ((((int)threadIdx.x) + 4) % 9))) && (((((int)threadIdx.x) + 4) % 9) < 8)) ? data[(((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 616) / 9) * 7)) + (ry_outer_outer * 7)) + ((((int)threadIdx.x) + 4) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 672)] = (((((1 <= ((((((int)threadIdx.x) + 42) % 63) / 9) + ry_outer_outer)) && (((((((int)threadIdx.x) + 42) % 63) / 9) + ry_outer_outer) < 8)) && (1 <= ((((int)threadIdx.x) + 6) % 9))) && (((((int)threadIdx.x) + 6) % 9) < 8)) ? data[(((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 672) / 9) * 7)) + (ry_outer_outer * 7)) + ((((int)threadIdx.x) + 6) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 728)] = (((((1 <= ((((((int)threadIdx.x) + 35) % 63) / 9) + ry_outer_outer)) && (((((((int)threadIdx.x) + 35) % 63) / 9) + ry_outer_outer) < 8)) && (1 <= ((((int)threadIdx.x) + 8) % 9))) && (((((int)threadIdx.x) + 8) % 9) < 8)) ? data[(((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 728) / 9) * 7)) + (ry_outer_outer * 7)) + ((((int)threadIdx.x) + 8) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 784)] = (((((1 <= ((((((int)threadIdx.x) + 28) % 63) / 9) + ry_outer_outer)) && (((((((int)threadIdx.x) + 28) % 63) / 9) + ry_outer_outer) < 8)) && (1 <= ((((int)threadIdx.x) + 1) % 9))) && (((((int)threadIdx.x) + 1) % 9) < 8)) ? data[(((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 784) / 9) * 7)) + (ry_outer_outer * 7)) + ((((int)threadIdx.x) + 1) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 840)] = (((((1 <= ((((((int)threadIdx.x) + 21) % 63) / 9) + ry_outer_outer)) && (((((((int)threadIdx.x) + 21) % 63) / 9) + ry_outer_outer) < 8)) && (1 <= ((((int)threadIdx.x) + 3) % 9))) && (((((int)threadIdx.x) + 3) % 9) < 8)) ? data[(((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 840) / 9) * 7)) + (ry_outer_outer * 7)) + ((((int)threadIdx.x) + 3) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 896)] = (((((1 <= ((((((int)threadIdx.x) + 14) % 63) / 9) + ry_outer_outer)) && (((((((int)threadIdx.x) + 14) % 63) / 9) + ry_outer_outer) < 8)) && (1 <= ((((int)threadIdx.x) + 5) % 9))) && (((((int)threadIdx.x) + 5) % 9) < 8)) ? data[(((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 896) / 9) * 7)) + (ry_outer_outer * 7)) + ((((int)threadIdx.x) + 5) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 952)] = (((((1 <= (((((int)threadIdx.x) + 7) / 9) + ry_outer_outer)) && ((((((int)threadIdx.x) + 7) / 9) + ry_outer_outer) < 8)) && (1 <= ((((int)threadIdx.x) + 7) % 9))) && (((((int)threadIdx.x) + 7) % 9) < 8)) ? data[(((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 952) / 9) * 7)) + (ry_outer_outer * 7)) + ((((int)threadIdx.x) + 7) % 9)) - 8)] : 0.000000e+00f);
+ kernel_shared[((int)threadIdx.x)] = kernel[((((((((int)blockIdx.x) * 147456) + ((((int)threadIdx.x) / 48) * 4608)) + (rc_outer_outer * 144)) + (((((int)threadIdx.x) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + (((int)threadIdx.x) % 3))];
+ kernel_shared[(((((((int)threadIdx.x) + 56) / 48) * 48) + ((((((int)threadIdx.x) + 8) % 48) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 56) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 8) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + ((((int)threadIdx.x) + 2) % 3))];
+ kernel_shared[(((((((int)threadIdx.x) + 112) / 48) * 48) + ((((((int)threadIdx.x) + 16) % 48) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 112) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 16) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + ((((int)threadIdx.x) + 1) % 3))];
+ kernel_shared[(((((((int)threadIdx.x) + 168) / 48) * 48) + ((((((int)threadIdx.x) / 3) + 8) & 15) * 3)) + (((int)threadIdx.x) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 168) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) / 3) + 8) & 15) * 9)) + (ry_outer_outer * 3)) + (((int)threadIdx.x) % 3))];
+ kernel_shared[(((((((int)threadIdx.x) + 224) / 48) * 48) + ((((((int)threadIdx.x) + 32) % 48) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 224) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 32) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + ((((int)threadIdx.x) + 2) % 3))];
+ kernel_shared[(((((((int)threadIdx.x) + 280) / 48) * 48) + ((((((int)threadIdx.x) + 40) % 48) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 280) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 40) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + ((((int)threadIdx.x) + 1) % 3))];
+ kernel_shared[(((int)threadIdx.x) + 336)] = kernel[(((((((((int)blockIdx.x) * 147456) + ((((int)threadIdx.x) / 48) * 4608)) + (rc_outer_outer * 144)) + (((((int)threadIdx.x) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + (((int)threadIdx.x) % 3)) + 32256)];
+ kernel_shared[(((((((int)threadIdx.x) + 392) / 48) * 48) + ((((((int)threadIdx.x) + 8) % 48) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 392) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 8) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + ((((int)threadIdx.x) + 2) % 3))];
+ kernel_shared[(((((((int)threadIdx.x) + 448) / 48) * 48) + ((((((int)threadIdx.x) + 16) % 48) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 448) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 16) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + ((((int)threadIdx.x) + 1) % 3))];
+ kernel_shared[(((((((int)threadIdx.x) + 504) / 48) * 48) + ((((((int)threadIdx.x) / 3) + 8) & 15) * 3)) + (((int)threadIdx.x) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 504) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) / 3) + 8) & 15) * 9)) + (ry_outer_outer * 3)) + (((int)threadIdx.x) % 3))];
+ kernel_shared[(((((((int)threadIdx.x) + 560) / 48) * 48) + ((((((int)threadIdx.x) + 32) % 48) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 560) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 32) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + ((((int)threadIdx.x) + 2) % 3))];
+ kernel_shared[(((((((int)threadIdx.x) + 616) / 48) * 48) + ((((((int)threadIdx.x) + 40) % 48) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 616) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 40) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + ((((int)threadIdx.x) + 1) % 3))];
+ kernel_shared[(((int)threadIdx.x) + 672)] = kernel[(((((((((int)blockIdx.x) * 147456) + ((((int)threadIdx.x) / 48) * 4608)) + (rc_outer_outer * 144)) + (((((int)threadIdx.x) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + (((int)threadIdx.x) % 3)) + 64512)];
+ kernel_shared[(((((((int)threadIdx.x) + 728) / 48) * 48) + ((((((int)threadIdx.x) + 8) % 48) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 728) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 8) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + ((((int)threadIdx.x) + 2) % 3))];
+ kernel_shared[(((((((int)threadIdx.x) + 784) / 48) * 48) + ((((((int)threadIdx.x) + 16) % 48) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 784) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 16) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + ((((int)threadIdx.x) + 1) % 3))];
+ kernel_shared[(((((((int)threadIdx.x) + 840) / 48) * 48) + ((((((int)threadIdx.x) / 3) + 8) & 15) * 3)) + (((int)threadIdx.x) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 840) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) / 3) + 8) & 15) * 9)) + (ry_outer_outer * 3)) + (((int)threadIdx.x) % 3))];
+ kernel_shared[(((((((int)threadIdx.x) + 896) / 48) * 48) + ((((((int)threadIdx.x) + 32) % 48) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 896) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 32) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + ((((int)threadIdx.x) + 2) % 3))];
+ kernel_shared[(((((((int)threadIdx.x) + 952) / 48) * 48) + ((((((int)threadIdx.x) + 40) % 48) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 952) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 40) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + ((((int)threadIdx.x) + 1) % 3))];
+ kernel_shared[(((int)threadIdx.x) + 1008)] = kernel[(((((((((int)blockIdx.x) * 147456) + ((((int)threadIdx.x) / 48) * 4608)) + (rc_outer_outer * 144)) + (((((int)threadIdx.x) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + (((int)threadIdx.x) % 3)) + 96768)];
+ kernel_shared[(((((((int)threadIdx.x) + 1064) / 48) * 48) + ((((((int)threadIdx.x) + 8) % 48) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 1064) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 8) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + ((((int)threadIdx.x) + 2) % 3))];
+ kernel_shared[(((((((int)threadIdx.x) + 1120) / 48) * 48) + ((((((int)threadIdx.x) + 16) % 48) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 1120) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 16) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + ((((int)threadIdx.x) + 1) % 3))];
+ kernel_shared[(((((((int)threadIdx.x) + 1176) / 48) * 48) + ((((((int)threadIdx.x) / 3) + 8) & 15) * 3)) + (((int)threadIdx.x) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 1176) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) / 3) + 8) & 15) * 9)) + (ry_outer_outer * 3)) + (((int)threadIdx.x) % 3))];
+ kernel_shared[(((((((int)threadIdx.x) + 1232) / 48) * 48) + ((((((int)threadIdx.x) + 32) % 48) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 1232) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 32) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + ((((int)threadIdx.x) + 2) % 3))];
+ kernel_shared[(((((((int)threadIdx.x) + 1288) / 48) * 48) + ((((((int)threadIdx.x) + 40) % 48) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 1288) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 40) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + ((((int)threadIdx.x) + 1) % 3))];
+ kernel_shared[(((int)threadIdx.x) + 1344)] = kernel[(((((((((int)blockIdx.x) * 147456) + ((((int)threadIdx.x) / 48) * 4608)) + (rc_outer_outer * 144)) + (((((int)threadIdx.x) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + (((int)threadIdx.x) % 3)) + 129024)];
+ kernel_shared[(((((((int)threadIdx.x) + 1400) / 48) * 48) + ((((((int)threadIdx.x) + 8) % 48) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 1400) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 8) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + ((((int)threadIdx.x) + 2) % 3))];
+ kernel_shared[(((((((int)threadIdx.x) + 1456) / 48) * 48) + ((((((int)threadIdx.x) + 16) % 48) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 1456) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 16) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + ((((int)threadIdx.x) + 1) % 3))];
+ if (((int)threadIdx.x) < 24) {
+ kernel_shared[(((((((int)threadIdx.x) + 1512) / 48) * 48) + ((int)threadIdx.x)) + 24)] = kernel[(((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 1512) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((int)threadIdx.x) / 3) * 9)) + (ry_outer_outer * 3)) + (((int)threadIdx.x) % 3)) + 72)];
+ }
+ __syncthreads();
+ for (int rc_outer_inner = 0; rc_outer_inner < 2; ++rc_outer_inner) {
+ for (int rx_outer_inner = 0; rx_outer_inner < 3; ++rx_outer_inner) {
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7))] * kernel_shared[((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 3)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 72)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 3)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 81)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 3)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 3)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 3)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 3)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 3)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 6)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 6)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 144)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 6)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 153)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 6)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 162)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 6)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 6)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 6)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 9)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 9)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 9)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 9)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 225)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 9)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 234)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 9)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 243)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 9)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 12)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 12)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 12)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 12)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 12)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 12)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 306)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 12)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 15)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 324)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 15)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 333)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 15)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 342)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 15)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 351)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 15)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 360)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 15)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 369)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 15)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 18)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 387)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 18)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 396)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 18)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 405)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 18)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 414)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 18)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 423)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 18)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 432)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 18)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 21)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 450)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 21)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 459)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 21)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 468)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 21)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 477)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 21)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 486)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 21)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 495)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 21)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7))] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 48)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 48)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 48)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 48)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 48)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 48)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 48)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 51)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 72)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 51)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 81)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 51)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 51)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 51)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 51)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 51)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 54)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 54)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 144)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 54)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 153)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 54)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 162)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 54)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 54)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 54)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 57)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 57)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 57)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 57)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 225)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 57)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 234)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 57)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 243)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 57)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 60)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 60)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 60)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 60)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 60)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 60)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 306)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 60)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 63)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 324)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 63)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 333)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 63)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 342)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 63)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 351)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 63)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 360)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 63)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 369)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 63)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 66)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 387)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 66)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 396)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 66)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 405)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 66)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 414)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 66)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 423)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 66)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 432)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 66)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 69)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 450)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 69)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 459)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 69)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 468)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 69)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 477)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 69)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 486)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 69)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 495)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 69)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7))] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 96)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 96)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 96)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 96)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 96)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 96)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 96)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 99)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 72)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 99)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 81)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 99)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 99)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 99)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 99)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 99)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 102)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 102)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 144)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 102)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 153)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 102)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 162)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 102)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 102)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 102)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 105)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 105)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 105)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 105)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 225)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 105)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 234)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 105)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 243)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 105)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 108)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 108)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 108)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 108)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 108)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 108)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 306)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 108)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 111)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 324)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 111)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 333)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 111)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 342)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 111)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 351)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 111)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 360)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 111)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 369)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 111)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 114)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 387)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 114)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 396)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 114)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 405)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 114)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 414)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 114)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 423)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 114)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 432)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 114)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 117)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 450)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 117)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 459)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 117)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 468)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 117)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 477)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 117)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 486)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 117)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 495)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 117)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7))] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 144)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 144)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 144)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 144)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 144)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 144)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 144)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 147)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 72)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 147)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 81)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 147)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 147)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 147)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 147)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 147)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 150)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 150)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 144)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 150)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 153)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 150)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 162)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 150)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 150)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 150)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 153)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 153)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 153)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 153)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 225)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 153)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 234)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 153)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 243)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 153)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 156)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 156)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 156)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 156)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 156)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 156)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 306)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 156)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 159)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 324)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 159)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 333)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 159)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 342)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 159)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 351)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 159)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 360)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 159)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 369)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 159)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 162)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 387)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 162)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 396)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 162)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 405)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 162)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 414)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 162)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 423)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 162)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 432)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 162)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 165)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 450)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 165)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 459)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 165)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 468)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 165)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 477)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 165)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 486)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 165)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 495)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 165)]));
}
- __syncthreads();
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((int)threadIdx.x) % 7)] * kernel_shared[((((int)threadIdx.x) / 7) * 128)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 7)] * kernel_shared[((((int)threadIdx.x) / 7) * 128)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 14)] * kernel_shared[((((int)threadIdx.x) / 7) * 128)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 21)] * kernel_shared[((((int)threadIdx.x) / 7) * 128)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 28)] * kernel_shared[((((int)threadIdx.x) / 7) * 128)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 35)] * kernel_shared[((((int)threadIdx.x) / 7) * 128)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 42)] * kernel_shared[((((int)threadIdx.x) / 7) * 128)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 49)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 1)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 56)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 1)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 63)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 1)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 70)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 1)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 77)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 1)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 84)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 1)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 91)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 1)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 98)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 2)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 105)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 2)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 112)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 2)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 119)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 2)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 126)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 2)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 133)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 2)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 140)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 2)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 147)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 3)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 154)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 3)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 161)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 3)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 168)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 3)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 175)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 3)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 182)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 3)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 189)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 3)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 196)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 4)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 203)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 4)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 210)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 4)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 217)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 4)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 224)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 4)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 231)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 4)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 238)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 4)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 245)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 5)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 252)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 5)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 259)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 5)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 266)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 5)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 273)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 5)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 280)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 5)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 287)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 5)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 294)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 6)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 301)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 6)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 308)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 6)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 315)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 6)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 322)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 6)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 329)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 6)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 336)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 6)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 343)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 7)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 350)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 7)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 357)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 7)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 364)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 7)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 371)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 7)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 378)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 7)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 385)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 7)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 392)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 8)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 399)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 8)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 406)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 8)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 413)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 8)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 420)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 8)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 427)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 8)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 434)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 8)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 441)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 9)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 448)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 9)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 455)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 9)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 462)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 9)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 469)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 9)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 476)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 9)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 483)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 9)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 490)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 10)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 497)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 10)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 504)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 10)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 511)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 10)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 518)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 10)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 525)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 10)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 532)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 10)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 539)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 11)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 546)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 11)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 553)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 11)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 560)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 11)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 567)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 11)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 574)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 11)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 581)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 11)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 588)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 12)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 595)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 12)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 602)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 12)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 609)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 12)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 616)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 12)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 623)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 12)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 630)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 12)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 637)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 13)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 644)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 13)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 651)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 13)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 658)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 13)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 665)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 13)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 672)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 13)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 679)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 13)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 686)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 14)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 693)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 14)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 700)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 14)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 707)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 14)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 714)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 14)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 721)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 14)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 728)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 14)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 735)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 15)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 742)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 15)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 749)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 15)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 756)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 15)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 763)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 15)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 770)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 15)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 777)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 15)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 784)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 16)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 791)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 16)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 798)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 16)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 805)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 16)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 812)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 16)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 819)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 16)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 826)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 16)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 833)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 17)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 840)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 17)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 847)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 17)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 854)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 17)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 861)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 17)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 868)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 17)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 875)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 17)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 882)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 18)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 889)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 18)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 896)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 18)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 903)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 18)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 910)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 18)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 917)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 18)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 924)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 18)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 931)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 19)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 938)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 19)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 945)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 19)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 952)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 19)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 959)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 19)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 966)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 19)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 973)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 19)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 980)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 20)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 987)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 20)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 994)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 20)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1001)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 20)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1008)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 20)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1015)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 20)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1022)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 20)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1029)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 21)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1036)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 21)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1043)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 21)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1050)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 21)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1057)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 21)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1064)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 21)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1071)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 21)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1078)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 22)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1085)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 22)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1092)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 22)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1099)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 22)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1106)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 22)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1113)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 22)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1120)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 22)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1127)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 23)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1134)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 23)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1141)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 23)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1148)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 23)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1155)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 23)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1162)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 23)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1169)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 23)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1176)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 24)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1183)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 24)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1190)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 24)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1197)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 24)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1204)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 24)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1211)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 24)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1218)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 24)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1225)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 25)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1232)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 25)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1239)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 25)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1246)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 25)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1253)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 25)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1260)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 25)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1267)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 25)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1274)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 26)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1281)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 26)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1288)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 26)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1295)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 26)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1302)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 26)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1309)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 26)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1316)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 26)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1323)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 27)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1330)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 27)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1337)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 27)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1344)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 27)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1351)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 27)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1358)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 27)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1365)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 27)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1372)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 28)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1379)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 28)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1386)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 28)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1393)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 28)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1400)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 28)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1407)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 28)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1414)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 28)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1421)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 29)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1428)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 29)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1435)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 29)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1442)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 29)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1449)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 29)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1456)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 29)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1463)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 29)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1470)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 30)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1477)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 30)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1484)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 30)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1491)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 30)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1498)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 30)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1505)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 30)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1512)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 30)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1519)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 31)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1526)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 31)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1533)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 31)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1540)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 31)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1547)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 31)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1554)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 31)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1561)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 31)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1568)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 32)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1575)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 32)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1582)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 32)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1589)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 32)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1596)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 32)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1603)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 32)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1610)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 32)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1617)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 33)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1624)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 33)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1631)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 33)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1638)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 33)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1645)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 33)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1652)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 33)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1659)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 33)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1666)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 34)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1673)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 34)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1680)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 34)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1687)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 34)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1694)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 34)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1701)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 34)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1708)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 34)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1715)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 35)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1722)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 35)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1729)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 35)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1736)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 35)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1743)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 35)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1750)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 35)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1757)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 35)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1764)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 36)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1771)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 36)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1778)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 36)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1785)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 36)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1792)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 36)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1799)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 36)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1806)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 36)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1813)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 37)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1820)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 37)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1827)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 37)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1834)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 37)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1841)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 37)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1848)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 37)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1855)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 37)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1862)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 38)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1869)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 38)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1876)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 38)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1883)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 38)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1890)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 38)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1897)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 38)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1904)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 38)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1911)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 39)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1918)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 39)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1925)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 39)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1932)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 39)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1939)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 39)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1946)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 39)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1953)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 39)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1960)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 40)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1967)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 40)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1974)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 40)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1981)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 40)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1988)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 40)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1995)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 40)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2002)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 40)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2009)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 41)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2016)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 41)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2023)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 41)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2030)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 41)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2037)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 41)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2044)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 41)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2051)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 41)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2058)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 42)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2065)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 42)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2072)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 42)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2079)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 42)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2086)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 42)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2093)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 42)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2100)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 42)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2107)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 43)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2114)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 43)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2121)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 43)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2128)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 43)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2135)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 43)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2142)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 43)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2149)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 43)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2156)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 44)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2163)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 44)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2170)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 44)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2177)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 44)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2184)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 44)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2191)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 44)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2198)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 44)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2205)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 45)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2212)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 45)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2219)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 45)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2226)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 45)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2233)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 45)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2240)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 45)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2247)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 45)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2254)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 46)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2261)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 46)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2268)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 46)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2275)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 46)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2282)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 46)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2289)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 46)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2296)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 46)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2303)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 47)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2310)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 47)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2317)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 47)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2324)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 47)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2331)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 47)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2338)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 47)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2345)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 47)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2352)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 48)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2359)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 48)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2366)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 48)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2373)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 48)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2380)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 48)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2387)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 48)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2394)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 48)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2401)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 49)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2408)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 49)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2415)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 49)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2422)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 49)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2429)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 49)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2436)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 49)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2443)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 49)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2450)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 50)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2457)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 50)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2464)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 50)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2471)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 50)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2478)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 50)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2485)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 50)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2492)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 50)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2499)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 51)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2506)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 51)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2513)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 51)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2520)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 51)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2527)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 51)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2534)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 51)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2541)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 51)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2548)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 52)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2555)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 52)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2562)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 52)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2569)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 52)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2576)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 52)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2583)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 52)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2590)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 52)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2597)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 53)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2604)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 53)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2611)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 53)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2618)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 53)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2625)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 53)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2632)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 53)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2639)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 53)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2646)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 54)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2653)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 54)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2660)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 54)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2667)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 54)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2674)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 54)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2681)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 54)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2688)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 54)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2695)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 55)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2702)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 55)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2709)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 55)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2716)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 55)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2723)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 55)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2730)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 55)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2737)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 55)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2744)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 56)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2751)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 56)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2758)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 56)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2765)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 56)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2772)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 56)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2779)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 56)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2786)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 56)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2793)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 57)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2800)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 57)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2807)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 57)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2814)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 57)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2821)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 57)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2828)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 57)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2835)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 57)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2842)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 58)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2849)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 58)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2856)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 58)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2863)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 58)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2870)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 58)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2877)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 58)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2884)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 58)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2891)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 59)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2898)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 59)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2905)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 59)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2912)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 59)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2919)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 59)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2926)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 59)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2933)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 59)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2940)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 60)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2947)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 60)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2954)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 60)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2961)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 60)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2968)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 60)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2975)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 60)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2982)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 60)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2989)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 61)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2996)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 61)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3003)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 61)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3010)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 61)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3017)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 61)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3024)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 61)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3031)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 61)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3038)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 62)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3045)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 62)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3052)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 62)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3059)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 62)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3066)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 62)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3073)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 62)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3080)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 62)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3087)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 63)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3094)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 63)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3101)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 63)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3108)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 63)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3115)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 63)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3122)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 63)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3129)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 63)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3136)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 64)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3143)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 64)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3150)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 64)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3157)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 64)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3164)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 64)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3171)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 64)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3178)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 64)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3185)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 65)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3192)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 65)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3199)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 65)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3206)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 65)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3213)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 65)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3220)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 65)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3227)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 65)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3234)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 66)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3241)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 66)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3248)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 66)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3255)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 66)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3262)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 66)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3269)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 66)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3276)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 66)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3283)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 67)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3290)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 67)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3297)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 67)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3304)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 67)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3311)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 67)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3318)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 67)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3325)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 67)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3332)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 68)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3339)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 68)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3346)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 68)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3353)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 68)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3360)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 68)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3367)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 68)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3374)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 68)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3381)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 69)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3388)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 69)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3395)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 69)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3402)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 69)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3409)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 69)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3416)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 69)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3423)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 69)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3430)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 70)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3437)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 70)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3444)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 70)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3451)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 70)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3458)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 70)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3465)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 70)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3472)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 70)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3479)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 71)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3486)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 71)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3493)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 71)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3500)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 71)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3507)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 71)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3514)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 71)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3521)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 71)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3528)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 72)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3535)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 72)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3542)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 72)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3549)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 72)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3556)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 72)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3563)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 72)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3570)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 72)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3577)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 73)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3584)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 73)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3591)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 73)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3598)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 73)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3605)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 73)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3612)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 73)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3619)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 73)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3626)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 74)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3633)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 74)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3640)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 74)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3647)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 74)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3654)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 74)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3661)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 74)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3668)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 74)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3675)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 75)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3682)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 75)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3689)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 75)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3696)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 75)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3703)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 75)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3710)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 75)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3717)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 75)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3724)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 76)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3731)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 76)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3738)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 76)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3745)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 76)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3752)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 76)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3759)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 76)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3766)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 76)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3773)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 77)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3780)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 77)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3787)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 77)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3794)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 77)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3801)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 77)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3808)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 77)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3815)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 77)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3822)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 78)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3829)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 78)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3836)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 78)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3843)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 78)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3850)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 78)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3857)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 78)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3864)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 78)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3871)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 79)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3878)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 79)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3885)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 79)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3892)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 79)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3899)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 79)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3906)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 79)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3913)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 79)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3920)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 80)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3927)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 80)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3934)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 80)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3941)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 80)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3948)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 80)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3955)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 80)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3962)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 80)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3969)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 81)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3976)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 81)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3983)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 81)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3990)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 81)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3997)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 81)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4004)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 81)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4011)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 81)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4018)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 82)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4025)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 82)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4032)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 82)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4039)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 82)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4046)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 82)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4053)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 82)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4060)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 82)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4067)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 83)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4074)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 83)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4081)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 83)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4088)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 83)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4095)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 83)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4102)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 83)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4109)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 83)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4116)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 84)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4123)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 84)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4130)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 84)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4137)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 84)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4144)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 84)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4151)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 84)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4158)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 84)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4165)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 85)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4172)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 85)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4179)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 85)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4186)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 85)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4193)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 85)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4200)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 85)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4207)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 85)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4214)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 86)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4221)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 86)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4228)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 86)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4235)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 86)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4242)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 86)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4249)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 86)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4256)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 86)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4263)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 87)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4270)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 87)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4277)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 87)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4284)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 87)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4291)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 87)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4298)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 87)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4305)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 87)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4312)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 88)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4319)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 88)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4326)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 88)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4333)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 88)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4340)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 88)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4347)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 88)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4354)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 88)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4361)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 89)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4368)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 89)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4375)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 89)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4382)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 89)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4389)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 89)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4396)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 89)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4403)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 89)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4410)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 90)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4417)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 90)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4424)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 90)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4431)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 90)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4438)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 90)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4445)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 90)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4452)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 90)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4459)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 91)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4466)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 91)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4473)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 91)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4480)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 91)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4487)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 91)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4494)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 91)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4501)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 91)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4508)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 92)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4515)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 92)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4522)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 92)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4529)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 92)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4536)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 92)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4543)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 92)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4550)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 92)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4557)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 93)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4564)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 93)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4571)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 93)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4578)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 93)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4585)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 93)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4592)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 93)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4599)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 93)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4606)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 94)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4613)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 94)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4620)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 94)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4627)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 94)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4634)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 94)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4641)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 94)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4648)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 94)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4655)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 95)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4662)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 95)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4669)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 95)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4676)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 95)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4683)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 95)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4690)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 95)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4697)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 95)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4704)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 96)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4711)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 96)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4718)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 96)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4725)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 96)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4732)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 96)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4739)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 96)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4746)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 96)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4753)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 97)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4760)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 97)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4767)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 97)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4774)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 97)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4781)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 97)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4788)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 97)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4795)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 97)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4802)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 98)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4809)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 98)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4816)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 98)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4823)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 98)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4830)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 98)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4837)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 98)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4844)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 98)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4851)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 99)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4858)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 99)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4865)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 99)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4872)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 99)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4879)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 99)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4886)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 99)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4893)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 99)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4900)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 100)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4907)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 100)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4914)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 100)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4921)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 100)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4928)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 100)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4935)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 100)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4942)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 100)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4949)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 101)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4956)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 101)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4963)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 101)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4970)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 101)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4977)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 101)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4984)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 101)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4991)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 101)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4998)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 102)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5005)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 102)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5012)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 102)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5019)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 102)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5026)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 102)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5033)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 102)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5040)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 102)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5047)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 103)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5054)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 103)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5061)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 103)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5068)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 103)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5075)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 103)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5082)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 103)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5089)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 103)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5096)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 104)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5103)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 104)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5110)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 104)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5117)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 104)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5124)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 104)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5131)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 104)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5138)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 104)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5145)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 105)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5152)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 105)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5159)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 105)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5166)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 105)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5173)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 105)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5180)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 105)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5187)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 105)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5194)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 106)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5201)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 106)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5208)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 106)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5215)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 106)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5222)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 106)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5229)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 106)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5236)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 106)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5243)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 107)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5250)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 107)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5257)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 107)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5264)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 107)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5271)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 107)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5278)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 107)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5285)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 107)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5292)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 108)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5299)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 108)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5306)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 108)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5313)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 108)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5320)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 108)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5327)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 108)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5334)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 108)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5341)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 109)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5348)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 109)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5355)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 109)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5362)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 109)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5369)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 109)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5376)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 109)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5383)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 109)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5390)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 110)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5397)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 110)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5404)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 110)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5411)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 110)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5418)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 110)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5425)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 110)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5432)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 110)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5439)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 111)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5446)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 111)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5453)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 111)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5460)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 111)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5467)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 111)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5474)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 111)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5481)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 111)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5488)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 112)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5495)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 112)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5502)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 112)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5509)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 112)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5516)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 112)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5523)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 112)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5530)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 112)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5537)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 113)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5544)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 113)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5551)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 113)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5558)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 113)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5565)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 113)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5572)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 113)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5579)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 113)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5586)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 114)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5593)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 114)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5600)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 114)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5607)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 114)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5614)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 114)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5621)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 114)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5628)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 114)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5635)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 115)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5642)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 115)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5649)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 115)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5656)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 115)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5663)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 115)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5670)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 115)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5677)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 115)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5684)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 116)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5691)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 116)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5698)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 116)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5705)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 116)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5712)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 116)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5719)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 116)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5726)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 116)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5733)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 117)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5740)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 117)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5747)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 117)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5754)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 117)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5761)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 117)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5768)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 117)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5775)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 117)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5782)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 118)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5789)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 118)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5796)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 118)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5803)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 118)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5810)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 118)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5817)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 118)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5824)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 118)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5831)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 119)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5838)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 119)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5845)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 119)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5852)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 119)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5859)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 119)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5866)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 119)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5873)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 119)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5880)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 120)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5887)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 120)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5894)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 120)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5901)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 120)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5908)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 120)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5915)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 120)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5922)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 120)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5929)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 121)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5936)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 121)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5943)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 121)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5950)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 121)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5957)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 121)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5964)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 121)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5971)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 121)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5978)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 122)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5985)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 122)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5992)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 122)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5999)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 122)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6006)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 122)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6013)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 122)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6020)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 122)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6027)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 123)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6034)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 123)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6041)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 123)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6048)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 123)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6055)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 123)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6062)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 123)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6069)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 123)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6076)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 124)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6083)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 124)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6090)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 124)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6097)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 124)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6104)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 124)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6111)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 124)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6118)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 124)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6125)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 125)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6132)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 125)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6139)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 125)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6146)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 125)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6153)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 125)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6160)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 125)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6167)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 125)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6174)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 126)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6181)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 126)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6188)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 126)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6195)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 126)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6202)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 126)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6209)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 126)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6216)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 126)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6223)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 127)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6230)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 127)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6237)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 127)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6244)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 127)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6251)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 127)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6258)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 127)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6265)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 127)]));
}
}
}
- compute[(((((int)blockIdx.x) * 784) + ((((int)threadIdx.x) / 7) * 49)) + (((int)threadIdx.x) % 7))] = max((conv2d_nchw[0] + bias[((((int)blockIdx.x) * 16) + (((int)threadIdx.x) / 7))]), 0.000000e+00f);
- compute[((((((int)blockIdx.x) * 784) + ((((int)threadIdx.x) / 7) * 49)) + (((int)threadIdx.x) % 7)) + 7)] = max((conv2d_nchw[1] + bias[((((int)blockIdx.x) * 16) + (((int)threadIdx.x) / 7))]), 0.000000e+00f);
- compute[((((((int)blockIdx.x) * 784) + ((((int)threadIdx.x) / 7) * 49)) + (((int)threadIdx.x) % 7)) + 14)] = max((conv2d_nchw[2] + bias[((((int)blockIdx.x) * 16) + (((int)threadIdx.x) / 7))]), 0.000000e+00f);
- compute[((((((int)blockIdx.x) * 784) + ((((int)threadIdx.x) / 7) * 49)) + (((int)threadIdx.x) % 7)) + 21)] = max((conv2d_nchw[3] + bias[((((int)blockIdx.x) * 16) + (((int)threadIdx.x) / 7))]), 0.000000e+00f);
- compute[((((((int)blockIdx.x) * 784) + ((((int)threadIdx.x) / 7) * 49)) + (((int)threadIdx.x) % 7)) + 28)] = max((conv2d_nchw[4] + bias[((((int)blockIdx.x) * 16) + (((int)threadIdx.x) / 7))]), 0.000000e+00f);
- compute[((((((int)blockIdx.x) * 784) + ((((int)threadIdx.x) / 7) * 49)) + (((int)threadIdx.x) % 7)) + 35)] = max((conv2d_nchw[5] + bias[((((int)blockIdx.x) * 16) + (((int)threadIdx.x) / 7))]), 0.000000e+00f);
- compute[((((((int)blockIdx.x) * 784) + ((((int)threadIdx.x) / 7) * 49)) + (((int)threadIdx.x) % 7)) + 42)] = max((conv2d_nchw[6] + bias[((((int)blockIdx.x) * 16) + (((int)threadIdx.x) / 7))]), 0.000000e+00f);
+ for (int i1_inner = 0; i1_inner < 4; ++i1_inner) {
+ for (int i2_inner = 0; i2_inner < 7; ++i2_inner) {
+ compute[(((((((int)blockIdx.x) * 1568) + ((((int)threadIdx.x) / 7) * 196)) + (i1_inner * 49)) + (i2_inner * 7)) + (((int)threadIdx.x) % 7))] = max((conv2d_nchw[((i1_inner * 7) + i2_inner)] + bias[(((((int)blockIdx.x) * 32) + ((((int)threadIdx.x) / 7) * 4)) + i1_inner)]), 0.000000e+00f);
+ }
+ }
}
@@ -2654,7 +1165,7 @@ 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 30.519 seconds)
+ **Total running time of the script:** ( 6 minutes 17.263 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 8c065fc54d..b31bbd22f1 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)
- 8.0970 8.0959 8.1005 8.0947 0.0025
+ 8.1035 8.1016 8.1151 8.0939 0.0087
@@ -675,7 +675,7 @@ Other Tips
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 9.444 seconds)
+ **Total running time of the script:** ( 1 minutes 12.706 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 b26042848a..2807b47090 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)
- 719.8005 719.7170 720.9380 718.7466 0.8966
+ 759.3156 758.9744 760.6804 758.2919 1.0045
@@ -694,7 +694,7 @@ Other Tips
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 40.103 seconds)
+ **Total running time of the script:** ( 1 minutes 45.435 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 258e1b4d31..97f1b445d3 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
@@ -389,53 +389,26 @@ 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": T.bool(True), "global_symbol": "main", "tir.noalias": T.bool(True)})
- 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 i_outer_inner in range(64):
- cse_var_2: T.int32 = i_outer_inner * 8
- cse_var_1: T.int32 = i0_outer_i1_outer_fused % 64 // 2
- compute_2[cse_var_2] = T.float32(0)
- compute_2[cse_var_2 + 1] = T.float32(0)
- compute_2[cse_var_2 + 2] = T.float32(0)
- compute_2[cse_var_2 + 3] = T.float32(0)
- compute_2[cse_var_2 + 4] = T.float32(0)
- compute_2[cse_var_2 + 5] = T.float32(0)
- compute_2[cse_var_2 + 6] = T.float32(0)
- compute_2[cse_var_2 + 7] = T.float32(0)
- for elem_idx in range(placeholder_5[cse_var_1 + 1] - placeholder_5[cse_var_1]):
+ 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 nb_j_inner in range(2):
+ for i_inner_init, j_init in T.grid(64, 16):
+ compute_2[i_inner_init * 32 + nb_j_inner * 16 + j_init] = T.float32(0)
+ for elem_idx, i_inner, j in T.grid(T.Let(placeholder_5[cse_var_1 + 1] - placeholder_5[cse_var_1], where={cse_var_1: i0_outer_i1_outer_fused % 16 * 2 + nb_j_inner}), 64, 16):
+ cse_var_1 = T.int32()
placeholder_5 = T.Buffer((33,), "int32", data=placeholder_3.data)
+ cse_var_3: T.int32 = i0_outer_i1_outer_fused % 16 * 2 + nb_j_inner
+ cse_var_2: T.int32 = i_inner * 32 + nb_j_inner * 16 + j
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[cse_var_1 + 1] - placeholder_5[cse_var_1]):
- compute_2[cse_var_2] = compute_2[cse_var_2] + placeholder_6[placeholder_5[cse_var_1] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 2 * 8] * T.max(placeholder_7[i0_outer_i1_outer_fused // 64 * 16384 + i_outer_inner * 256 + placeholder_8[placeholder_5[cse_var_1] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[cse_var_1 + 1] - placeholder_5[cse_var_1]):
- cse_var_3: T.int32 = cse_var_2 + 1
- compute_2[cse_var_3] = compute_2[cse_var_3] + placeholder_6[placeholder_5[cse_var_1] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 2 * 8 + 1] * T.max(placeholder_7[i0_outer_i1_outer_fused // 64 * 16384 + i_outer_inner * 256 + placeholder_8[placeholder_5[cse_var_1] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[cse_var_1 + 1] - placeholder_5[cse_var_1]):
- cse_var_4: T.int32 = cse_var_2 + 2
- compute_2[cse_var_4] = compute_2[cse_var_4] + placeholder_6[placeholder_5[cse_var_1] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 2 * 8 + 2] * T.max(placeholder_7[i0_outer_i1_outer_fused // 64 * 16384 + i_outer_inner * 256 + placeholder_8[placeholder_5[cse_var_1] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[cse_var_1 + 1] - placeholder_5[cse_var_1]):
- cse_var_5: T.int32 = cse_var_2 + 3
- compute_2[cse_var_5] = compute_2[cse_var_5] + placeholder_6[placeholder_5[cse_var_1] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 2 * 8 + 3] * T.max(placeholder_7[i0_outer_i1_outer_fused // 64 * 16384 + i_outer_inner * 256 + placeholder_8[placeholder_5[cse_var_1] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[cse_var_1 + 1] - placeholder_5[cse_var_1]):
- cse_var_6: T.int32 = cse_var_2 + 4
- compute_2[cse_var_6] = compute_2[cse_var_6] + placeholder_6[placeholder_5[cse_var_1] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 2 * 8 + 4] * T.max(placeholder_7[i0_outer_i1_outer_fused // 64 * 16384 + i_outer_inner * 256 + placeholder_8[placeholder_5[cse_var_1] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[cse_var_1 + 1] - placeholder_5[cse_var_1]):
- cse_var_7: T.int32 = cse_var_2 + 5
- compute_2[cse_var_7] = compute_2[cse_var_7] + placeholder_6[placeholder_5[cse_var_1] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 2 * 8 + 5] * T.max(placeholder_7[i0_outer_i1_outer_fused // 64 * 16384 + i_outer_inner * 256 + placeholder_8[placeholder_5[cse_var_1] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[cse_var_1 + 1] - placeholder_5[cse_var_1]):
- cse_var_8: T.int32 = cse_var_2 + 6
- compute_2[cse_var_8] = compute_2[cse_var_8] + placeholder_6[placeholder_5[cse_var_1] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 2 * 8 + 6] * T.max(placeholder_7[i0_outer_i1_outer_fused // 64 * 16384 + i_outer_inner * 256 + placeholder_8[placeholder_5[cse_var_1] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[cse_var_1 + 1] - placeholder_5[cse_var_1]):
- cse_var_9: T.int32 = cse_var_2 + 7
- compute_2[cse_var_9] = compute_2[cse_var_9] + placeholder_6[placeholder_5[cse_var_1] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 2 * 8 + 7] * T.max(placeholder_7[i0_outer_i1_outer_fused // 64 * 16384 + i_outer_inner * 256 + placeholder_8[placeholder_5[cse_var_1] + elem_idx]], T.float32(0))
+ compute_2[cse_var_2] = compute_2[cse_var_2] + placeholder_6[placeholder_5[cse_var_3] * 16 + elem_idx * 16 + j] * T.max(placeholder_7[i0_outer_i1_outer_fused // 16 * 16384 + i_inner * 256 + placeholder_8[placeholder_5[cse_var_3] + elem_idx]], T.float32(0))
for i0_inner in range(64):
- cse_var_10: T.int32 = i0_outer_i1_outer_fused // 64 * 32768 + i0_inner * 512 + i0_outer_i1_outer_fused % 64 * 8
+ cse_var_4: T.int32 = i0_outer_i1_outer_fused // 16 * 32768 + 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_10:cse_var_10 + 8] = T.max(compute_2[i0_inner * 8:i0_inner * 8 + 8] + placeholder_5[cse_var_10:cse_var_10 + 8], T.Broadcast(T.float32(0), 8))
+ compute_3[cse_var_4:cse_var_4 + 32] = T.max(compute_2[i0_inner * 32:i0_inner * 32 + 32] + placeholder_5[cse_var_4:cse_var_4 + 32], T.Broadcast(T.float32(0), 32))
@@ -485,7 +458,7 @@ We build the binary and check its correctness and performance.
.. code-block:: none
- Execution time of this operator: 2.490 ms
+ Execution time of this operator: 1.800 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 3390f6f78d..c1e90af510 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
=================
-**00:39.175** total execution time for **how_to_tune_with_autotvm** files:
+**00:45.955** 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``) | 00:39.137 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autotvm_tune_conv2d_cuda.py` (``tune_conv2d_cuda.py``) | 00:45.916 | 0.0 MB |
+--------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autotvm_tune_relay_x86.py` (``tune_relay_x86.py``) | 00:00.023 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autotvm_tune_relay_x86.py` (``tune_relay_x86.py``) | 00:00.024 | 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_mobile_gpu.py` (``tune_relay_mobile_gpu.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_mobile_gpu.py` (``tune_relay_mobile_gpu.py``) | 00:00.005 | 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 be9e71e20f..34a6433c51 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
@@ -268,8 +268,8 @@ for this template
waiting for device...
device available
Get devices for measurement successfully!
- No: 1 GFLOPS: 19.81/19.81 result: MeasureResult(costs=(0.011685259727272728,), error_no=MeasureErrorNo.NO_ERROR, all_cost=2.231394052505493, timestamp=1681665130.3004358) [('tile_f', [-1, 32, 2, 2]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 2, 2]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,5847229
- No: 2 GFLOPS: 0.00/19.81 result: Traceback (most recent call last):
+ No: 1 GFLOPS: 108.31/108.31 result: MeasureResult(costs=(0.0021374299215686276,), error_no=MeasureErrorNo.NO_ERROR, all_cost=4.775891065597534, timestamp=1681675917.6439672) [('tile_f', [-1, 4, 1, 1]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 1, 2]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,10103502
+ No: 2 GFLOPS: 0.00/108.31 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
@@ -391,8 +391,9 @@ 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, 1, 4, 128]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 1, 64]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,4998835
- No: 3 GFLOPS: 0.00/19.81 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 1, 256, 2]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 32, 1]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,4664539
+ No: 3 GFLOPS: 81.83/108.31 result: MeasureResult(costs=(0.0028289757358490567,), error_no=MeasureErrorNo.NO_ERROR, all_cost=5.083340644836426, timestamp=1681675920.383326) [('tile_f', [-1, 2, 4, 16]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 8, 1]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,8917656
+ No: 4 GFLOPS: 0.00/108.31 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
@@ -514,8 +515,8 @@ 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, 4, 1]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 128, 2]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,7223285
- No: 4 GFLOPS: 0.00/19.81 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 2, 8]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 32, 8]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,4762264
+ No: 5 GFLOPS: 0.00/108.31 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
@@ -637,8 +638,8 @@ 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, 2, 2]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 1, 512]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,9290890
- No: 5 GFLOPS: 0.00/19.81 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 4, 1, 128]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 64, 8]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,7280012
+ No: 6 GFLOPS: 0.00/108.31 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
@@ -760,377 +761,8 @@ 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, 8, 1, 64]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 32, 8]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,502243
- No: 6 GFLOPS: 0.00/19.81 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:401
- 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:387
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:282
- 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:101
- 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:401
- 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:387
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:282
- 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:101
- 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, 1, 512]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 4, 8]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,492139
- No: 7 GFLOPS: 0.00/19.81 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:401
- 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:387
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:282
- 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:101
- 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:401
- 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:387
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:282
- 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:101
- 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, 8, 16, 4]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 8, 64]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,5785029
- No: 8 GFLOPS: 0.00/19.81 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:401
- 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:387
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:282
- 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:101
- 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:401
- 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:387
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:282
- 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:101
- 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, 4, 4, 16]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 32, 16]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,1299277
- No: 9 GFLOPS: 0.00/19.81 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, 7, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 32, 4]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,1635393
+ No: 7 GFLOPS: 0.00/108.31 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
@@ -1218,7 +850,7 @@ for this template
15: _PyEval_EvalFrameDefault
14: 0x0000000000537c30
13: _PyObject_FastCallKeywords
- 12: 0x00007f4651985fa2
+ 12: 0x00007f211f670fa2
11: _ctypes_callproc
10: ffi_call
9: ffi_call_unix64
@@ -1282,132 +914,8 @@ for this template
22: _PyEval_EvalFrameDefault
21: _PyFunction_FastCallKeywords
20: _PyEval_EvalFrameDefault
- 19: _PyFunction_FastCall [('tile_f', [-1, 4, 1, 16]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 1, 16]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,8447946
- No: 10 GFLOPS: 43.92/43.92 result: MeasureResult(costs=(0.005271183210526315,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.4004976749420166, timestamp=1681665144.7520602) [('tile_f', [-1, 2, 4, 1]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 2, 1]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,3294740
- No: 11 GFLOPS: 0.00/43.92 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
- 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:401
- 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:387
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:282
- 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:101
- 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:401
- 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:387
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:282
- 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:101
- 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, 4, 32, 1]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 2, 32]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,8663202
- No: 12 GFLOPS: 0.00/43.92 result: Traceback (most recent call last):
+ 19: _PyFunction_FastCall [('tile_f', [-1, 4, 1, 8]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 4, 2]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,5271118
+ No: 8 GFLOPS: 0.00/108.31 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
@@ -1529,8 +1037,8 @@ 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, 2, 1, 128]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 8, 64]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,169171
- No: 13 GFLOPS: 0.00/43.92 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 1, 1, 2]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 128, 4]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,675455
+ No: 9 GFLOPS: 0.00/108.31 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
@@ -1652,8 +1160,8 @@ 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, 4, 32, 2]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 32, 2]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,6637712
- No: 14 GFLOPS: 0.00/43.92 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 1, 8, 64]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 8, 2]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,9533689
+ No: 10 GFLOPS: 0.00/108.31 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
@@ -1775,8 +1283,8 @@ 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, 32, 1, 2]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 512, 1]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,5067760
- No: 15 GFLOPS: 0.00/43.92 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 32, 16, 1]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 1, 8]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,2033279
+ No: 11 GFLOPS: 0.00/108.31 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
@@ -1898,8 +1406,13 @@ 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, 8, 4]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 2, 256]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,6578125
- No: 16 GFLOPS: 0.00/43.92 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 4, 8, 16]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 8, 1]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,6206161
+ No: 12 GFLOPS: 19.48/108.31 result: MeasureResult(costs=(0.011882424999999999,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.517064094543457, timestamp=1681675931.8390515) [('tile_f', [-1, 1, 2, 32]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 8, 4]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,6468850
+ No: 13 GFLOPS: 171.17/171.17 result: MeasureResult(costs=(0.0013524473611111112,), error_no=MeasureErrorNo.NO_ERROR, all_cost=3.153829336166382, timestamp=1681675936.8126795) [('tile_f', [-1, 2, 16, 2]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 32, 4]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,1829606
+ No: 14 GFLOPS: 42.38/171.17 result: MeasureResult(costs=(0.005463093789473685,), error_no=MeasureErrorNo.NO_ERROR, all_cost=2.143911600112915, timestamp=1681675937.7011867) [('tile_f', [-1, 1, 8, 4]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 1, 16]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,1669041
+ No: 15 GFLOPS: 6.46/171.17 result: MeasureResult(costs=(0.0358309935,), error_no=MeasureErrorNo.NO_ERROR, all_cost=4.595254182815552, timestamp=1681675938.7023196) [('tile_f', [-1, 64, 8, 1]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 2, 2]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,7202833
+ No: 16 GFLOPS: 11.54/171.17 result: MeasureResult(costs=(0.020058811166666666,), error_no=MeasureErrorNo.NO_ERROR, all_cost=2.235470771789551, timestamp=1681675939.692564) [('tile_f', [-1, 1, 8, 64]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 1, 4]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,2973289
+ No: 17 GFLOPS: 0.00/171.17 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
@@ -2021,8 +1534,8 @@ 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, 4, 16, 2]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 1, 128]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,10239767
- No: 17 GFLOPS: 0.00/43.92 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 4, 16, 2]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 1, 256]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,6573907
+ No: 18 GFLOPS: 0.00/171.17 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
@@ -2144,8 +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, 128, 1]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 256, 2]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,7228149
- No: 18 GFLOPS: 0.00/43.92 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 4, 8, 2]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 16, 2]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,8377461
+ No: 19 GFLOPS: 0.00/171.17 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
@@ -2267,8 +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, 4, 8]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 8, 32]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,2477573
- No: 19 GFLOPS: 0.00/43.92 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 16, 8]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 2, 64]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,7521519
+ No: 20 GFLOPS: 0.00/171.17 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
@@ -2390,8 +1903,7 @@ 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, 1]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 512, 1]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,3712290
- No: 20 GFLOPS: 3.16/43.92 result: MeasureResult(costs=(0.07319751725000001,), error_no=MeasureErrorNo.NO_ERROR, all_cost=3.892284393310547, timestamp=1681665149.061125) [('tile_f', [-1, 1, 1, 4]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 4, 32]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,7120840
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 8, 64, 1]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 2, 64]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,8100008
@@ -2446,9 +1958,9 @@ and measure running time.
Finish loading 20 records
Best config:
- [('tile_f', [-1, 2, 4, 1]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 2, 1]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,3294740
+ [('tile_f', [-1, 2, 16, 2]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 32, 4]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,1829606
Finish loading 20 records
- Time cost of this operator: 0.005648
+ Time cost of this operator: 0.001788
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 5d48a52175..a6a656bda0 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 313.3 98.754 (1, 2, 10, 10, 3) 2 1 [313.3]
- tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 3.024 0.953 (1, 6, 10, 10) 1 1 [3.024]
- tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.93 0.293 (1, 1, 10, 10, 3) 1 1 [0.93]
- Total_time - 317.253 - - - - -
+ tvmgen_default_fused_nn_contrib_conv2d_NCHWc tvmgen_default_fused_nn_contrib_conv2d_NCHWc 315.4 98.599 (1, 2, 10, 10, 3) 2 1 [315.4]
+ tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 3.506 1.096 (1, 6, 10, 10) 1 1 [3.506]
+ tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.976 0.305 (1, 1, 10, 10, 3) 1 1 [0.976]
+ Total_time - 319.882 - - - - -
@@ -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 138.6 98.19 (1, 6, 10, 10, 1) 2 1 [138.6]
- tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 1.716 1.216 (1, 6, 10, 10) 1 1 [1.716]
- tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.839 0.594 (1, 3, 10, 10, 1) 1 1 [0.839]
- Total_time - 141.155 - - - - -
+ tvmgen_default_fused_nn_contrib_conv2d_NCHWc tvmgen_default_fused_nn_contrib_conv2d_NCHWc 139.8 98.163 (1, 6, 10, 10, 1) 2 1 [139.8]
+ tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 1.777 1.248 (1, 6, 10, 10) 1 1 [1.777]
+ tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.839 0.589 (1, 3, 10, 10, 1) 1 1 [0.839]
+ Total_time - 142.417 - - - - -
@@ -439,7 +439,7 @@ Timing the tuned program
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 20.283 seconds)
+ **Total running time of the script:** ( 1 minutes 24.734 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 590dcdaba5..9e7754a353 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]
100%|##########| 3.42M/3.42M [00:00<00:00, 47.9MB/s]
+
0%| | 0.00/3.42M [00:00<?, ?B/s]
61%|###### | 2.09M/3.42M [00:00<00:00, 14.2MB/s]
100%|##########| 3.42M/3.42M [00:00<00:00, 22.2MB/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 15.917 seconds)
+ **Total running time of the script:** ( 1 minutes 20.209 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 c97f156d83..84f95161a8 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
@@ -217,7 +217,7 @@ take about **2 minutes** to download the Stanford Cars, while COCO 2017 validati
.. code-block:: none
- '/tmp/tmp4zj673b7/images/random'
+ '/tmp/tmph78n3crc/images/random'
@@ -308,7 +308,7 @@ 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], [0.0, 1.0], [0.0, 1.0], [1.0, 0.0], [1.0, 0.0], [0.0, 1.0], [1.0, 0.0], [1.0, 0.0], [0.0, 1.0], [1.0, 0.0]
+ :alt: [0.0, 1.0], [0.0, 1.0], [1.0, 0.0], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0], [1.0, 0.0], [1.0, 0.0]
:srcset: /how_to/work_with_microtvm/images/sphx_glr_micro_train_001.png
:class: sphx-glr-single-img
@@ -317,8 +317,8 @@ objects to other stuff? We can display some examples from our datasets using ``m
.. code-block:: none
- /tmp/tmp4zj673b7/images/target contains 8144 images
- /tmp/tmp4zj673b7/images/random contains 5000 images
+ /tmp/tmph78n3crc/images/target contains 8144 images
+ /tmp/tmph78n3crc/images/random contains 5000 images
@@ -493,13 +493,13 @@ the time on our validation set).
.. code-block:: none
Epoch 1/3
- 328/328 - 42s - loss: 0.2285 - accuracy: 0.9209 - val_loss: 0.1201 - val_accuracy: 0.9551 - 42s/epoch - 127ms/step
+ 328/328 - 41s - loss: 0.2109 - accuracy: 0.9263 - val_loss: 0.1025 - val_accuracy: 0.9653 - 41s/epoch - 125ms/step
Epoch 2/3
- 328/328 - 34s - loss: 0.1074 - accuracy: 0.9607 - val_loss: 0.0938 - val_accuracy: 0.9713 - 34s/epoch - 103ms/step
+ 328/328 - 35s - loss: 0.1018 - accuracy: 0.9636 - val_loss: 0.1084 - val_accuracy: 0.9645 - 35s/epoch - 106ms/step
Epoch 3/3
- 328/328 - 34s - loss: 0.0593 - accuracy: 0.9780 - val_loss: 0.1151 - val_accuracy: 0.9656 - 34s/epoch - 103ms/step
+ 328/328 - 35s - loss: 0.0701 - accuracy: 0.9755 - val_loss: 0.1198 - val_accuracy: 0.9619 - 35s/epoch - 105ms/step
- <keras.callbacks.History object at 0x7f1c35760750>
+ <keras.callbacks.History object at 0x7f817d328f90>
@@ -860,7 +860,7 @@ 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 15.285 seconds)
+ **Total running time of the script:** ( 4 minutes 23.735 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 a80dd241d1..397177d913 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
@@ -5,20 +5,20 @@
Computation times
=================
-**07:16.771** total execution time for **how_to_work_with_microtvm** files:
+**07:35.026** total execution time for **how_to_work_with_microtvm** files:
+-----------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_microtvm_micro_train.py` (``micro_train.py``) | 04:15.285 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_microtvm_micro_train.py` (``micro_train.py``) | 04:23.735 | 0.0 MB |
+-----------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_microtvm_micro_autotune.py` (``micro_autotune.py``) | 01:20.283 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_microtvm_micro_autotune.py` (``micro_autotune.py``) | 01:24.734 | 0.0 MB |
+-----------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_microtvm_micro_pytorch.py` (``micro_pytorch.py``) | 01:15.917 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_microtvm_micro_pytorch.py` (``micro_pytorch.py``) | 01:20.209 | 0.0 MB |
+-----------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_microtvm_micro_aot.py` (``micro_aot.py``) | 00:09.971 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_microtvm_micro_aot.py` (``micro_aot.py``) | 00:10.629 | 0.0 MB |
+-----------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_microtvm_micro_custom_ide.py` (``micro_custom_ide.py``) | 00:07.717 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_microtvm_micro_custom_ide.py` (``micro_custom_ide.py``) | 00:08.199 | 0.0 MB |
+-----------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_microtvm_micro_tflite.py` (``micro_tflite.py``) | 00:07.599 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_microtvm_micro_tflite.py` (``micro_tflite.py``) | 00:07.520 | 0.0 MB |
+-----------------------------------------------------------------------------------------+-----------+--------+
| :ref:`sphx_glr_how_to_work_with_microtvm_micro_ethosu.py` (``micro_ethosu.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 d9a6b198e1..ea794fa985 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
=================
-**00:35.635** total execution time for **how_to_work_with_relay** files:
+**00:37.041** 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``) | 00:31.166 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_relay_using_pipeline_executor.py` (``using_pipeline_executor.py``) | 00:32.542 | 0.0 MB |
+----------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_relay_using_external_lib.py` (``using_external_lib.py``) | 00:02.777 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_relay_using_external_lib.py` (``using_external_lib.py``) | 00:02.929 | 0.0 MB |
+----------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_relay_build_gcn.py` (``build_gcn.py``) | 00:01.686 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_relay_build_gcn.py` (``build_gcn.py``) | 00:01.563 | 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 |
+| :ref:`sphx_glr_how_to_work_with_relay_using_relay_viz.py` (``using_relay_viz.py``) | 00:00.007 | 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 2564cb6e43..8aee85dff3 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
@@ -278,7 +278,7 @@ The following example customizes CUDA lowering rule for :code:`exp`.
.. code-block:: none
- <function my_cuda_math_rule at 0x7f18646ddd40>
+ <function my_cuda_math_rule at 0x7f80481a47a0>
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 3def8c34e7..dfe051a08f 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
=================
-**00:08.355** total execution time for **how_to_work_with_schedules** files:
+**00:08.600** total execution time for **how_to_work_with_schedules** files:
+------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_intrin_math.py` (``intrin_math.py``) | 00:05.530 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_intrin_math.py` (``intrin_math.py``) | 00:05.824 | 0.0 MB |
+------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_tensorize.py` (``tensorize.py``) | 00:01.371 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_tensorize.py` (``tensorize.py``) | 00:01.257 | 0.0 MB |
+------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_reduction.py` (``reduction.py``) | 00:00.601 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_reduction.py` (``reduction.py``) | 00:00.627 | 0.0 MB |
+------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_scan.py` (``scan.py``) | 00:00.583 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_scan.py` (``scan.py``) | 00:00.608 | 0.0 MB |
+------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_extern_op.py` (``extern_op.py``) | 00:00.124 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_extern_op.py` (``extern_op.py``) | 00:00.133 | 0.0 MB |
+------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_tedd.py` (``tedd.py``) | 00:00.061 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_tedd.py` (``tedd.py``) | 00:00.064 | 0.0 MB |
+------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_schedule_primitives.py` (``schedule_primitives.py``) | 00:00.054 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_schedule_primitives.py` (``schedule_primitives.py``) | 00:00.057 | 0.0 MB |
+------------------------------------------------------------------------------------------------+-----------+--------+
| :ref:`sphx_glr_how_to_work_with_schedules_tuple_inputs.py` (``tuple_inputs.py``) | 00:00.030 | 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 f438d07a54..8a7541e9b5 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
=================
-**00:30.200** total execution time for **topic_vta_tutorials_autotvm** files:
+**00:32.142** total execution time for **topic_vta_tutorials_autotvm** files:
+---------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_autotvm_tune_relay_vta.py` (``tune_relay_vta.py``) | 00:30.193 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_autotvm_tune_relay_vta.py` (``tune_relay_vta.py``) | 00:32.135 | 0.0 MB |
+---------------------------------------------------------------------------------------+-----------+--------+
| :ref:`sphx_glr_topic_vta_tutorials_autotvm_tune_alu_vta.py` (``tune_alu_vta.py``) | 00:00.007 | 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 76f9138ebd..cd27c6d238 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 31.97s!
+ resnet18_v1 inference graph built in 34.43s!
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 d36e269f0a..8bfa5a12e3 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 21.87s!
+ yolov3-tiny inference graph built in 23.36s!
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 80ce091ac2..562cfcb041 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
=================
-**01:37.201** total execution time for **topic_vta_tutorials_frontend** files:
+**01:41.680** total execution time for **topic_vta_tutorials_frontend** files:
+------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_frontend_deploy_classification.py` (``deploy_classification.py``) | 00:48.640 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_frontend_deploy_classification.py` (``deploy_classification.py``) | 00:51.273 | 0.0 MB |
+------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_frontend_deploy_detection.py` (``deploy_detection.py``) | 00:48.562 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_frontend_deploy_detection.py` (``deploy_detection.py``) | 00:50.408 | 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 470d97cb9b..8389cff067 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
=================
-**00:03.132** total execution time for **topic_vta_tutorials_optimize** files:
+**00:03.241** total execution time for **topic_vta_tutorials_optimize** files:
+--------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_optimize_convolution_opt.py` (``convolution_opt.py``) | 00:02.659 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_optimize_convolution_opt.py` (``convolution_opt.py``) | 00:02.750 | 0.0 MB |
+--------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_optimize_matrix_multiply_opt.py` (``matrix_multiply_opt.py``) | 00:00.473 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_optimize_matrix_multiply_opt.py` (``matrix_multiply_opt.py``) | 00:00.491 | 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 5cf6f3bac4..0cefa2b726 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
=================
-**00:00.799** total execution time for **topic_vta_tutorials** files:
+**00:00.832** total execution time for **topic_vta_tutorials** files:
+---------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_matrix_multiply.py` (``matrix_multiply.py``) | 00:00.413 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_matrix_multiply.py` (``matrix_multiply.py``) | 00:00.433 | 0.0 MB |
+---------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_vta_get_started.py` (``vta_get_started.py``) | 00:00.386 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_vta_get_started.py` (``vta_get_started.py``) | 00:00.400 | 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 f593d2b253..cbc5ccfaef 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: 96.502 ms
+ Execution time of this operator: 97.692 ms
@@ -434,7 +434,7 @@ operations.
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 24.173 seconds)
+ **Total running time of the script:** ( 1 minutes 33.956 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 121eee1c4c..a003bc3c00 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: 9.62/9.62 result: MeasureResult(costs=(0.0279181416,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.9916150569915771, timestamp=1681663511.3304734) [('tile_y', [-1, 2]), ('tile_x', [-1, 128])],None,71
- No: 2 GFLOPS: 9.12/9.62 result: MeasureResult(costs=(0.0294408222,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.1057593822479248, timestamp=1681663512.059216) [('tile_y', [-1, 16]), ('tile_x', [-1, 128])],None,74
- No: 3 GFLOPS: 0.49/9.62 result: MeasureResult(costs=(0.5518579202,), error_no=MeasureErrorNo.NO_ERROR, all_cost=9.058472633361816, timestamp=1681663522.3478136) [('tile_y', [-1, 512]), ('tile_x', [-1, 1])],None,9
- No: 4 GFLOPS: 10.39/10.39 result: MeasureResult(costs=(0.025825719199999996,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.7059674263000488, timestamp=1681663524.215304) [('tile_y', [-1, 16]), ('tile_x', [-1, 32])],None,54
- No: 5 GFLOPS: 4.65/10.39 result: MeasureResult(costs=(0.0577035498,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.1835944652557373, timestamp=1681663525.5682073) [('tile_y', [-1, 16]), ('tile_x', [-1, 16])],None,44
- No: 6 GFLOPS: 11.20/11.20 result: MeasureResult(costs=(0.0239624332,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.6260159015655518, timestamp=1681663527.4112155) [('tile_y', [-1, 2]), ('tile_x', [-1, 32])],None,51
- No: 7 GFLOPS: 12.09/12.09 result: MeasureResult(costs=(0.022197886,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.6628327369689941, timestamp=1681663528.0178668) [('tile_y', [-1, 64]), ('tile_x', [-1, 32])],None,56
- No: 8 GFLOPS: 3.41/12.09 result: MeasureResult(costs=(0.0787082696,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.5164477825164795, timestamp=1681663529.5281386) [('tile_y', [-1, 64]), ('tile_x', [-1, 8])],None,36
- No: 9 GFLOPS: 3.72/12.09 result: MeasureResult(costs=(0.0722422336,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.3671281337738037, timestamp=1681663531.0101504) [('tile_y', [-1, 128]), ('tile_x', [-1, 16])],None,47
- No: 10 GFLOPS: 11.70/12.09 result: MeasureResult(costs=(0.0229524498,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.5744786262512207, timestamp=1681663531.6360884) [('tile_y', [-1, 256]), ('tile_x', [-1, 512])],None,98
+ No: 1 GFLOPS: 12.70/12.70 result: MeasureResult(costs=(0.0211406588,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.6140856742858887, timestamp=1681674238.3986027) [('tile_y', [-1, 32]), ('tile_x', [-1, 128])],None,75
+ No: 2 GFLOPS: 12.89/12.89 result: MeasureResult(costs=(0.020830415999999997,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.7135822772979736, timestamp=1681674238.994832) [('tile_y', [-1, 64]), ('tile_x', [-1, 128])],None,76
+ No: 3 GFLOPS: 14.75/14.75 result: MeasureResult(costs=(0.018199289400000002,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.8523674011230469, timestamp=1681674240.785946) [('tile_y', [-1, 64]), ('tile_x', [-1, 64])],None,66
+ No: 4 GFLOPS: 3.07/14.75 result: MeasureResult(costs=(0.0873224262,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.657752275466919, timestamp=1681674243.7411017) [('tile_y', [-1, 256]), ('tile_x', [-1, 8])],None,38
+ No: 5 GFLOPS: 8.88/14.75 result: MeasureResult(costs=(0.0302279796,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.7226266860961914, timestamp=1681674245.9529366) [('tile_y', [-1, 1]), ('tile_x', [-1, 32])],None,50
+ No: 6 GFLOPS: 10.07/14.75 result: MeasureResult(costs=(0.026652299,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.7300765514373779, timestamp=1681674246.6403675) [('tile_y', [-1, 16]), ('tile_x', [-1, 32])],None,54
+ No: 7 GFLOPS: 10.29/14.75 result: MeasureResult(costs=(0.026077315399999995,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.6747543811798096, timestamp=1681674247.3188195) [('tile_y', [-1, 2]), ('tile_x', [-1, 64])],None,61
+ No: 8 GFLOPS: 10.36/14.75 result: MeasureResult(costs=(0.0259077772,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.6880345344543457, timestamp=1681674247.9964845) [('tile_y', [-1, 8]), ('tile_x', [-1, 32])],None,53
+ No: 9 GFLOPS: 11.13/14.75 result: MeasureResult(costs=(0.024123681,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.6246585845947266, timestamp=1681674248.7397454) [('tile_y', [-1, 256]), ('tile_x', [-1, 512])],None,98
+ No: 10 GFLOPS: 2.46/14.75 result: MeasureResult(costs=(0.1090250572,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.9821467399597168, timestamp=1681674250.744545) [('tile_y', [-1, 2]), ('tile_x', [-1, 4])],None,21
diff --git a/docs/_sources/tutorial/autotvm_relay_x86.rst.txt b/docs/_sources/tutorial/autotvm_relay_x86.rst.txt
index 2b8d41f41e..218b8841b1 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': 509.8071522700002, 'median': 509.3824774999973, 'std': 2.8923425636596294}
+ {'mean': 515.4545443699999, 'median': 515.1880160499957, 'std': 2.972500288970956}
@@ -582,31 +582,30 @@ 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: 11.26/ 18.06 GFLOPS | Progress: (4/20) | 11.17 s
[Task 1/25] Current/Best: 7.28/ 23.68 GFLOPS | Progress: (8/20) | 14.45 s
[Task 1/25] Current/Best: 23.86/ 23.86 GFLOPS | Progress: (12/20) | 18.42 s
[Task 1/25] Current/Best: 11.03/ 23.86 GFLOPS | Progress: (16/20) | 22.12 s
[Task 1/25] Current/Best: 12.55/ 23.86 GFLOPS | Progress: (20/20) | 25.89 s Done.
-
[Task 2/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 2/25] Current/Best: 14.99/ 14.99 GFLOPS | Progress: (4/20) | 4.31 s
[Task 2/25] Current/Best: 9.30/ 19.10 GFLOPS | Progress: (8/20) | 5.97 s
[Task 2/25] Current/Best: 10.93/ 19.10 GFLOPS | Progress: (12/20) | 8.12 s
[Task 2/25] Current/Best: 13.77/ 23.42 GFLOPS | Progress: (16/20) | 9.74 s
[Task 2/25] Current/Best: 4.96/ 23.42 GFLOPS | Progress: (20/20) | 12.08 s Done.
-
[Task 3/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 3/25] Current/Best: 16.42/ 16.42 GFLOPS | Progress: (4/20) | 4.95 s
[Task 3/25] Current/Best: 13.70/ 16.42 GFLOPS | Progress: (8/20) | 7.54 s
[Task 3/25] Current/Best: 9.32/ 16.78 GFLOPS | Progress: (12/20) | 10.10 s
[Task 3/25] Current/Best: 10.25/ 16.78 GFLOPS | Progress: (16/20) | 12.80 s
[Task 3/25] Current/Best: 10.90/ 16.78 GFLOPS | Progress: (20/20) | 16.60 s Done.
-
[Task 4/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 4/25] Current/Best: 12.51/ 16.58 GFLOPS | Progress: (4/20) | 4.69 s
[Task 4/25] Current/Best: 10.26/ 16.58 GFLOPS | Progress: (8/20) | 9.82 s
[Task 4/25] Current/Best: 17.47/ 20.53 GFLOPS | Progress: (12/20) | 11.49 s
[Task 4/25] Current/Best: 17.35/ 20.53 GFLOPS | Progress: (16/20) | 13.70 s
[Task 4/25] Current/Best: 18.74/ 20.53 GFLOPS | Progress: (20/20) | 16.84 s Done.
-
[Task 5/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 5/25] Current/Best: 2.90/ 14.13 GFLOPS | Progress: (4/20) | 5.32 s
[Task 5/25] Current/Best: 18.30/ 18.30 GFLOPS | Progress: (8/20) | 7.70 s
[Task 5/25] Current/Best: 18.54/ 19.99 GFLOPS | Progress: (12/20) | 9.71 s
[Task 5/25] Current/Best: 18.34/ 19.99 GFLOPS | Progress: (16/20) | 11.89 s
[Task 5/25] Current/Best: 18.17/ 19.99 GFLOPS | Progress: (20/20) | 13.72 s Done.
-
[Task 6/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 6/25] Current/Best: 17.90/ 19.08 GFLOPS | Progress: (4/20) | 4.58 s
[Task 6/25] Current/Best: 16.56/ 19.08 GFLOPS | Progress: (8/20) | 6.95 s
[Task 6/25] Current/Best: 14.58/ 20.46 GFLOPS | Progress: (12/20) | 9.76 s
[Task 6/25] Current/Best: 13.86/ 20.46 GFLOPS | Progress: (16/20) | 12.70 s
[Task 6/25] Current/Best: 6.20/ 20.46 GFLOPS | Progress: (20/20) | 15.85 s Done.
-
[Task 7/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 7/25] Current/Best: 18.85/ 18.85 GFLOPS | Progress: (4/20) | 5.28 s
[Task 7/25] Current/Best: 7.96/ 18.85 GFLOPS | Progress: (8/20) | 7.88 s
[Task 7/25] Current/Best: 12.89/ 19.99 GFLOPS | Progress: (12/20) | 10.15 s
[Task 7/25] Current/Best: 7.91/ 19.99 GFLOPS | Progress: (16/20) | 13.21 s
[Task 7/25] Current/Best: 12.58/ 19.99 GFLOPS | Progress: (20/20) | 17.23 s Done.
-
[Task 8/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 8/25] Current/Best: 13.72/ 13.72 GFLOPS | Progress: (4/20) | 5.94 s
[Task 8/25] Current/Best: 8.76/ 17.78 GFLOPS | Progress: (8/20) | 10.87 s
[Task 8/25] Current/Best: 8.28/ 17.78 GFLOPS | Progress: (12/20) | 19.63 s
[Task 8/25] Current/Best: 4.21/ 17.78 GFLOPS | Progress: (16/20) | 24.04 s
[Task 8/25] Current/Best: 5.67/ 17.78 GFLOPS | Progress: (20/20) | 27.06 s Done.
-
[Task 9/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 9/25] Current/Best: 14.22/ 14.22 GFLOPS | Progress: (4/20) | 6.53 s
[Task 9/25] Current/Best: 11.66/ 16.12 GFLOPS | Progress: (8/20) | 10.22 s
[Task 9/25] Current/Best: 11.69/ 16.12 GFLOPS | Progress: (12/20) | 15.59 s
[Task 9/25] Current/Best: 6.89/ 16.12 GFLOPS | Progress: (16/20) | 26.87 s
[Task 9/25] Current/Best: 3.46/ 22.21 GFLOPS | Progress: (20/20) | 29.89 s
[Task 10/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 10/25] Current/Best: 6.15/ 22.77 GFLOPS | Progress: (4/20) | 4.41 s Done.
-
[Task 10/25] Current/Best: 13.72/ 22.77 GFLOPS | Progress: (8/20) | 6.58 s
[Task 10/25] Current/Best: 5.32/ 22.77 GFLOPS | Progress: (12/20) | 8.84 s
[Task 10/25] Current/Best: 8.82/ 22.77 GFLOPS | Progress: (16/20) | 11.05 s
[Task 10/25] Current/Best: 5.03/ 22.77 GFLOPS | Progress: (20/20) | 12.87 s Done.
-
[Task 11/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 11/25] Current/Best: 11.39/ 19.50 GFLOPS | Progress: (4/20) | 5.55 s
[Task 11/25] Current/Best: 14.45/ 19.50 GFLOPS | Progress: (8/20) | 8.51 s
[Task 11/25] Current/Best: 12.52/ 19.50 GFLOPS | Progress: (12/20) | 10.59 s
[Task 11/25] Current/Best: 13.47/ 19.50 GFLOPS | Progress: (16/20) | 14.56 s
[Task 11/25] Current/Best: 21.44/ 21.44 GFLOPS | Progress: (20/20) | 16.57 s Done.
-
[Task 12/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 12/25] Current/Best: 14.41/ 18.45 GFLOPS | Progress: (4/20) | 5.47 s
[Task 12/25] Current/Best: 18.21/ 18.45 GFLOPS | Progress: (8/20) | 10.29 s
[Task 12/25] Current/Best: 13.98/ 18.45 GFLOPS | Progress: (12/20) | 13.77 s
[Task 12/25] Current/Best: 14.36/ 18.45 GFLOPS | Progress: (16/20) | 15.92 s
[Task 12/25] Current/Best: 1.64/ 18.45 GFLOPS | Progress: (20/20) | 19.77 s Done.
-
[Task 13/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 13/25] Current/Best: 21.02/ 21.02 GFLOPS | Progress: (4/20) | 5.94 s
[Task 13/25] Current/Best: 18.96/ 21.02 GFLOPS | Progress: (8/20) | 8.58 s
[Task 13/25] Current/Best: 6.18/ 22.16 GFLOPS | Progress: (12/20) | 12.54 s
[Task 13/25] Current/Best: 5.28/ 22.16 GFLOPS | Progress: (16/20) | 15.64 s
[Task 13/25] Current/Best: 12.48/ 22.16 GFLOPS | Progress: (20/20) | 17.90 s Done.
-
[Task 14/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 14/25] Current/Best: 14.11/ 19.64 GFLOPS | Progress: (4/20) | 5.26 s
[Task 14/25] Current/Best: 1.61/ 19.64 GFLOPS | Progress: (8/20) | 9.12 s
[Task 14/25] Current/Best: 3.03/ 19.64 GFLOPS | Progress: (12/20) | 11.80 s
[Task 14/25] Current/Best: 6.44/ 19.64 GFLOPS | Progress: (16/20) | 13.87 s
[Task 14/25] Current/Best: 5.89/ 19.64 GFLOPS | Progress: (20/20) | 21.45 s
[Task 15/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 15/25] Current/Best: 16.38/ 21.59 GFLOPS | Progress: (4/20) | 4.77 s
[Task 15/25] Current/Best: 5.87/ 22.12 GFLOPS | Progress: (8/20) | 6.78 s
[Task 15/25] Current/Best: 18.55/ 22.12 GFLOPS | Progress: (12/20) | 8.77 s
[Task 15/25] Current/Best: 8.88/ 22.12 GFLOPS | Progress: (16/20) | 11.77 s
[Task 15/25] Current/Best: 13.40/ 22.12 GFLOPS | Progress: (20/20)
| 14.22 s
[Task 16/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 16/25] Current/Best: 12.99/ 15.76 GFLOPS | Progress: (4/20) | 6.20 s
[Task 16/25] Current/Best: 10.58/ 17.89 GFLOPS | Progress: (8/20) | 8.70 s
[Task 16/25] Current/Best: 9.03/ 17.99 GFLOPS | Progress: (12/20) | 10.33 s
[Task 16/25] Current/Best: 17.39/ 17.99 GFLOPS | Progress: (16/20) | 12.05 s
[Task 16/25] Current/Best: 13.47/ 23.58 GFLOPS | Progress: (20/20) | 13.57 s Done.
-
[Task 17/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 17/25] Current/Best: 22.64/ 22.64 GFLOPS | Progress: (4/20) | 5.37 s
[Task 17/25] Current/Best: 10.63/ 22.64 GFLOPS | Progress: (8/20) | 8.62 s
[Task 17/25] Current/Best: 19.83/ 22.64 GFLOPS | Progress: (12/20) | 10.50 s
[Task 17/25] Current/Best: 20.94/ 22.64 GFLOPS | Progress: (16/20) | 12.77 s
[Task 17/25] Current/Best: 19.99/ 22.64 GFLOPS | Progress: (20/20) | 15.56 s Done.
-
[Task 18/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 18/25] Current/Best: 12.78/ 13.64 GFLOPS | Progress: (4/20) | 5.47 s
[Task 18/25] Current/Best: 9.18/ 18.28 GFLOPS | Progress: (8/20) | 9.88 s
[Task 18/25] Current/Best: 6.77/ 18.28 GFLOPS | Progress: (12/20) | 12.46 s
[Task 18/25] Current/Best: 6.58/ 18.28 GFLOPS | Progress: (16/20) | 15.89 s
[Task 18/25] Current/Best: 10.27/ 18.28 GFLOPS | Progress: (20/20) | 19.99 s Done.
-
[Task 19/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 19/25] Current/Best: 22.98/ 22.98 GFLOPS | Progress: (4/20) | 6.18 s
[Task 19/25] Current/Best: 13.98/ 22.98 GFLOPS | Progress: (8/20) | 10.56 s
[Task 19/25] Current/Best: 11.72/ 22.98 GFLOPS | Progress: (12/20) | 14.38 s
[Task 19/25] Current/Best: 1.58/ 22.98 GFLOPS | Progress: (16/20) | 19.22 s
[Task 19/25] Current/Best: 6.59/ 22.98 GFLOPS | Progress: (20/20) | 22.80 s Done.
-
[Task 20/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 20/25] Current/Best: 12.88/ 20.16 GFLOPS | Progress: (4/20) | 4.80 s
[Task 20/25] Current/Best: 14.35/ 20.16 GFLOPS | Progress: (8/20) | 8.04 s Done.
- Done.
-
[Task 20/25] Current/Best: 21.44/ 21.44 GFLOPS | Progress: (12/20) | 11.95 s
[Task 20/25] Current/Best: 17.09/ 21.44 GFLOPS | Progress: (16/20) | 14.83 s
[Task 20/25] Current/Best: 13.29/ 21.44 GFLOPS | Progress: (20/20) | 19.29 s
[Task 21/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 21/25] Current/Best: 4.82/ 16.74 GFLOPS | Progress: (4/20) | 4.43 s
[Task 21/25] Current/Best: 4.66/ 16.74 GFLOPS | Progress: (8/20) | 7.07 s
[Task 21/25] Current/Best: 5.48/ 19.27 GFLOPS | Progress: (12/20) | 10.56 s
[Task 21/25] Current/Best: 7.69/ 19.27 GFLOPS | Progress: (16/20) | 13.85 s
[Task 21/25] Current/Best: 17.46/ 19.27 GFLOPS | Progress: (20/20) | 15.39 s
[Task 22/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 22/25] Current/Best: 5.40/ 11.10 GFLOPS | Progress: (4/20) | 8.85 s
[Task 22/25] Current/Best: 5.35/ 18.57 GFLOPS | Progress: (8/2
0) | 12.84 s Done.
- Done.
-
[Task 22/25] Current/Best: 9.26/ 18.57 GFLOPS | Progress: (12/20) | 14.66 s
[Task 22/25] Current/Best: 14.41/ 18.57 GFLOPS | Progress: (16/20) | 17.21 s
[Task 22/25] Current/Best: 11.31/ 18.57 GFLOPS | Progress: (20/20) | 19.49 s Done.
-
[Task 23/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 23/25] Current/Best: 21.22/ 21.22 GFLOPS | Progress: (4/20) | 6.88 s
[Task 23/25] Current/Best: 10.37/ 22.79 GFLOPS | Progress: (8/20) | 9.58 s
[Task 23/25] Current/Best: 11.03/ 22.79 GFLOPS | Progress: (12/20) | 13.76 s
[Task 23/25] Current/Best: 10.10/ 22.79 GFLOPS | Progress: (16/20) | 16.29 s
[Task 23/25] Current/Best: 1.57/ 22.79 GFLOPS | Progress: (20/20) | 20.77 s Done.
-
[Task 24/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 24/25] Current/Best: 5.93/ 5.93 GFLOPS | Progress: (4/20) | 8.75 s
[Task 24/25] Current/Best: 1.74/ 9.65 GFLOPS | Progress: (8/20) | 19.56 s
[Task 24/25] Current/Best: 5.77/ 10.28 GFLOPS | Progress: (12/20) | 22.85 s
[Task 24/25] Current/Best: 3.78/ 10.28 GFLOPS | Progress: (16/20) | 33.81 s
[Task 24/25] Current/Best: 8.29/ 10.28 GFLOPS | Progress: (20/20) | 45.77 s
[Task 25/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s Done.
-
[Task 25/25] Current/Best: 2.62/ 8.00 GFLOPS | Progress: (4/20) | 15.58 s
[Task 25/25] Current/Best: 5.69/ 8.14 GFLOPS | Progress: (8/20) | 26.25 s
[Task 25/25] Current/Best: 6.53/ 8.14 GFLOPS | Progress: (12/20) | 38.79 s
[Task 25/25] Current/Best: 9.97/ 9.97 GFLOPS | Progress: (16/20) | 43.75 s
[Task 25/25] Current/Best: 8.09/ 10.77 GFLOPS | Progress: (20/20) | 49.69 s
+
[Task 1/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 1/25] Current/Best: 14.97/ 15.03 GFLOPS | Progress: (4/20) | 11.05 s
[Task 1/25] Current/Best: 14.62/ 22.80 GFLOPS | Progress: (8/20) | 14.57 s
[Task 1/25] Current/Best: 14.32/ 22.80 GFLOPS | Progress: (12/20) | 17.66 s
[Task 1/25] Current/Best: 17.45/ 22.80 GFLOPS | Progress: (16/20) | 21.68 s
[Task 1/25] Current/Best: 23.65/ 23.65 GFLOPS | Progress: (20/20) | 25.29 s Done.
+
[Task 2/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 2/25] Current/Best: 12.05/ 15.56 GFLOPS | Progress: (4/20) | 4.49 s
[Task 2/25] Current/Best: 8.53/ 15.56 GFLOPS | Progress: (8/20) | 6.14 s
[Task 2/25] Current/Best: 14.33/ 17.73 GFLOPS | Progress: (12/20) | 7.65 s
[Task 2/25] Current/Best: 12.68/ 17.73 GFLOPS | Progress: (16/20) | 9.52 s
[Task 2/25] Current/Best: 16.46/ 17.73 GFLOPS | Progress: (20/20) | 11.73 s Done.
+
[Task 3/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 3/25] Current/Best: 10.98/ 15.42 GFLOPS | Progress: (4/20) | 5.89 s
[Task 3/25] Current/Best: 3.20/ 17.65 GFLOPS | Progress: (8/20) | 8.72 s
[Task 3/25] Current/Best: 8.16/ 17.65 GFLOPS | Progress: (12/20) | 11.36 s
[Task 3/25] Current/Best: 14.69/ 17.65 GFLOPS | Progress: (16/20) | 13.66 s
[Task 3/25] Current/Best: 8.87/ 17.65 GFLOPS | Progress: (20/20) | 15.83 s Done.
+
[Task 4/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 4/25] Current/Best: 16.29/ 16.29 GFLOPS | Progress: (4/20) | 4.95 s
[Task 4/25] Current/Best: 17.53/ 17.53 GFLOPS | Progress: (8/20) | 7.93 s
[Task 4/25] Current/Best: 12.68/ 17.53 GFLOPS | Progress: (12/20) | 9.86 s
[Task 4/25] Current/Best: 15.93/ 20.84 GFLOPS | Progress: (16/20) | 12.08 s
[Task 4/25] Current/Best: 9.66/ 20.84 GFLOPS | Progress: (20/20) | 17.32 s Done.
+
[Task 5/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 5/25] Current/Best: 13.03/ 13.03 GFLOPS | Progress: (4/20) | 5.02 s
[Task 5/25] Current/Best: 17.93/ 19.98 GFLOPS | Progress: (8/20) | 6.87 s
[Task 5/25] Current/Best: 9.99/ 19.98 GFLOPS | Progress: (12/20) | 8.87 s
[Task 5/25] Current/Best: 14.64/ 19.98 GFLOPS | Progress: (16/20) | 11.28 s
[Task 5/25] Current/Best: 2.69/ 19.98 GFLOPS | Progress: (20/20) | 13.78 s Done.
+
[Task 6/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 6/25] Current/Best: 15.32/ 15.68 GFLOPS | Progress: (4/20) | 5.12 s
[Task 6/25] Current/Best: 8.07/ 16.11 GFLOPS | Progress: (8/20) | 8.35 s
[Task 6/25] Current/Best: 13.47/ 17.58 GFLOPS | Progress: (12/20) | 10.72 s
[Task 6/25] Current/Best: 17.77/ 17.77 GFLOPS | Progress: (16/20) | 13.14 s
[Task 6/25] Current/Best: 12.53/ 17.77 GFLOPS | Progress: (20/20) | 15.94 s Done.
+
[Task 7/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 7/25] Current/Best: 15.27/ 19.44 GFLOPS | Progress: (4/20) | 5.44 s
[Task 7/25] Current/Best: 16.67/ 19.44 GFLOPS | Progress: (8/20) | 8.28 s
[Task 7/25] Current/Best: 13.38/ 19.44 GFLOPS | Progress: (12/20) | 10.63 s
[Task 7/25] Current/Best: 15.50/ 19.44 GFLOPS | Progress: (16/20) | 13.14 s
[Task 7/25] Current/Best: 10.37/ 19.44 GFLOPS | Progress: (20/20) | 15.89 s Done.
+
[Task 8/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 8/25] Current/Best: 13.89/ 13.89 GFLOPS | Progress: (4/20) | 14.38 s
[Task 8/25] Current/Best: 11.46/ 13.89 GFLOPS | Progress: (8/20) | 18.18 s
[Task 8/25] Current/Best: 14.30/ 14.30 GFLOPS | Progress: (12/20) | 21.21 s
[Task 8/25] Current/Best: 18.39/ 19.43 GFLOPS | Progress: (16/20) | 23.34 s
[Task 8/25] Current/Best: 10.40/ 19.43 GFLOPS | Progress: (20/20) | 32.32 s
[Task 9/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 9/25] Current/Best: 22.98/ 22.98 GFLOPS | Progress: (4/20) | 7.48 s Done.
+
[Task 9/25] Current/Best: 21.97/ 22.98 GFLOPS | Progress: (8/20) | 9.47 s
[Task 9/25] Current/Best: 10.44/ 22.98 GFLOPS | Progress: (12/20) | 15.46 s
[Task 9/25] Current/Best: 17.06/ 22.98 GFLOPS | Progress: (16/20) | 21.49 s
[Task 9/25] Current/Best: 8.65/ 22.98 GFLOPS | Progress: (20/20) | 23.49 s Done.
+
[Task 10/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 10/25] Current/Best: 15.38/ 15.38 GFLOPS | Progress: (4/20) | 5.72 s
[Task 10/25] Current/Best: 14.04/ 18.59 GFLOPS | Progress: (8/20) | 9.23 s
[Task 10/25] Current/Best: 18.25/ 18.59 GFLOPS | Progress: (12/20) | 10.83 s
[Task 10/25] Current/Best: 20.71/ 20.71 GFLOPS | Progress: (16/20) | 12.98 s
[Task 10/25] Current/Best: 11.69/ 20.71 GFLOPS | Progress: (20/20) | 15.07 s Done.
+
[Task 11/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 11/25] Current/Best: 8.75/ 18.95 GFLOPS | Progress: (4/20) | 5.17 s
[Task 11/25] Current/Best: 11.61/ 18.95 GFLOPS | Progress: (8/20) | 7.93 s
[Task 11/25] Current/Best: 18.11/ 22.95 GFLOPS | Progress: (12/20) | 10.46 s
[Task 11/25] Current/Best: 14.33/ 22.95 GFLOPS | Progress: (16/20) | 13.26 s
[Task 11/25] Current/Best: 19.89/ 22.95 GFLOPS | Progress: (20/20) | 15.41 s Done.
+
[Task 12/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 12/25] Current/Best: 12.75/ 12.75 GFLOPS | Progress: (4/20) | 5.41 s
[Task 12/25] Current/Best: 3.13/ 12.75 GFLOPS | Progress: (8/20) | 11.11 s
[Task 12/25] Current/Best: 12.78/ 16.44 GFLOPS | Progress: (12/20) | 14.32 s
[Task 12/25] Current/Best: 9.95/ 16.77 GFLOPS | Progress: (16/20) | 17.35 s
[Task 12/25] Current/Best: 10.67/ 16.77 GFLOPS | Progress: (20/20) | 20.58 s Done.
+
[Task 13/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 13/25] Current/Best: 12.15/ 18.23 GFLOPS | Progress: (4/20) | 5.97 s
[Task 13/25] Current/Best: 11.77/ 18.23 GFLOPS | Progress: (8/20) | 10.01 s
[Task 13/25] Current/Best: 17.23/ 22.64 GFLOPS | Progress: (12/20) | 13.13 s
[Task 13/25] Current/Best: 12.20/ 22.64 GFLOPS | Progress: (16/20) | 18.41 s
[Task 13/25] Current/Best: 3.08/ 22.64 GFLOPS | Progress: (20/20) | 21.77 s Done.
+
[Task 14/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 14/25] Current/Best: 16.82/ 16.82 GFLOPS | Progress: (4/20) | 4.80 s
[Task 14/25] Current/Best: 13.44/ 16.82 GFLOPS | Progress: (8/20) | 13.41 s
[Task 14/25] Current/Best: 8.31/ 18.13 GFLOPS | Progress: (12/20) | 15.40 s
[Task 14/25] Current/Best: 11.76/ 18.13 GFLOPS | Progress: (16/20) | 19.25 s
[Task 14/25] Current/Best: 13.82/ 18.13 GFLOPS | Progress: (20/20) | 22.76 s
[Task 15/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 15/25] Current/Best: 4.93/ 15.88 GFLOPS | Progress: (4/20) | 6.59 s
[Task 15/25] Current/Best: 13.44/ 19.57 GFLOPS | Progress: (8/20) | 9.42 s
[Task 15/25] Current/Best: 15.85/ 19.75 GFLOPS | Progress: (12/20) | 15.83 s
[Task 15/25] Current/Best: 20.28/ 20.28 GFLOPS | Progress: (16/20) | 18.24 s
[Task 15/25] Current/Best: 22.05/ 22.05 GFLOPS | Progress: (20/2
0) | 21.59 s Done.
+
[Task 16/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 16/25] Current/Best: 10.30/ 15.23 GFLOPS | Progress: (4/20) | 5.01 s
[Task 16/25] Current/Best: 11.17/ 16.81 GFLOPS | Progress: (8/20) | 8.04 s
[Task 16/25] Current/Best: 3.05/ 19.53 GFLOPS | Progress: (12/20) | 9.90 s
[Task 16/25] Current/Best: 12.21/ 19.53 GFLOPS | Progress: (16/20) | 12.30 s
[Task 16/25] Current/Best: 17.33/ 19.53 GFLOPS | Progress: (20/20) | 13.97 s Done.
+
[Task 17/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 17/25] Current/Best: 12.28/ 18.18 GFLOPS | Progress: (4/20) | 5.36 s
[Task 17/25] Current/Best: 15.07/ 22.50 GFLOPS | Progress: (8/20) | 7.96 s
[Task 17/25] Current/Best: 9.10/ 22.50 GFLOPS | Progress: (12/20) | 11.51 s
[Task 17/25] Current/Best: 6.43/ 22.50 GFLOPS | Progress: (16/20) | 14.75 s
[Task 17/25] Current/Best: 11.78/ 22.50 GFLOPS | Progress: (20/20) | 19.34 s Done.
+
[Task 18/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 18/25] Current/Best: 15.38/ 15.38 GFLOPS | Progress: (4/20) | 5.45 s
[Task 18/25] Current/Best: 15.90/ 21.35 GFLOPS | Progress: (8/20) | 7.62 s
[Task 18/25] Current/Best: 8.40/ 21.35 GFLOPS | Progress: (12/20) | 15.82 s
[Task 18/25] Current/Best: 6.00/ 21.35 GFLOPS | Progress: (16/20) | 25.99 s
[Task 18/25] Current/Best: 9.36/ 21.35 GFLOPS | Progress: (20/20) | 33.00 s Done.
+
[Task 19/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 19/25] Current/Best: 8.62/ 22.12 GFLOPS | Progress: (4/20) | 5.48 s
[Task 19/25] Current/Best: 13.80/ 22.12 GFLOPS | Progress: (8/20) | 7.74 s
[Task 19/25] Current/Best: 21.11/ 23.47 GFLOPS | Progress: (12/20) | 12.25 s
[Task 19/25] Current/Best: 19.41/ 23.47 GFLOPS | Progress: (16/20) | 15.00 s
[Task 19/25] Current/Best: 10.22/ 23.47 GFLOPS | Progress: (20/20) | 17.85 s Done.
+
[Task 20/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 20/25] Current/Best: 11.00/ 16.21 GFLOPS | Progress: (4/20) | 5.52 s
[Task 20/25] Current/Best: 3.49/ 16.21 GFLOPS | Progress: (8/20) | 9.85 s
[Task 20/25] Current/Best: 13.16/ 16.28 GFLOPS | Progress: (12/20) | 15.08 s
[Task 20/25] Current/Best: 8.67/ 16.87 GFLOPS | Progress: (16/20) | 18.22 s
[Task 20/25] Current/Best: 16.93/ 17.95 GFLOPS | Progress: (20/20) | 20.23 s Done.
+
[Task 21/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 21/25] Current/Best: 6.32/ 6.32 GFLOPS | Progress: (4/20) | 4.43 s
[Task 21/25] Current/Best: 10.63/ 18.93 GFLOPS | Progress: (8/20) | 8.07 s Done.
+
[Task 21/25] Current/Best: 16.32/ 20.53 GFLOPS | Progress: (12/20) | 11.02 s
[Task 21/25] Current/Best: 2.74/ 20.53 GFLOPS | Progress: (16/20) | 14.71 s
[Task 21/25] Current/Best: 3.14/ 20.53 GFLOPS | Progress: (20/20) | 17.61 s
[Task 22/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 22/25] Current/Best: 12.32/ 16.70 GFLOPS | Progress: (4/20) | 5.89 s
[Task 22/25] Current/Best: 5.18/ 20.45 GFLOPS | Progress: (8/20) | 8.36 s
[Task 22/25] Current/Best: 6.81/ 20.45 GFLOPS | Progress: (12/20) | 12.19 s
[Task 22/25] Current/Best: 16.41/ 20.45 GFLOPS | Progress: (16/20) | 13.86 s
[Task 22/25] Current/Best: 6.92/ 20.45 GFLOPS | Progress: (20/20) | 16.69 s Done.
+
[Task 23/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 23/25] Current/Best: 9.29/ 18.56 GFLOPS | Progress: (4/20) | 5.61 s
[Task 23/25] Current/Best: 3.08/ 18.56 GFLOPS | Progress: (8/20) | 11.84 s
[Task 23/25] Current/Best: 13.69/ 19.23 GFLOPS | Progress: (12/20) | 16.14 s
[Task 23/25] Current/Best: 11.70/ 19.23 GFLOPS | Progress: (16/20) | 19.52 s
[Task 23/25] Current/Best: 10.48/ 20.03 GFLOPS | Progress: (20/20) | 25.04 s Done.
+
[Task 24/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 24/25] Current/Best: 1.73/ 3.52 GFLOPS | Progress: (4/20) | 13.48 s
[Task 24/25] Current/Best: 3.65/ 3.65 GFLOPS | Progress: (8/20) | 24.44 s
[Task 24/25] Current/Best: 8.38/ 8.38 GFLOPS | Progress: (12/20) | 35.44 s
[Task 24/25] Current/Best: 2.73/ 8.38 GFLOPS | Progress: (16/20) | 48.18 s Done.
+
[Task 24/25] Current/Best: 1.87/ 10.10 GFLOPS | Progress: (20/20) | 58.87 s
[Task 25/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 25/25] Current/Best: 2.54/ 7.99 GFLOPS | Progress: (4/20) | 6.76 s
[Task 25/25] Current/Best: 8.04/ 8.04 GFLOPS | Progress: (8/20) | 17.43 s
[Task 25/25] Current/Best: 5.82/ 8.41 GFLOPS | Progress: (12/20) | 20.21 s
[Task 25/25] Current/Best: 4.92/ 8.41 GFLOPS | Progress: (16/20) | 25.03 s
[Task 25/25] Current/Best: 5.48/ 8.41 GFLOPS | Progress: (20/20) | 36.02 s
@@ -703,7 +702,7 @@ 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=0.356379
+ class='n02123159 tiger cat' with probability=0.356378
class='n02124075 Egyptian cat' with probability=0.019712
class='n02129604 tiger, Panthera tigris' with probability=0.001215
class='n04040759 radiator' with probability=0.000262
@@ -760,8 +759,8 @@ improvement in comparing the optimized model to the unoptimized model.
.. code-block:: none
- optimized: {'mean': 409.2955450000022, 'median': 408.3869461000063, 'std': 2.8880201223233275}
- unoptimized: {'mean': 509.8071522700002, 'median': 509.3824774999973, 'std': 2.8923425636596294}
+ optimized: {'mean': 427.5091489600004, 'median': 427.1601727499956, 'std': 4.275411909121886}
+ unoptimized: {'mean': 515.4545443699999, 'median': 515.1880160499957, 'std': 2.972500288970956}
@@ -784,7 +783,7 @@ profiling/benchmarking.
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 12 minutes 30.527 seconds)
+ **Total running time of the script:** ( 13 minutes 2.787 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 fd54a79edb..8667222837 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
- 1.276e-07 secs/op
+ 1.181e-07 secs/op
diff --git a/docs/_sources/tutorial/intro_topi.rst.txt b/docs/_sources/tutorial/intro_topi.rst.txt
index bd975fc3dd..50d83da8f9 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, 0x5306b20)), stage(b, placeholder(b, 0x8b65a50)), 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.R [...]
+ [stage(a, placeholder(a, 0x1b587790)), stage(b, placeholder(b, 0x22e368a0)), 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 4bd3664c3c..51500a1bd9 100644
--- a/docs/_sources/tutorial/sg_execution_times.rst.txt
+++ b/docs/_sources/tutorial/sg_execution_times.rst.txt
@@ -5,29 +5,29 @@
Computation times
=================
-**15:59.420** total execution time for **tutorial** files:
+**16:37.810** total execution time for **tutorial** files:
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_autotvm_relay_x86.py` (``autotvm_relay_x86.py``) | 12:30.527 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_autotvm_relay_x86.py` (``autotvm_relay_x86.py``) | 13:02.787 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_auto_scheduler_matmul_x86.py` (``auto_scheduler_matmul_x86.py``) | 01:24.173 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_auto_scheduler_matmul_x86.py` (``auto_scheduler_matmul_x86.py``) | 01:33.956 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_tensor_expr_get_started.py` (``tensor_expr_get_started.py``) | 00:58.155 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_tensor_expr_get_started.py` (``tensor_expr_get_started.py``) | 01:01.094 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_relay_quick_start.py` (``relay_quick_start.py``) | 00:35.970 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_relay_quick_start.py` (``relay_quick_start.py``) | 00:37.168 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_autotvm_matmul_x86.py` (``autotvm_matmul_x86.py``) | 00:27.695 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_autotvm_matmul_x86.py` (``autotvm_matmul_x86.py``) | 00:20.100 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_tensor_ir_blitz_course.py` (``tensor_ir_blitz_course.py``) | 00:01.893 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_tensor_ir_blitz_course.py` (``tensor_ir_blitz_course.py``) | 00:01.668 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_intro_topi.py` (``intro_topi.py``) | 00:00.841 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_intro_topi.py` (``intro_topi.py``) | 00:00.855 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_cross_compilation_and_rpc.py` (``cross_compilation_and_rpc.py``) | 00:00.166 | 0.0 MB |
-+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_introduction.py` (``introduction.py``) | 00:00.000 | 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_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 |
diff --git a/docs/_sources/tutorial/tensor_expr_get_started.rst.txt b/docs/_sources/tutorial/tensor_expr_get_started.rst.txt
index 64dbbfa887..cfea3bcacb 100644
--- a/docs/_sources/tutorial/tensor_expr_get_started.rst.txt
+++ b/docs/_sources/tutorial/tensor_expr_get_started.rst.txt
@@ -285,8 +285,8 @@ helper function to run a profile of the TVM generated code.
.. code-block:: none
- Numpy running time: 0.000006
- naive: 0.000009
+ 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: 0.000008
+ parallel: 0.000007
@@ -444,7 +444,7 @@ factor to be the number of threads on your CPU.
.. code-block:: none
- vector: 0.000043
+ vector: 0.000040
# 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 6.390940000073897e-06 1.0
- naive 8.651900000000001e-06 1.3537758138708798
- parallel 7.872299999999999e-06 1.231790628594381
- vector 4.2759699999999997e-05 6.690674611169182
+ numpy 6.925449999926059e-06 1.0
+ naive 6.7094000000000005e-06 0.9688034712649192
+ parallel 6.9884e-06 1.0090896620543954
+ vector 4.03558e-05 5.827173685526697
@@ -922,7 +922,7 @@ matrix multiplication.
.. code-block:: none
- Numpy running time: 0.018258
+ Numpy running time: 0.018210
@@ -980,7 +980,7 @@ optimizations.
.. code-block:: none
- none: 3.225776
+ none: 3.418322
@@ -1080,7 +1080,7 @@ schedule.
.. code-block:: none
- blocking: 0.289393
+ blocking: 0.308912
@@ -1164,7 +1164,7 @@ already cache friendly from our previous optimizations.
.. code-block:: none
- vectorization: 0.324637
+ vectorization: 0.335391
# 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: 0.113190
+ loop permutation: 0.116179
# 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: 0.107750
+ array packing: 0.108260
# 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: 0.108013
+ block caching: 0.110334
# 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: 0.143291
+ parallelization: 0.146405
# 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 3.2257764807999996 1.0
- blocking 0.28939330209999997 0.08971275716792064
- vectorization 0.3246370389 0.10063841708570251
- loop permutation 0.1131898727 0.03508918654894795
- array packing 0.1077499762 0.03340280296583902
- block caching 0.1080131402 0.033484384563809734
- parallelization 0.1432905827 0.0444204933456715
+ none 3.4183219973 1.0
+ blocking 0.3089119253 0.09036946359763579
+ vectorization 0.3353911454 0.09811572627298201
+ loop permutation 0.11617881919999999 0.03398709053499498
+ array packing 0.1082596446 0.0316704057386958
+ block caching 0.11033356269999998 0.03227711221679765
+ parallelization 0.1464052159 0.04282955672860538
@@ -1594,6 +1594,11 @@ 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 1.094 seconds)
+
+
.. _sphx_glr_download_tutorial_tensor_expr_get_started.py:
.. only:: html
diff --git a/docs/commit_hash b/docs/commit_hash
index c07e8635d8..74de98e5d2 100644
--- a/docs/commit_hash
+++ b/docs/commit_hash
@@ -1 +1 @@
-62bffbb4cdf8302bfca7af492cd668207d040b6d
+0d51fbbecde0cefc4da8ef5c171c2fbd22655e30
diff --git a/docs/how_to/compile_models/from_darknet.html b/docs/how_to/compile_models/from_darknet.html
index a4617adacd..7961ebb8c7 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 19.028 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 20.948 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_mxnet.html b/docs/how_to/compile_models/from_mxnet.html
index 09042d95bc..a00c0f1c28 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.zip14f4d7c9-c966-4257-bf78-b7900fd652e8 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.zipfe19e3f0-490e-4e8c-b4a5-cffbb6c3277d 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 47b4508f61..df6b18f689 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, 58.8MB/s]
- 41%|#### | 16.9M/41.5M [00:00<00:00, 76.0MB/s]
- 59%|#####8 | 24.4M/41.5M [00:00<00:00, 75.8MB/s]
- 77%|#######7 | 32.0M/41.5M [00:00<00:00, 71.2MB/s]
- 94%|#########3| 38.9M/41.5M [00:00<00:00, 62.1MB/s]
-100%|##########| 41.5M/41.5M [00:00<00:00, 61.4MB/s]
+ 19%|#9 | 7.99M/41.5M [00:00<00:00, 47.9MB/s]
+ 39%|###8 | 16.0M/41.5M [00:00<00:00, 49.3MB/s]
+ 62%|######2 | 25.8M/41.5M [00:00<00:00, 66.9MB/s]
+ 88%|########7 | 36.3M/41.5M [00:00<00:00, 80.8MB/s]
+100%|##########| 41.5M/41.5M [00:00<00:00, 65.0MB/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 e233ee1929..065254dfe5 100644
--- a/docs/how_to/compile_models/from_pytorch.html
+++ b/docs/how_to/compile_models/from_pytorch.html
@@ -437,13 +437,12 @@ 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%|#7 | 7.99M/44.7M [00:00<00:00, 73.4MB/s]
- 34%|###3 | 15.0M/44.7M [00:00<00:00, 57.3MB/s]
- 49%|####8 | 21.8M/44.7M [00:00<00:00, 62.7MB/s]
- 63%|######2 | 28.0M/44.7M [00:00<00:00, 56.1MB/s]
- 86%|########5 | 38.3M/44.7M [00:00<00:00, 48.8MB/s]
- 99%|#########8| 44.2M/44.7M [00:00<00:00, 50.4MB/s]
-100%|##########| 44.7M/44.7M [00:00<00:00, 53.7MB/s]
+ 18%|#7 | 7.99M/44.7M [00:00<00:00, 77.3MB/s]
+ 36%|###5 | 16.0M/44.7M [00:00<00:00, 80.4MB/s]
+ 58%|#####8 | 26.1M/44.7M [00:00<00:00, 83.8MB/s]
+ 76%|#######6 | 34.0M/44.7M [00:00<00:00, 83.1MB/s]
+ 96%|#########6| 43.0M/44.7M [00:00<00:00, 86.8MB/s]
+100%|##########| 44.7M/44.7M [00:00<00:00, 87.0MB/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 6375302d4f..c309bed728 100644
--- a/docs/how_to/compile_models/from_tensorflow.html
+++ b/docs/how_to/compile_models/from_tensorflow.html
@@ -657,7 +657,7 @@ 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 32.584 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 32.337 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 c1afeafe80..6a672f7750 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">ΒΆ</a></h1>
-<p><strong>06:46.135</strong> total execution time for <strong>how_to_compile_models</strong> files:</p>
+<p><strong>06:57.086</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>01:32.584</p></td>
+<td><p>01:32.337</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>01:19.028</p></td>
+<td><p>01:20.948</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>00:56.006</p></td>
+<td><p>00:58.825</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>00:37.565</p></td>
+<td><p>00:38.579</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>00:31.604</p></td>
+<td><p>00:33.235</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>00:29.816</p></td>
+<td><p>00:30.773</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>00:28.016</p></td>
+<td><p>00:28.641</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>00:26.519</p></td>
+<td><p>00:26.838</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>00:22.363</p></td>
+<td><p>00:24.163</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>00:02.633</p></td>
+<td><p>00:02.747</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 7746c3e2b0..b2eb67b459 100644
--- a/docs/how_to/deploy_models/deploy_model_on_adreno.html
+++ b/docs/how_to/deploy_models/deploy_model_on_adreno.html
@@ -835,7 +835,7 @@ Top5 predictions:
Evaluate inference time cost...
Execution time summary:
mean (ms) median (ms) max (ms) min (ms) std (ms)
- 2639.3694 2638.2291 2644.0245 2636.1597 2.7886
+ 2679.1327 2678.2908 2682.9453 2676.3431 2.4263
</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 5a5122bfba..b8fa7a14ae 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
@@ -443,25 +443,28 @@ 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
- 7249920/102967424 [=>............................] - ETA: 0s
- 10575872/102967424 [==>...........................] - ETA: 1s
- 13828096/102967424 [===>..........................] - ETA: 1s
- 16482304/102967424 [===>..........................] - ETA: 1s
- 25616384/102967424 [======>.......................] - ETA: 1s
- 35422208/102967424 [=========>....................] - ETA: 0s
- 46432256/102967424 [============>.................] - ETA: 0s
- 52477952/102967424 [==============>...............] - ETA: 0s
- 58228736/102967424 [===============>..............] - ETA: 0s
- 63225856/102967424 [=================>............] - ETA: 0s
- 68239360/102967424 [==================>...........] - ETA: 0s
- 73261056/102967424 [====================>.........] - ETA: 0s
- 78217216/102967424 [=====================>........] - ETA: 0s
- 83255296/102967424 [=======================>......] - ETA: 0s
- 84746240/102967424 [=======================>......] - ETA: 0s
- 92995584/102967424 [==========================>...] - ETA: 0s
- 98312192/102967424 [===========================>..] - ETA: 0s
+ 8380416/102967424 [=>............................] - ETA: 1s
+ 15024128/102967424 [===>..........................] - ETA: 1s
+ 16769024/102967424 [===>..........................] - ETA: 1s
+ 23412736/102967424 [=====>........................] - ETA: 1s
+ 25157632/102967424 [======>.......................] - ETA: 1s
+ 33546240/102967424 [========>.....................] - ETA: 1s
+ 40189952/102967424 [==========>...................] - ETA: 1s
+ 41934848/102967424 [===========>..................] - ETA: 1s
+ 50323456/102967424 [=============>................] - ETA: 0s
+ 58712064/102967424 [================>.............] - ETA: 0s
+ 65355776/102967424 [==================>...........] - ETA: 0s
+ 67100672/102967424 [==================>...........] - ETA: 0s
+ 69296128/102967424 [===================>..........] - ETA: 0s
+ 75489280/102967424 [====================>.........] - ETA: 0s
+ 82124800/102967424 [======================>.......] - ETA: 0s
+ 83877888/102967424 [=======================>......] - ETA: 0s
+ 90521600/102967424 [=========================>....] - ETA: 0s
+ 92266496/102967424 [=========================>....] - ETA: 0s
+ 94191616/102967424 [==========================>...] - ETA: 0s
+100646912/102967424 [============================>.] - ETA: 0s
102850560/102967424 [============================>.] - ETA: 0s
-102967424/102967424 [==============================] - 1s 0us/step
+102967424/102967424 [==============================] - 2s 0us/step
</pre></div>
</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 dab28091e4..9fa171b932 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)
- 14.5475 14.4685 15.0415 14.4153 0.1812
+ 16.1558 15.9476 17.0428 15.8976 0.4230
</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 2994ecaaef..a65f0fc769 100644
--- a/docs/how_to/deploy_models/deploy_object_detection_pytorch.html
+++ b/docs/how_to/deploy_models/deploy_object_detection_pytorch.html
@@ -459,33 +459,28 @@ 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]
- 5%|4 | 7.99M/170M [00:00<00:03, 52.9MB/s]
- 8%|8 | 14.3M/170M [00:00<00:02, 55.5MB/s]
- 12%|#1 | 19.6M/170M [00:00<00:03, 46.3MB/s]
- 14%|#4 | 24.1M/170M [00:00<00:03, 46.3MB/s]
- 19%|#8 | 32.0M/170M [00:00<00:03, 47.5MB/s]
- 24%|##3 | 40.0M/170M [00:00<00:02, 53.7MB/s]
- 28%|##8 | 48.0M/170M [00:00<00:02, 55.1MB/s]
- 33%|###2 | 56.0M/170M [00:01<00:02, 57.5MB/s]
- 37%|###6 | 62.3M/170M [00:01<00:02, 49.1MB/s]
- 40%|###9 | 67.2M/170M [00:01<00:02, 49.4MB/s]
- 42%|####2 | 72.0M/170M [00:01<00:02, 48.7MB/s]
- 47%|####7 | 80.0M/170M [00:01<00:02, 46.7MB/s]
- 50%|####9 | 84.5M/170M [00:01<00:02, 44.5MB/s]
- 52%|#####2 | 88.8M/170M [00:01<00:02, 40.2MB/s]
- 57%|#####6 | 96.0M/170M [00:02<00:01, 45.9MB/s]
- 61%|######1 | 104M/170M [00:02<00:01, 44.0MB/s]
- 66%|######6 | 112M/170M [00:02<00:01, 53.3MB/s]
- 70%|######9 | 118M/170M [00:02<00:01, 53.7MB/s]
- 74%|#######3 | 125M/170M [00:02<00:00, 58.0MB/s]
- 77%|#######7 | 131M/170M [00:02<00:00, 47.9MB/s]
- 80%|######## | 136M/170M [00:02<00:00, 42.2MB/s]
- 85%|########5 | 145M/170M [00:03<00:00, 54.0MB/s]
- 89%|########9 | 152M/170M [00:03<00:00, 48.4MB/s]
- 93%|#########3| 158M/170M [00:03<00:00, 50.8MB/s]
- 96%|#########6| 164M/170M [00:03<00:00, 48.2MB/s]
- 99%|#########9| 169M/170M [00:03<00:00, 48.3MB/s]
-100%|##########| 170M/170M [00:03<00:00, 49.5MB/s]
+ 5%|4 | 7.99M/170M [00:00<00:03, 48.5MB/s]
+ 9%|9 | 16.0M/170M [00:00<00:03, 45.9MB/s]
+ 14%|#4 | 24.0M/170M [00:00<00:03, 47.0MB/s]
+ 19%|#8 | 32.0M/170M [00:00<00:02, 53.4MB/s]
+ 24%|##3 | 40.0M/170M [00:00<00:02, 60.1MB/s]
+ 30%|##9 | 50.6M/170M [00:00<00:01, 73.9MB/s]
+ 38%|###7 | 64.0M/170M [00:00<00:01, 92.3MB/s]
+ 43%|####3 | 73.5M/170M [00:01<00:01, 73.4MB/s]
+ 48%|####7 | 81.4M/170M [00:01<00:01, 73.1MB/s]
+ 52%|#####2 | 89.0M/170M [00:01<00:01, 64.4MB/s]
+ 56%|#####6 | 95.8M/170M [00:01<00:01, 62.8MB/s]
+ 60%|###### | 102M/170M [00:01<00:01, 48.4MB/s]
+ 63%|######3 | 108M/170M [00:01<00:01, 46.4MB/s]
+ 67%|######6 | 113M/170M [00:02<00:01, 48.7MB/s]
+ 71%|####### | 120M/170M [00:02<00:00, 53.9MB/s]
+ 74%|#######4 | 126M/170M [00:02<00:00, 56.7MB/s]
+ 79%|#######9 | 134M/170M [00:02<00:00, 63.7MB/s]
+ 85%|########4 | 144M/170M [00:02<00:00, 62.5MB/s]
+ 89%|########9 | 152M/170M [00:02<00:00, 67.1MB/s]
+ 94%|#########4| 160M/170M [00:02<00:00, 62.3MB/s]
+ 98%|#########7| 166M/170M [00:02<00:00, 59.8MB/s]
+100%|##########| 170M/170M [00:02<00:00, 60.1MB/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=& [...]
@@ -583,7 +578,7 @@ 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 24.688 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 3 minutes 38.754 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 85dc6a8066..cc1b3f067c 100644
--- a/docs/how_to/deploy_models/deploy_prequantized.html
+++ b/docs/how_to/deploy_models/deploy_prequantized.html
@@ -500,10 +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]
- 47%|####6 | 6.30M/13.6M [00:00<00:00, 40.3MB/s]
- 75%|#######4 | 10.2M/13.6M [00:00<00:00, 32.6MB/s]
- 98%|#########7| 13.2M/13.6M [00:00<00:00, 31.6MB/s]
-100%|##########| 13.6M/13.6M [00:00<00:00, 33.5MB/s]
+ 59%|#####8 | 7.99M/13.6M [00:00<00:00, 54.8MB/s]
+100%|##########| 13.6M/13.6M [00:00<00:00, 79.6MB/s]
</pre></div>
</div>
</div>
@@ -594,7 +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)
- 88.9007 88.8215 92.7114 88.4916 0.4311
+ 90.1986 90.1078 94.0665 89.9954 0.4309
</pre></div>
</div>
<div class="admonition note">
@@ -633,7 +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">ΒΆ</a></h2>
<p>TODO</p>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 15.619 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 19.336 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 929a9f4079..ba067bbef4 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)
- 114.4671 114.2424 119.0063 113.4299 0.8437
+ 119.8069 119.4192 155.3561 118.6067 3.6082
</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 34.480 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 2 minutes 36.460 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 a78bb99209..b3d17b4500 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 39.726 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 38.680 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 042dbd73cd..7be9fc43c3 100644
--- a/docs/how_to/deploy_models/deploy_ssd_gluoncv.html
+++ b/docs/how_to/deploy_models/deploy_ssd_gluoncv.html
@@ -468,23 +468,25 @@ 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 | 4959/132723 [00:00<00:02, 49583.07KB/s]
- 10%|9 | 13134/132723 [00:00<00:01, 68502.69KB/s]
- 16%|#6 | 21402/132723 [00:00<00:01, 74975.06KB/s]
- 22%|##2 | 29630/132723 [00:00<00:01, 77855.61KB/s]
- 29%|##8 | 37962/132723 [00:00<00:01, 79825.08KB/s]
- 35%|###4 | 46265/132723 [00:00<00:01, 80913.76KB/s]
- 41%|####1 | 54510/132723 [00:00<00:00, 81413.65KB/s]
- 47%|####7 | 62777/132723 [00:00<00:00, 81811.70KB/s]
- 54%|#####3 | 71113/132723 [00:00<00:00, 82294.62KB/s]
- 60%|#####9 | 79443/132723 [00:01<00:00, 82604.16KB/s]
- 66%|######6 | 87771/132723 [00:01<00:00, 82810.17KB/s]
- 72%|#######2 | 96053/132723 [00:01<00:00, 82780.65KB/s]
- 79%|#######8 | 104332/132723 [00:01<00:00, 79493.09KB/s]
- 85%|########4 | 112309/132723 [00:01<00:00, 70754.81KB/s]
- 90%|######### | 119568/132723 [00:01<00:00, 51841.35KB/s]
- 96%|#########6| 127808/132723 [00:01<00:00, 58567.59KB/s]
-100%|##########| 132723/132723 [00:01<00:00, 71581.58KB/s]
+ 4%|3 | 5216/132723 [00:00<00:02, 52110.85KB/s]
+ 9%|9 | 12469/132723 [00:00<00:01, 64115.06KB/s]
+ 15%|#4 | 19783/132723 [00:00<00:01, 68231.77KB/s]
+ 20%|## | 27099/132723 [00:00<00:01, 70174.57KB/s]
+ 26%|##5 | 34365/132723 [00:00<00:01, 71068.11KB/s]
+ 31%|###1 | 41777/132723 [00:00<00:01, 72093.28KB/s]
+ 37%|###6 | 49038/132723 [00:00<00:01, 72259.21KB/s]
+ 42%|####2 | 56372/132723 [00:00<00:01, 72601.32KB/s]
+ 48%|####8 | 63712/132723 [00:00<00:00, 72734.40KB/s]
+ 54%|#####3 | 71159/132723 [00:01<00:00, 73268.73KB/s]
+ 59%|#####9 | 78629/132723 [00:01<00:00, 73701.27KB/s]
+ 65%|######4 | 86152/132723 [00:01<00:00, 74016.32KB/s]
+ 71%|####### | 93749/132723 [00:01<00:00, 74603.37KB/s]
+ 76%|#######6 | 101347/132723 [00:01<00:00, 75017.03KB/s]
+ 82%|########2 | 108849/132723 [00:01<00:00, 74753.18KB/s]
+ 88%|########7 | 116480/132723 [00:01<00:00, 75218.67KB/s]
+ 93%|#########3| 124003/132723 [00:01<00:00, 74792.79KB/s]
+ 99%|#########9| 131998/132723 [00:01<00:00, 76330.82KB/s]
+100%|##########| 132723/132723 [00:01<00:00, 73109.71KB/s]
</pre></div>
</div>
<p>Create TVM runtime and do inference
@@ -523,7 +525,7 @@ 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> ( 3 minutes 52.236 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> ( 4 minutes 2.527 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 d086bdf487..62ac37bacf 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">ΒΆ</a></h1>
-<p><strong>16:10.207</strong> total execution time for <strong>how_to_deploy_models</strong> files:</p>
+<p><strong>16:48.689</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>03:52.236</p></td>
+<td><p>04:02.527</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>03:24.688</p></td>
+<td><p>03:38.754</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>02:34.480</p></td>
+<td><p>02:36.460</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>01:39.726</p></td>
+<td><p>01:38.680</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>01:15.619</p></td>
+<td><p>01:19.336</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 Adrenoβ’</span></a> (<code class="docutils literal notranslate"><span class="pre">deploy_model_on_adreno.py</span></code>)</p></td>
-<td><p>00:55.581</p></td>
+<td><p>00:57.116</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 Adrenoβ’ 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>00:50.240</p></td>
+<td><p>00:53.472</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>00:41.837</p></td>
+<td><p>00:43.972</p></td>
<td><p>0.0 MB</p></td>
</tr>
<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>00:28.124</p></td>
+<td><p>00:29.303</p></td>
<td><p>0.0 MB</p></td>
</tr>
<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>00:27.672</p></td>
+<td><p>00:29.063</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 c93b4f9569..4c27a20810 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.zipb9b51972-6301-4f46-b7b0-f27570bb132d 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.zip0953cac7-8696-4cb7-8883-8dfc06a203a6 from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/mobilenet0.25-9f83e440.zip...
</pre></div>
</div>
<p>Itβs 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 eddf2b1fac..c6060aed29 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">ΒΆ</a></h1>
-<p><strong>00:52.436</strong> total execution time for <strong>how_to_extend_tvm</strong> files:</p>
+<p><strong>00:55.939</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>00:48.822</p></td>
+<td><p>00:52.041</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>00:02.596</p></td>
+<td><p>00:02.794</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>00:01.011</p></td>
+<td><p>00:01.097</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>00:00.007</p></td>
+<td><p>00:00.008</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 be1f95b903..4deedbd019 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: 22011us [22011us] (48.70%; 48.70%)
-FoldScaleAxis: 23182us [6us] (51.30%; 51.30%)
- FoldConstant: 23176us [1688us] (51.28%; 99.97%)
- InferType: 21488us [21488us] (47.55%; 92.72%)
+InferType: 22554us [22554us] (47.83%; 47.83%)
+FoldScaleAxis: 24604us [7us] (52.17%; 52.17%)
+ FoldConstant: 24597us [1674us] (52.16%; 99.97%)
+ InferType: 22923us [22923us] (48.61%; 93.19%)
</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: 21512us [21512us] (48.59%; 48.59%)
-FoldScaleAxis: 22759us [4us] (51.41%; 51.41%)
- FoldConstant: 22754us [1566us] (51.40%; 99.98%)
- InferType: 21188us [21188us] (47.86%; 93.12%)
+InferType: 22398us [22398us] (48.48%; 48.48%)
+FoldScaleAxis: 23806us [6us] (51.52%; 51.52%)
+ FoldConstant: 23800us [1770us] (51.51%; 99.97%)
+ InferType: 22031us [22031us] (47.68%; 92.56%)
</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 429fc11371..f21fdd4ccf 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: 53.520542 ms
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Convolution: 53.530464 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 66d3cd5d49..7f9c6bc2e9 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: 11.141734 ms
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>conv2d with tensor core: 12.245008 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 1a0ab986c2..dfeeacea3f 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: 0.015934
-Baseline: 3.237478
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Numpy running time: 0.018680
+Baseline: 3.376687
</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: 0.288589
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt1: 0.304087
</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: 0.323071
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt2: 0.343316
</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: 0.111656
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt3: 0.124391
</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: 0.107130
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt4: 0.110405
</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: 0.100487
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt5: 0.111236
</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: 0.133279
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt6: 0.147561
</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 63cfe70735..bdbc509625 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">ΒΆ</a></h1>
-<p><strong>00:33.258</strong> total execution time for <strong>how_to_optimize_operators</strong> files:</p>
+<p><strong>00:35.503</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>00:30.380</p></td>
+<td><p>00:32.496</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>00:01.800</p></td>
+<td><p>00:01.864</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>00:01.077</p></td>
+<td><p>00:01.143</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 f86285a3fc..3c2af8d619 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">ΒΆ</a></h1>
-<p><strong>10:18.423</strong> total execution time for <strong>how_to_tune_with_autoscheduler</strong> files:</p>
+<p><strong>10:29.512</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>06:30.519</p></td>
+<td><p>06:17.263</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>01:40.103</p></td>
+<td><p>01:45.435</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>01:09.444</p></td>
+<td><p>01:12.706</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>00:31.296</p></td>
+<td><p>00:45.695</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>00:13.816</p></td>
+<td><p>00:14.585</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>00:13.244</p></td>
+<td><p>00:13.829</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 fdb2c40c8d..1122756096 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
@@ -510,12 +510,12 @@ 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": T.bool(True), "global_symbol": "main", "tir.noalias": T.bool(True)})
- blockIdx_x = T.launch_thread("blockIdx.x", 32)
- conv2d_nchw = T.allocate([7], "float32", "local")
- pad_temp_shared = T.allocate([6272], "float32", "shared")
- kernel_shared = T.allocate([2048], "float32", "shared")
- threadIdx_x = T.launch_thread("threadIdx.x", 112)
- conv2d_nchw_1 = T.Buffer((1,), data=conv2d_nchw, scope="local", align=4)
+ blockIdx_x = T.launch_thread("blockIdx.x", 16)
+ conv2d_nchw = T.allocate([28], "float32", "local")
+ pad_temp_shared = T.allocate([1008], "float32", "shared")
+ kernel_shared = T.allocate([1536], "float32", "shared")
+ threadIdx_x = T.launch_thread("threadIdx.x", 56)
+ conv2d_nchw_1 = T.Buffer((28,), data=conv2d_nchw, scope="local")
conv2d_nchw_1[0] = T.float32(0)
conv2d_nchw_1[1] = T.float32(0)
conv2d_nchw_1[2] = T.float32(0)
@@ -523,1094 +523,360 @@ class Module:
conv2d_nchw_1[4] = T.float32(0)
conv2d_nchw_1[5] = T.float32(0)
conv2d_nchw_1[6] = T.float32(0)
- for rc_outer_outer, ry_outer_outer, rx_outer_outer in T.grid(4, 3, 3):
+ conv2d_nchw_1[7] = T.float32(0)
+ conv2d_nchw_1[8] = T.float32(0)
+ conv2d_nchw_1[9] = T.float32(0)
+ conv2d_nchw_1[10] = T.float32(0)
+ conv2d_nchw_1[11] = T.float32(0)
+ conv2d_nchw_1[12] = T.float32(0)
+ conv2d_nchw_1[13] = T.float32(0)
+ conv2d_nchw_1[14] = T.float32(0)
+ conv2d_nchw_1[15] = T.float32(0)
+ conv2d_nchw_1[16] = T.float32(0)
+ conv2d_nchw_1[17] = T.float32(0)
+ conv2d_nchw_1[18] = T.float32(0)
+ conv2d_nchw_1[19] = T.float32(0)
+ conv2d_nchw_1[20] = T.float32(0)
+ conv2d_nchw_1[21] = T.float32(0)
+ conv2d_nchw_1[22] = T.float32(0)
+ conv2d_nchw_1[23] = T.float32(0)
+ conv2d_nchw_1[24] = T.float32(0)
+ conv2d_nchw_1[25] = T.float32(0)
+ conv2d_nchw_1[26] = T.float32(0)
+ conv2d_nchw_1[27] = T.float32(0)
+ for rc_outer_outer, ry_outer_outer in T.grid(32, 3):
+ cse_var_4: T.int32 = rc_outer_outer * 784
+ cse_var_3: T.int32 = ry_outer_outer * 7
+ cse_var_2: T.int32 = rc_outer_outer * 144
+ cse_var_1: T.int32 = ry_outer_outer * 3
threadIdx_x_1 = T.env_thread("threadIdx.x")
- pad_temp_shared_1 = T.Buffer((6272,), data=pad_temp_shared, scope="shared")
+ pad_temp_shared_1 = T.Buffer((1008,), data=pad_temp_shared, scope="shared")
data_1 = T.Buffer((25088,), data=data.data)
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1] = T.if_then_else(1 <= threadIdx_x_1 % 49 // 7 + ry_outer_outer and threadIdx_x_1 % 49 // 7 + ry_outer_outer < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer - 8], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 112] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 104], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 224] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 216], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 336] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 328], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 448] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 440], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 560] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 552], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 672] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 664], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 784] = T.if_then_else(1 <= threadIdx_x_1 % 49 // 7 + ry_outer_outer and threadIdx_x_1 % 49 // 7 + ry_outer_outer < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 776], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 896] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 888], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 1008] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 1000], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 1120] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 1112], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 1232] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 1224], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 1344] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 1336], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 1456] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 1448], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 1568] = T.if_then_else(1 <= threadIdx_x_1 % 49 // 7 + ry_outer_outer and threadIdx_x_1 % 49 // 7 + ry_outer_outer < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 1560], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 1680] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 1672], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 1792] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 1784], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 1904] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 1896], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 2016] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 2008], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 2128] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 2120], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 2240] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 2232], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 2352] = T.if_then_else(1 <= threadIdx_x_1 % 49 // 7 + ry_outer_outer and threadIdx_x_1 % 49 // 7 + ry_outer_outer < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 2344], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 2464] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 2456], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 2576] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 2568], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 2688] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 2680], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 2800] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 2792], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 2912] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 2904], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 3024] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 3016], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 3136] = T.if_then_else(1 <= threadIdx_x_1 % 49 // 7 + ry_outer_outer and threadIdx_x_1 % 49 // 7 + ry_outer_outer < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 3128], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 3248] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 3240], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 3360] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 3352], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 3472] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 3464], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 3584] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 3576], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 3696] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 3688], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 3808] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 3800], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 3920] = T.if_then_else(1 <= threadIdx_x_1 % 49 // 7 + ry_outer_outer and threadIdx_x_1 % 49 // 7 + ry_outer_outer < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 3912], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 4032] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 4024], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 4144] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 4136], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 4256] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 4248], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 4368] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 4360], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 4480] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 4472], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 4592] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 4584], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 4704] = T.if_then_else(1 <= threadIdx_x_1 % 49 // 7 + ry_outer_outer and threadIdx_x_1 % 49 // 7 + ry_outer_outer < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 4696], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 4816] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 4808], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 4928] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 4920], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 5040] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 5032], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 5152] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 5144], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 5264] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 5256], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 5376] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 5368], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 5488] = T.if_then_else(1 <= threadIdx_x_1 % 49 // 7 + ry_outer_outer and threadIdx_x_1 % 49 // 7 + ry_outer_outer < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 5480], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 5600] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 5592], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 5712] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 5704], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 5824] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 5816], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 5936] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 5928], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 6048] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 6040], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 6160] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 6272 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 6152], T.float32(0))
- kernel_shared_1 = T.Buffer((2048,), data=kernel_shared, scope="shared")
- with T.launch_thread("threadIdx.x", 112) as threadIdx_x_2:
- kernel_1 = T.Buffer((2359296,), data=kernel.data)
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 1] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 9]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 2] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 18]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 3] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 27]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 4] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 36]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 5] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 45]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 6] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 54]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 7] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 63]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 8] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 72]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 9] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 81]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 10] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 90]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 11] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 99]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 12] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 108]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 13] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 117]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 14] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 126]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 15] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 135]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 16] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 144]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 17] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 153]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 18] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 162]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 19] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 171]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 20] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 180]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 21] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 189]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 22] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 198]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 23] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 207]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 24] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 216]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 25] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 225]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 26] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 234]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 27] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 243]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 28] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 252]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 29] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 261]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 30] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 270]
- if T.likely(threadIdx_x_2 < 64):
- kernel_shared_1[threadIdx_x_2 * 32 + 31] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 // 4 * 4608 + rc_outer_outer * 1152 + threadIdx_x_2 % 4 * 288 + ry_outer_outer * 3 + rx_outer_outer + 279]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7] * kernel_shared_1[threadIdx_x // 7 * 128]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 7] * kernel_shared_1[threadIdx_x // 7 * 128]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 14] * kernel_shared_1[threadIdx_x // 7 * 128]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 21] * kernel_shared_1[threadIdx_x // 7 * 128]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 28] * kernel_shared_1[threadIdx_x // 7 * 128]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 35] * kernel_shared_1[threadIdx_x // 7 * 128]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 42] * kernel_shared_1[threadIdx_x // 7 * 128]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 49] * kernel_shared_1[threadIdx_x // 7 * 128 + 1]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 56] * kernel_shared_1[threadIdx_x // 7 * 128 + 1]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 63] * kernel_shared_1[threadIdx_x // 7 * 128 + 1]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 70] * kernel_shared_1[threadIdx_x // 7 * 128 + 1]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 77] * kernel_shared_1[threadIdx_x // 7 * 128 + 1]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 84] * kernel_shared_1[threadIdx_x // 7 * 128 + 1]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 91] * kernel_shared_1[threadIdx_x // 7 * 128 + 1]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 98] * kernel_shared_1[threadIdx_x // 7 * 128 + 2]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 105] * kernel_shared_1[threadIdx_x // 7 * 128 + 2]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 112] * kernel_shared_1[threadIdx_x // 7 * 128 + 2]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 119] * kernel_shared_1[threadIdx_x // 7 * 128 + 2]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 126] * kernel_shared_1[threadIdx_x // 7 * 128 + 2]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 133] * kernel_shared_1[threadIdx_x // 7 * 128 + 2]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 140] * kernel_shared_1[threadIdx_x // 7 * 128 + 2]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 147] * kernel_shared_1[threadIdx_x // 7 * 128 + 3]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 154] * kernel_shared_1[threadIdx_x // 7 * 128 + 3]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 161] * kernel_shared_1[threadIdx_x // 7 * 128 + 3]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 168] * kernel_shared_1[threadIdx_x // 7 * 128 + 3]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 175] * kernel_shared_1[threadIdx_x // 7 * 128 + 3]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 182] * kernel_shared_1[threadIdx_x // 7 * 128 + 3]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 189] * kernel_shared_1[threadIdx_x // 7 * 128 + 3]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 196] * kernel_shared_1[threadIdx_x // 7 * 128 + 4]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 203] * kernel_shared_1[threadIdx_x // 7 * 128 + 4]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 210] * kernel_shared_1[threadIdx_x // 7 * 128 + 4]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 217] * kernel_shared_1[threadIdx_x // 7 * 128 + 4]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 224] * kernel_shared_1[threadIdx_x // 7 * 128 + 4]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 231] * kernel_shared_1[threadIdx_x // 7 * 128 + 4]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 238] * kernel_shared_1[threadIdx_x // 7 * 128 + 4]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 245] * kernel_shared_1[threadIdx_x // 7 * 128 + 5]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 252] * kernel_shared_1[threadIdx_x // 7 * 128 + 5]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 259] * kernel_shared_1[threadIdx_x // 7 * 128 + 5]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 266] * kernel_shared_1[threadIdx_x // 7 * 128 + 5]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 273] * kernel_shared_1[threadIdx_x // 7 * 128 + 5]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 280] * kernel_shared_1[threadIdx_x // 7 * 128 + 5]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 287] * kernel_shared_1[threadIdx_x // 7 * 128 + 5]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 294] * kernel_shared_1[threadIdx_x // 7 * 128 + 6]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 301] * kernel_shared_1[threadIdx_x // 7 * 128 + 6]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 308] * kernel_shared_1[threadIdx_x // 7 * 128 + 6]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 315] * kernel_shared_1[threadIdx_x // 7 * 128 + 6]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 322] * kernel_shared_1[threadIdx_x // 7 * 128 + 6]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 329] * kernel_shared_1[threadIdx_x // 7 * 128 + 6]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 336] * kernel_shared_1[threadIdx_x // 7 * 128 + 6]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 343] * kernel_shared_1[threadIdx_x // 7 * 128 + 7]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 350] * kernel_shared_1[threadIdx_x // 7 * 128 + 7]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 357] * kernel_shared_1[threadIdx_x // 7 * 128 + 7]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 364] * kernel_shared_1[threadIdx_x // 7 * 128 + 7]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 371] * kernel_shared_1[threadIdx_x // 7 * 128 + 7]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 378] * kernel_shared_1[threadIdx_x // 7 * 128 + 7]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 385] * kernel_shared_1[threadIdx_x // 7 * 128 + 7]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 392] * kernel_shared_1[threadIdx_x // 7 * 128 + 8]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 399] * kernel_shared_1[threadIdx_x // 7 * 128 + 8]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 406] * kernel_shared_1[threadIdx_x // 7 * 128 + 8]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 413] * kernel_shared_1[threadIdx_x // 7 * 128 + 8]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 420] * kernel_shared_1[threadIdx_x // 7 * 128 + 8]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 427] * kernel_shared_1[threadIdx_x // 7 * 128 + 8]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 434] * kernel_shared_1[threadIdx_x // 7 * 128 + 8]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 441] * kernel_shared_1[threadIdx_x // 7 * 128 + 9]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 448] * kernel_shared_1[threadIdx_x // 7 * 128 + 9]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 455] * kernel_shared_1[threadIdx_x // 7 * 128 + 9]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 462] * kernel_shared_1[threadIdx_x // 7 * 128 + 9]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 469] * kernel_shared_1[threadIdx_x // 7 * 128 + 9]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 476] * kernel_shared_1[threadIdx_x // 7 * 128 + 9]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 483] * kernel_shared_1[threadIdx_x // 7 * 128 + 9]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 490] * kernel_shared_1[threadIdx_x // 7 * 128 + 10]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 497] * kernel_shared_1[threadIdx_x // 7 * 128 + 10]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 504] * kernel_shared_1[threadIdx_x // 7 * 128 + 10]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 511] * kernel_shared_1[threadIdx_x // 7 * 128 + 10]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 518] * kernel_shared_1[threadIdx_x // 7 * 128 + 10]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 525] * kernel_shared_1[threadIdx_x // 7 * 128 + 10]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 532] * kernel_shared_1[threadIdx_x // 7 * 128 + 10]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 539] * kernel_shared_1[threadIdx_x // 7 * 128 + 11]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 546] * kernel_shared_1[threadIdx_x // 7 * 128 + 11]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 553] * kernel_shared_1[threadIdx_x // 7 * 128 + 11]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 560] * kernel_shared_1[threadIdx_x // 7 * 128 + 11]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 567] * kernel_shared_1[threadIdx_x // 7 * 128 + 11]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 574] * kernel_shared_1[threadIdx_x // 7 * 128 + 11]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 581] * kernel_shared_1[threadIdx_x // 7 * 128 + 11]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 588] * kernel_shared_1[threadIdx_x // 7 * 128 + 12]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 595] * kernel_shared_1[threadIdx_x // 7 * 128 + 12]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 602] * kernel_shared_1[threadIdx_x // 7 * 128 + 12]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 609] * kernel_shared_1[threadIdx_x // 7 * 128 + 12]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 616] * kernel_shared_1[threadIdx_x // 7 * 128 + 12]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 623] * kernel_shared_1[threadIdx_x // 7 * 128 + 12]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 630] * kernel_shared_1[threadIdx_x // 7 * 128 + 12]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 637] * kernel_shared_1[threadIdx_x // 7 * 128 + 13]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 644] * kernel_shared_1[threadIdx_x // 7 * 128 + 13]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 651] * kernel_shared_1[threadIdx_x // 7 * 128 + 13]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 658] * kernel_shared_1[threadIdx_x // 7 * 128 + 13]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 665] * kernel_shared_1[threadIdx_x // 7 * 128 + 13]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 672] * kernel_shared_1[threadIdx_x // 7 * 128 + 13]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 679] * kernel_shared_1[threadIdx_x // 7 * 128 + 13]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 686] * kernel_shared_1[threadIdx_x // 7 * 128 + 14]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 693] * kernel_shared_1[threadIdx_x // 7 * 128 + 14]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 700] * kernel_shared_1[threadIdx_x // 7 * 128 + 14]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 707] * kernel_shared_1[threadIdx_x // 7 * 128 + 14]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 714] * kernel_shared_1[threadIdx_x // 7 * 128 + 14]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 721] * kernel_shared_1[threadIdx_x // 7 * 128 + 14]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 728] * kernel_shared_1[threadIdx_x // 7 * 128 + 14]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 735] * kernel_shared_1[threadIdx_x // 7 * 128 + 15]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 742] * kernel_shared_1[threadIdx_x // 7 * 128 + 15]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 749] * kernel_shared_1[threadIdx_x // 7 * 128 + 15]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 756] * kernel_shared_1[threadIdx_x // 7 * 128 + 15]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 763] * kernel_shared_1[threadIdx_x // 7 * 128 + 15]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 770] * kernel_shared_1[threadIdx_x // 7 * 128 + 15]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 777] * kernel_shared_1[threadIdx_x // 7 * 128 + 15]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 784] * kernel_shared_1[threadIdx_x // 7 * 128 + 16]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 791] * kernel_shared_1[threadIdx_x // 7 * 128 + 16]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 798] * kernel_shared_1[threadIdx_x // 7 * 128 + 16]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 805] * kernel_shared_1[threadIdx_x // 7 * 128 + 16]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 812] * kernel_shared_1[threadIdx_x // 7 * 128 + 16]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 819] * kernel_shared_1[threadIdx_x // 7 * 128 + 16]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 826] * kernel_shared_1[threadIdx_x // 7 * 128 + 16]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 833] * kernel_shared_1[threadIdx_x // 7 * 128 + 17]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 840] * kernel_shared_1[threadIdx_x // 7 * 128 + 17]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 847] * kernel_shared_1[threadIdx_x // 7 * 128 + 17]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 854] * kernel_shared_1[threadIdx_x // 7 * 128 + 17]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 861] * kernel_shared_1[threadIdx_x // 7 * 128 + 17]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 868] * kernel_shared_1[threadIdx_x // 7 * 128 + 17]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 875] * kernel_shared_1[threadIdx_x // 7 * 128 + 17]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 882] * kernel_shared_1[threadIdx_x // 7 * 128 + 18]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 889] * kernel_shared_1[threadIdx_x // 7 * 128 + 18]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 896] * kernel_shared_1[threadIdx_x // 7 * 128 + 18]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 903] * kernel_shared_1[threadIdx_x // 7 * 128 + 18]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 910] * kernel_shared_1[threadIdx_x // 7 * 128 + 18]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 917] * kernel_shared_1[threadIdx_x // 7 * 128 + 18]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 924] * kernel_shared_1[threadIdx_x // 7 * 128 + 18]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 931] * kernel_shared_1[threadIdx_x // 7 * 128 + 19]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 938] * kernel_shared_1[threadIdx_x // 7 * 128 + 19]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 945] * kernel_shared_1[threadIdx_x // 7 * 128 + 19]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 952] * kernel_shared_1[threadIdx_x // 7 * 128 + 19]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 959] * kernel_shared_1[threadIdx_x // 7 * 128 + 19]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 966] * kernel_shared_1[threadIdx_x // 7 * 128 + 19]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 973] * kernel_shared_1[threadIdx_x // 7 * 128 + 19]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 980] * kernel_shared_1[threadIdx_x // 7 * 128 + 20]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 987] * kernel_shared_1[threadIdx_x // 7 * 128 + 20]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 994] * kernel_shared_1[threadIdx_x // 7 * 128 + 20]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1001] * kernel_shared_1[threadIdx_x // 7 * 128 + 20]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1008] * kernel_shared_1[threadIdx_x // 7 * 128 + 20]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1015] * kernel_shared_1[threadIdx_x // 7 * 128 + 20]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1022] * kernel_shared_1[threadIdx_x // 7 * 128 + 20]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1029] * kernel_shared_1[threadIdx_x // 7 * 128 + 21]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1036] * kernel_shared_1[threadIdx_x // 7 * 128 + 21]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1043] * kernel_shared_1[threadIdx_x // 7 * 128 + 21]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1050] * kernel_shared_1[threadIdx_x // 7 * 128 + 21]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1057] * kernel_shared_1[threadIdx_x // 7 * 128 + 21]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1064] * kernel_shared_1[threadIdx_x // 7 * 128 + 21]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1071] * kernel_shared_1[threadIdx_x // 7 * 128 + 21]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1078] * kernel_shared_1[threadIdx_x // 7 * 128 + 22]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1085] * kernel_shared_1[threadIdx_x // 7 * 128 + 22]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1092] * kernel_shared_1[threadIdx_x // 7 * 128 + 22]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1099] * kernel_shared_1[threadIdx_x // 7 * 128 + 22]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1106] * kernel_shared_1[threadIdx_x // 7 * 128 + 22]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1113] * kernel_shared_1[threadIdx_x // 7 * 128 + 22]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1120] * kernel_shared_1[threadIdx_x // 7 * 128 + 22]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1127] * kernel_shared_1[threadIdx_x // 7 * 128 + 23]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1134] * kernel_shared_1[threadIdx_x // 7 * 128 + 23]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1141] * kernel_shared_1[threadIdx_x // 7 * 128 + 23]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1148] * kernel_shared_1[threadIdx_x // 7 * 128 + 23]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1155] * kernel_shared_1[threadIdx_x // 7 * 128 + 23]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1162] * kernel_shared_1[threadIdx_x // 7 * 128 + 23]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1169] * kernel_shared_1[threadIdx_x // 7 * 128 + 23]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1176] * kernel_shared_1[threadIdx_x // 7 * 128 + 24]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1183] * kernel_shared_1[threadIdx_x // 7 * 128 + 24]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1190] * kernel_shared_1[threadIdx_x // 7 * 128 + 24]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1197] * kernel_shared_1[threadIdx_x // 7 * 128 + 24]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1204] * kernel_shared_1[threadIdx_x // 7 * 128 + 24]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1211] * kernel_shared_1[threadIdx_x // 7 * 128 + 24]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1218] * kernel_shared_1[threadIdx_x // 7 * 128 + 24]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1225] * kernel_shared_1[threadIdx_x // 7 * 128 + 25]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1232] * kernel_shared_1[threadIdx_x // 7 * 128 + 25]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1239] * kernel_shared_1[threadIdx_x // 7 * 128 + 25]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1246] * kernel_shared_1[threadIdx_x // 7 * 128 + 25]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1253] * kernel_shared_1[threadIdx_x // 7 * 128 + 25]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1260] * kernel_shared_1[threadIdx_x // 7 * 128 + 25]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1267] * kernel_shared_1[threadIdx_x // 7 * 128 + 25]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1274] * kernel_shared_1[threadIdx_x // 7 * 128 + 26]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1281] * kernel_shared_1[threadIdx_x // 7 * 128 + 26]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1288] * kernel_shared_1[threadIdx_x // 7 * 128 + 26]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1295] * kernel_shared_1[threadIdx_x // 7 * 128 + 26]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1302] * kernel_shared_1[threadIdx_x // 7 * 128 + 26]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1309] * kernel_shared_1[threadIdx_x // 7 * 128 + 26]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1316] * kernel_shared_1[threadIdx_x // 7 * 128 + 26]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1323] * kernel_shared_1[threadIdx_x // 7 * 128 + 27]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1330] * kernel_shared_1[threadIdx_x // 7 * 128 + 27]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1337] * kernel_shared_1[threadIdx_x // 7 * 128 + 27]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1344] * kernel_shared_1[threadIdx_x // 7 * 128 + 27]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1351] * kernel_shared_1[threadIdx_x // 7 * 128 + 27]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1358] * kernel_shared_1[threadIdx_x // 7 * 128 + 27]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1365] * kernel_shared_1[threadIdx_x // 7 * 128 + 27]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1372] * kernel_shared_1[threadIdx_x // 7 * 128 + 28]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1379] * kernel_shared_1[threadIdx_x // 7 * 128 + 28]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1386] * kernel_shared_1[threadIdx_x // 7 * 128 + 28]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1393] * kernel_shared_1[threadIdx_x // 7 * 128 + 28]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1400] * kernel_shared_1[threadIdx_x // 7 * 128 + 28]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1407] * kernel_shared_1[threadIdx_x // 7 * 128 + 28]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1414] * kernel_shared_1[threadIdx_x // 7 * 128 + 28]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1421] * kernel_shared_1[threadIdx_x // 7 * 128 + 29]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1428] * kernel_shared_1[threadIdx_x // 7 * 128 + 29]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1435] * kernel_shared_1[threadIdx_x // 7 * 128 + 29]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1442] * kernel_shared_1[threadIdx_x // 7 * 128 + 29]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1449] * kernel_shared_1[threadIdx_x // 7 * 128 + 29]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1456] * kernel_shared_1[threadIdx_x // 7 * 128 + 29]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1463] * kernel_shared_1[threadIdx_x // 7 * 128 + 29]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1470] * kernel_shared_1[threadIdx_x // 7 * 128 + 30]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1477] * kernel_shared_1[threadIdx_x // 7 * 128 + 30]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1484] * kernel_shared_1[threadIdx_x // 7 * 128 + 30]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1491] * kernel_shared_1[threadIdx_x // 7 * 128 + 30]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1498] * kernel_shared_1[threadIdx_x // 7 * 128 + 30]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1505] * kernel_shared_1[threadIdx_x // 7 * 128 + 30]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1512] * kernel_shared_1[threadIdx_x // 7 * 128 + 30]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1519] * kernel_shared_1[threadIdx_x // 7 * 128 + 31]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1526] * kernel_shared_1[threadIdx_x // 7 * 128 + 31]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1533] * kernel_shared_1[threadIdx_x // 7 * 128 + 31]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1540] * kernel_shared_1[threadIdx_x // 7 * 128 + 31]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1547] * kernel_shared_1[threadIdx_x // 7 * 128 + 31]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1554] * kernel_shared_1[threadIdx_x // 7 * 128 + 31]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1561] * kernel_shared_1[threadIdx_x // 7 * 128 + 31]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1568] * kernel_shared_1[threadIdx_x // 7 * 128 + 32]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1575] * kernel_shared_1[threadIdx_x // 7 * 128 + 32]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1582] * kernel_shared_1[threadIdx_x // 7 * 128 + 32]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1589] * kernel_shared_1[threadIdx_x // 7 * 128 + 32]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1596] * kernel_shared_1[threadIdx_x // 7 * 128 + 32]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1603] * kernel_shared_1[threadIdx_x // 7 * 128 + 32]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1610] * kernel_shared_1[threadIdx_x // 7 * 128 + 32]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1617] * kernel_shared_1[threadIdx_x // 7 * 128 + 33]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1624] * kernel_shared_1[threadIdx_x // 7 * 128 + 33]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1631] * kernel_shared_1[threadIdx_x // 7 * 128 + 33]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1638] * kernel_shared_1[threadIdx_x // 7 * 128 + 33]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1645] * kernel_shared_1[threadIdx_x // 7 * 128 + 33]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1652] * kernel_shared_1[threadIdx_x // 7 * 128 + 33]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1659] * kernel_shared_1[threadIdx_x // 7 * 128 + 33]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1666] * kernel_shared_1[threadIdx_x // 7 * 128 + 34]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1673] * kernel_shared_1[threadIdx_x // 7 * 128 + 34]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1680] * kernel_shared_1[threadIdx_x // 7 * 128 + 34]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1687] * kernel_shared_1[threadIdx_x // 7 * 128 + 34]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1694] * kernel_shared_1[threadIdx_x // 7 * 128 + 34]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1701] * kernel_shared_1[threadIdx_x // 7 * 128 + 34]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1708] * kernel_shared_1[threadIdx_x // 7 * 128 + 34]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1715] * kernel_shared_1[threadIdx_x // 7 * 128 + 35]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1722] * kernel_shared_1[threadIdx_x // 7 * 128 + 35]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1729] * kernel_shared_1[threadIdx_x // 7 * 128 + 35]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1736] * kernel_shared_1[threadIdx_x // 7 * 128 + 35]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1743] * kernel_shared_1[threadIdx_x // 7 * 128 + 35]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1750] * kernel_shared_1[threadIdx_x // 7 * 128 + 35]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1757] * kernel_shared_1[threadIdx_x // 7 * 128 + 35]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1764] * kernel_shared_1[threadIdx_x // 7 * 128 + 36]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1771] * kernel_shared_1[threadIdx_x // 7 * 128 + 36]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1778] * kernel_shared_1[threadIdx_x // 7 * 128 + 36]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1785] * kernel_shared_1[threadIdx_x // 7 * 128 + 36]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1792] * kernel_shared_1[threadIdx_x // 7 * 128 + 36]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1799] * kernel_shared_1[threadIdx_x // 7 * 128 + 36]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1806] * kernel_shared_1[threadIdx_x // 7 * 128 + 36]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1813] * kernel_shared_1[threadIdx_x // 7 * 128 + 37]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1820] * kernel_shared_1[threadIdx_x // 7 * 128 + 37]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1827] * kernel_shared_1[threadIdx_x // 7 * 128 + 37]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1834] * kernel_shared_1[threadIdx_x // 7 * 128 + 37]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1841] * kernel_shared_1[threadIdx_x // 7 * 128 + 37]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1848] * kernel_shared_1[threadIdx_x // 7 * 128 + 37]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1855] * kernel_shared_1[threadIdx_x // 7 * 128 + 37]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1862] * kernel_shared_1[threadIdx_x // 7 * 128 + 38]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1869] * kernel_shared_1[threadIdx_x // 7 * 128 + 38]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1876] * kernel_shared_1[threadIdx_x // 7 * 128 + 38]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1883] * kernel_shared_1[threadIdx_x // 7 * 128 + 38]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1890] * kernel_shared_1[threadIdx_x // 7 * 128 + 38]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1897] * kernel_shared_1[threadIdx_x // 7 * 128 + 38]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1904] * kernel_shared_1[threadIdx_x // 7 * 128 + 38]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1911] * kernel_shared_1[threadIdx_x // 7 * 128 + 39]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1918] * kernel_shared_1[threadIdx_x // 7 * 128 + 39]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1925] * kernel_shared_1[threadIdx_x // 7 * 128 + 39]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1932] * kernel_shared_1[threadIdx_x // 7 * 128 + 39]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1939] * kernel_shared_1[threadIdx_x // 7 * 128 + 39]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1946] * kernel_shared_1[threadIdx_x // 7 * 128 + 39]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 1953] * kernel_shared_1[threadIdx_x // 7 * 128 + 39]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 1960] * kernel_shared_1[threadIdx_x // 7 * 128 + 40]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 1967] * kernel_shared_1[threadIdx_x // 7 * 128 + 40]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 1974] * kernel_shared_1[threadIdx_x // 7 * 128 + 40]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 1981] * kernel_shared_1[threadIdx_x // 7 * 128 + 40]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 1988] * kernel_shared_1[threadIdx_x // 7 * 128 + 40]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 1995] * kernel_shared_1[threadIdx_x // 7 * 128 + 40]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2002] * kernel_shared_1[threadIdx_x // 7 * 128 + 40]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2009] * kernel_shared_1[threadIdx_x // 7 * 128 + 41]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2016] * kernel_shared_1[threadIdx_x // 7 * 128 + 41]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2023] * kernel_shared_1[threadIdx_x // 7 * 128 + 41]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2030] * kernel_shared_1[threadIdx_x // 7 * 128 + 41]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2037] * kernel_shared_1[threadIdx_x // 7 * 128 + 41]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2044] * kernel_shared_1[threadIdx_x // 7 * 128 + 41]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2051] * kernel_shared_1[threadIdx_x // 7 * 128 + 41]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2058] * kernel_shared_1[threadIdx_x // 7 * 128 + 42]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2065] * kernel_shared_1[threadIdx_x // 7 * 128 + 42]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2072] * kernel_shared_1[threadIdx_x // 7 * 128 + 42]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2079] * kernel_shared_1[threadIdx_x // 7 * 128 + 42]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2086] * kernel_shared_1[threadIdx_x // 7 * 128 + 42]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2093] * kernel_shared_1[threadIdx_x // 7 * 128 + 42]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2100] * kernel_shared_1[threadIdx_x // 7 * 128 + 42]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2107] * kernel_shared_1[threadIdx_x // 7 * 128 + 43]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2114] * kernel_shared_1[threadIdx_x // 7 * 128 + 43]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2121] * kernel_shared_1[threadIdx_x // 7 * 128 + 43]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2128] * kernel_shared_1[threadIdx_x // 7 * 128 + 43]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2135] * kernel_shared_1[threadIdx_x // 7 * 128 + 43]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2142] * kernel_shared_1[threadIdx_x // 7 * 128 + 43]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2149] * kernel_shared_1[threadIdx_x // 7 * 128 + 43]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2156] * kernel_shared_1[threadIdx_x // 7 * 128 + 44]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2163] * kernel_shared_1[threadIdx_x // 7 * 128 + 44]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2170] * kernel_shared_1[threadIdx_x // 7 * 128 + 44]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2177] * kernel_shared_1[threadIdx_x // 7 * 128 + 44]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2184] * kernel_shared_1[threadIdx_x // 7 * 128 + 44]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2191] * kernel_shared_1[threadIdx_x // 7 * 128 + 44]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2198] * kernel_shared_1[threadIdx_x // 7 * 128 + 44]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2205] * kernel_shared_1[threadIdx_x // 7 * 128 + 45]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2212] * kernel_shared_1[threadIdx_x // 7 * 128 + 45]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2219] * kernel_shared_1[threadIdx_x // 7 * 128 + 45]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2226] * kernel_shared_1[threadIdx_x // 7 * 128 + 45]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2233] * kernel_shared_1[threadIdx_x // 7 * 128 + 45]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2240] * kernel_shared_1[threadIdx_x // 7 * 128 + 45]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2247] * kernel_shared_1[threadIdx_x // 7 * 128 + 45]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2254] * kernel_shared_1[threadIdx_x // 7 * 128 + 46]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2261] * kernel_shared_1[threadIdx_x // 7 * 128 + 46]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2268] * kernel_shared_1[threadIdx_x // 7 * 128 + 46]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2275] * kernel_shared_1[threadIdx_x // 7 * 128 + 46]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2282] * kernel_shared_1[threadIdx_x // 7 * 128 + 46]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2289] * kernel_shared_1[threadIdx_x // 7 * 128 + 46]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2296] * kernel_shared_1[threadIdx_x // 7 * 128 + 46]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2303] * kernel_shared_1[threadIdx_x // 7 * 128 + 47]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2310] * kernel_shared_1[threadIdx_x // 7 * 128 + 47]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2317] * kernel_shared_1[threadIdx_x // 7 * 128 + 47]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2324] * kernel_shared_1[threadIdx_x // 7 * 128 + 47]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2331] * kernel_shared_1[threadIdx_x // 7 * 128 + 47]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2338] * kernel_shared_1[threadIdx_x // 7 * 128 + 47]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2345] * kernel_shared_1[threadIdx_x // 7 * 128 + 47]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2352] * kernel_shared_1[threadIdx_x // 7 * 128 + 48]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2359] * kernel_shared_1[threadIdx_x // 7 * 128 + 48]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2366] * kernel_shared_1[threadIdx_x // 7 * 128 + 48]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2373] * kernel_shared_1[threadIdx_x // 7 * 128 + 48]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2380] * kernel_shared_1[threadIdx_x // 7 * 128 + 48]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2387] * kernel_shared_1[threadIdx_x // 7 * 128 + 48]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2394] * kernel_shared_1[threadIdx_x // 7 * 128 + 48]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2401] * kernel_shared_1[threadIdx_x // 7 * 128 + 49]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2408] * kernel_shared_1[threadIdx_x // 7 * 128 + 49]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2415] * kernel_shared_1[threadIdx_x // 7 * 128 + 49]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2422] * kernel_shared_1[threadIdx_x // 7 * 128 + 49]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2429] * kernel_shared_1[threadIdx_x // 7 * 128 + 49]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2436] * kernel_shared_1[threadIdx_x // 7 * 128 + 49]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2443] * kernel_shared_1[threadIdx_x // 7 * 128 + 49]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2450] * kernel_shared_1[threadIdx_x // 7 * 128 + 50]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2457] * kernel_shared_1[threadIdx_x // 7 * 128 + 50]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2464] * kernel_shared_1[threadIdx_x // 7 * 128 + 50]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2471] * kernel_shared_1[threadIdx_x // 7 * 128 + 50]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2478] * kernel_shared_1[threadIdx_x // 7 * 128 + 50]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2485] * kernel_shared_1[threadIdx_x // 7 * 128 + 50]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2492] * kernel_shared_1[threadIdx_x // 7 * 128 + 50]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2499] * kernel_shared_1[threadIdx_x // 7 * 128 + 51]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2506] * kernel_shared_1[threadIdx_x // 7 * 128 + 51]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2513] * kernel_shared_1[threadIdx_x // 7 * 128 + 51]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2520] * kernel_shared_1[threadIdx_x // 7 * 128 + 51]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2527] * kernel_shared_1[threadIdx_x // 7 * 128 + 51]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2534] * kernel_shared_1[threadIdx_x // 7 * 128 + 51]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2541] * kernel_shared_1[threadIdx_x // 7 * 128 + 51]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2548] * kernel_shared_1[threadIdx_x // 7 * 128 + 52]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2555] * kernel_shared_1[threadIdx_x // 7 * 128 + 52]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2562] * kernel_shared_1[threadIdx_x // 7 * 128 + 52]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2569] * kernel_shared_1[threadIdx_x // 7 * 128 + 52]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2576] * kernel_shared_1[threadIdx_x // 7 * 128 + 52]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2583] * kernel_shared_1[threadIdx_x // 7 * 128 + 52]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2590] * kernel_shared_1[threadIdx_x // 7 * 128 + 52]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2597] * kernel_shared_1[threadIdx_x // 7 * 128 + 53]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2604] * kernel_shared_1[threadIdx_x // 7 * 128 + 53]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2611] * kernel_shared_1[threadIdx_x // 7 * 128 + 53]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2618] * kernel_shared_1[threadIdx_x // 7 * 128 + 53]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2625] * kernel_shared_1[threadIdx_x // 7 * 128 + 53]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2632] * kernel_shared_1[threadIdx_x // 7 * 128 + 53]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2639] * kernel_shared_1[threadIdx_x // 7 * 128 + 53]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2646] * kernel_shared_1[threadIdx_x // 7 * 128 + 54]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2653] * kernel_shared_1[threadIdx_x // 7 * 128 + 54]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2660] * kernel_shared_1[threadIdx_x // 7 * 128 + 54]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2667] * kernel_shared_1[threadIdx_x // 7 * 128 + 54]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2674] * kernel_shared_1[threadIdx_x // 7 * 128 + 54]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2681] * kernel_shared_1[threadIdx_x // 7 * 128 + 54]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2688] * kernel_shared_1[threadIdx_x // 7 * 128 + 54]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2695] * kernel_shared_1[threadIdx_x // 7 * 128 + 55]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2702] * kernel_shared_1[threadIdx_x // 7 * 128 + 55]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2709] * kernel_shared_1[threadIdx_x // 7 * 128 + 55]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2716] * kernel_shared_1[threadIdx_x // 7 * 128 + 55]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2723] * kernel_shared_1[threadIdx_x // 7 * 128 + 55]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2730] * kernel_shared_1[threadIdx_x // 7 * 128 + 55]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2737] * kernel_shared_1[threadIdx_x // 7 * 128 + 55]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2744] * kernel_shared_1[threadIdx_x // 7 * 128 + 56]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2751] * kernel_shared_1[threadIdx_x // 7 * 128 + 56]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2758] * kernel_shared_1[threadIdx_x // 7 * 128 + 56]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2765] * kernel_shared_1[threadIdx_x // 7 * 128 + 56]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2772] * kernel_shared_1[threadIdx_x // 7 * 128 + 56]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2779] * kernel_shared_1[threadIdx_x // 7 * 128 + 56]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2786] * kernel_shared_1[threadIdx_x // 7 * 128 + 56]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2793] * kernel_shared_1[threadIdx_x // 7 * 128 + 57]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2800] * kernel_shared_1[threadIdx_x // 7 * 128 + 57]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2807] * kernel_shared_1[threadIdx_x // 7 * 128 + 57]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2814] * kernel_shared_1[threadIdx_x // 7 * 128 + 57]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2821] * kernel_shared_1[threadIdx_x // 7 * 128 + 57]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2828] * kernel_shared_1[threadIdx_x // 7 * 128 + 57]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2835] * kernel_shared_1[threadIdx_x // 7 * 128 + 57]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2842] * kernel_shared_1[threadIdx_x // 7 * 128 + 58]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2849] * kernel_shared_1[threadIdx_x // 7 * 128 + 58]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2856] * kernel_shared_1[threadIdx_x // 7 * 128 + 58]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2863] * kernel_shared_1[threadIdx_x // 7 * 128 + 58]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2870] * kernel_shared_1[threadIdx_x // 7 * 128 + 58]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2877] * kernel_shared_1[threadIdx_x // 7 * 128 + 58]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2884] * kernel_shared_1[threadIdx_x // 7 * 128 + 58]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2891] * kernel_shared_1[threadIdx_x // 7 * 128 + 59]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2898] * kernel_shared_1[threadIdx_x // 7 * 128 + 59]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2905] * kernel_shared_1[threadIdx_x // 7 * 128 + 59]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2912] * kernel_shared_1[threadIdx_x // 7 * 128 + 59]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2919] * kernel_shared_1[threadIdx_x // 7 * 128 + 59]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2926] * kernel_shared_1[threadIdx_x // 7 * 128 + 59]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2933] * kernel_shared_1[threadIdx_x // 7 * 128 + 59]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2940] * kernel_shared_1[threadIdx_x // 7 * 128 + 60]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2947] * kernel_shared_1[threadIdx_x // 7 * 128 + 60]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 2954] * kernel_shared_1[threadIdx_x // 7 * 128 + 60]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 2961] * kernel_shared_1[threadIdx_x // 7 * 128 + 60]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 2968] * kernel_shared_1[threadIdx_x // 7 * 128 + 60]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 2975] * kernel_shared_1[threadIdx_x // 7 * 128 + 60]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 2982] * kernel_shared_1[threadIdx_x // 7 * 128 + 60]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 2989] * kernel_shared_1[threadIdx_x // 7 * 128 + 61]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 2996] * kernel_shared_1[threadIdx_x // 7 * 128 + 61]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3003] * kernel_shared_1[threadIdx_x // 7 * 128 + 61]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3010] * kernel_shared_1[threadIdx_x // 7 * 128 + 61]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3017] * kernel_shared_1[threadIdx_x // 7 * 128 + 61]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3024] * kernel_shared_1[threadIdx_x // 7 * 128 + 61]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3031] * kernel_shared_1[threadIdx_x // 7 * 128 + 61]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3038] * kernel_shared_1[threadIdx_x // 7 * 128 + 62]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3045] * kernel_shared_1[threadIdx_x // 7 * 128 + 62]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3052] * kernel_shared_1[threadIdx_x // 7 * 128 + 62]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3059] * kernel_shared_1[threadIdx_x // 7 * 128 + 62]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3066] * kernel_shared_1[threadIdx_x // 7 * 128 + 62]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3073] * kernel_shared_1[threadIdx_x // 7 * 128 + 62]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3080] * kernel_shared_1[threadIdx_x // 7 * 128 + 62]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3087] * kernel_shared_1[threadIdx_x // 7 * 128 + 63]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3094] * kernel_shared_1[threadIdx_x // 7 * 128 + 63]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3101] * kernel_shared_1[threadIdx_x // 7 * 128 + 63]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3108] * kernel_shared_1[threadIdx_x // 7 * 128 + 63]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3115] * kernel_shared_1[threadIdx_x // 7 * 128 + 63]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3122] * kernel_shared_1[threadIdx_x // 7 * 128 + 63]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3129] * kernel_shared_1[threadIdx_x // 7 * 128 + 63]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3136] * kernel_shared_1[threadIdx_x // 7 * 128 + 64]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3143] * kernel_shared_1[threadIdx_x // 7 * 128 + 64]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3150] * kernel_shared_1[threadIdx_x // 7 * 128 + 64]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3157] * kernel_shared_1[threadIdx_x // 7 * 128 + 64]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3164] * kernel_shared_1[threadIdx_x // 7 * 128 + 64]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3171] * kernel_shared_1[threadIdx_x // 7 * 128 + 64]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3178] * kernel_shared_1[threadIdx_x // 7 * 128 + 64]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3185] * kernel_shared_1[threadIdx_x // 7 * 128 + 65]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3192] * kernel_shared_1[threadIdx_x // 7 * 128 + 65]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3199] * kernel_shared_1[threadIdx_x // 7 * 128 + 65]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3206] * kernel_shared_1[threadIdx_x // 7 * 128 + 65]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3213] * kernel_shared_1[threadIdx_x // 7 * 128 + 65]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3220] * kernel_shared_1[threadIdx_x // 7 * 128 + 65]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3227] * kernel_shared_1[threadIdx_x // 7 * 128 + 65]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3234] * kernel_shared_1[threadIdx_x // 7 * 128 + 66]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3241] * kernel_shared_1[threadIdx_x // 7 * 128 + 66]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3248] * kernel_shared_1[threadIdx_x // 7 * 128 + 66]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3255] * kernel_shared_1[threadIdx_x // 7 * 128 + 66]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3262] * kernel_shared_1[threadIdx_x // 7 * 128 + 66]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3269] * kernel_shared_1[threadIdx_x // 7 * 128 + 66]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3276] * kernel_shared_1[threadIdx_x // 7 * 128 + 66]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3283] * kernel_shared_1[threadIdx_x // 7 * 128 + 67]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3290] * kernel_shared_1[threadIdx_x // 7 * 128 + 67]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3297] * kernel_shared_1[threadIdx_x // 7 * 128 + 67]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3304] * kernel_shared_1[threadIdx_x // 7 * 128 + 67]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3311] * kernel_shared_1[threadIdx_x // 7 * 128 + 67]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3318] * kernel_shared_1[threadIdx_x // 7 * 128 + 67]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3325] * kernel_shared_1[threadIdx_x // 7 * 128 + 67]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3332] * kernel_shared_1[threadIdx_x // 7 * 128 + 68]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3339] * kernel_shared_1[threadIdx_x // 7 * 128 + 68]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3346] * kernel_shared_1[threadIdx_x // 7 * 128 + 68]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3353] * kernel_shared_1[threadIdx_x // 7 * 128 + 68]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3360] * kernel_shared_1[threadIdx_x // 7 * 128 + 68]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3367] * kernel_shared_1[threadIdx_x // 7 * 128 + 68]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3374] * kernel_shared_1[threadIdx_x // 7 * 128 + 68]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3381] * kernel_shared_1[threadIdx_x // 7 * 128 + 69]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3388] * kernel_shared_1[threadIdx_x // 7 * 128 + 69]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3395] * kernel_shared_1[threadIdx_x // 7 * 128 + 69]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3402] * kernel_shared_1[threadIdx_x // 7 * 128 + 69]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3409] * kernel_shared_1[threadIdx_x // 7 * 128 + 69]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3416] * kernel_shared_1[threadIdx_x // 7 * 128 + 69]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3423] * kernel_shared_1[threadIdx_x // 7 * 128 + 69]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3430] * kernel_shared_1[threadIdx_x // 7 * 128 + 70]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3437] * kernel_shared_1[threadIdx_x // 7 * 128 + 70]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3444] * kernel_shared_1[threadIdx_x // 7 * 128 + 70]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3451] * kernel_shared_1[threadIdx_x // 7 * 128 + 70]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3458] * kernel_shared_1[threadIdx_x // 7 * 128 + 70]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3465] * kernel_shared_1[threadIdx_x // 7 * 128 + 70]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3472] * kernel_shared_1[threadIdx_x // 7 * 128 + 70]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3479] * kernel_shared_1[threadIdx_x // 7 * 128 + 71]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3486] * kernel_shared_1[threadIdx_x // 7 * 128 + 71]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3493] * kernel_shared_1[threadIdx_x // 7 * 128 + 71]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3500] * kernel_shared_1[threadIdx_x // 7 * 128 + 71]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3507] * kernel_shared_1[threadIdx_x // 7 * 128 + 71]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3514] * kernel_shared_1[threadIdx_x // 7 * 128 + 71]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3521] * kernel_shared_1[threadIdx_x // 7 * 128 + 71]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3528] * kernel_shared_1[threadIdx_x // 7 * 128 + 72]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3535] * kernel_shared_1[threadIdx_x // 7 * 128 + 72]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3542] * kernel_shared_1[threadIdx_x // 7 * 128 + 72]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3549] * kernel_shared_1[threadIdx_x // 7 * 128 + 72]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3556] * kernel_shared_1[threadIdx_x // 7 * 128 + 72]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3563] * kernel_shared_1[threadIdx_x // 7 * 128 + 72]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3570] * kernel_shared_1[threadIdx_x // 7 * 128 + 72]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3577] * kernel_shared_1[threadIdx_x // 7 * 128 + 73]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3584] * kernel_shared_1[threadIdx_x // 7 * 128 + 73]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3591] * kernel_shared_1[threadIdx_x // 7 * 128 + 73]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3598] * kernel_shared_1[threadIdx_x // 7 * 128 + 73]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3605] * kernel_shared_1[threadIdx_x // 7 * 128 + 73]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3612] * kernel_shared_1[threadIdx_x // 7 * 128 + 73]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3619] * kernel_shared_1[threadIdx_x // 7 * 128 + 73]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3626] * kernel_shared_1[threadIdx_x // 7 * 128 + 74]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3633] * kernel_shared_1[threadIdx_x // 7 * 128 + 74]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3640] * kernel_shared_1[threadIdx_x // 7 * 128 + 74]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3647] * kernel_shared_1[threadIdx_x // 7 * 128 + 74]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3654] * kernel_shared_1[threadIdx_x // 7 * 128 + 74]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3661] * kernel_shared_1[threadIdx_x // 7 * 128 + 74]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3668] * kernel_shared_1[threadIdx_x // 7 * 128 + 74]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3675] * kernel_shared_1[threadIdx_x // 7 * 128 + 75]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3682] * kernel_shared_1[threadIdx_x // 7 * 128 + 75]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3689] * kernel_shared_1[threadIdx_x // 7 * 128 + 75]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3696] * kernel_shared_1[threadIdx_x // 7 * 128 + 75]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3703] * kernel_shared_1[threadIdx_x // 7 * 128 + 75]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3710] * kernel_shared_1[threadIdx_x // 7 * 128 + 75]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3717] * kernel_shared_1[threadIdx_x // 7 * 128 + 75]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3724] * kernel_shared_1[threadIdx_x // 7 * 128 + 76]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3731] * kernel_shared_1[threadIdx_x // 7 * 128 + 76]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3738] * kernel_shared_1[threadIdx_x // 7 * 128 + 76]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3745] * kernel_shared_1[threadIdx_x // 7 * 128 + 76]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3752] * kernel_shared_1[threadIdx_x // 7 * 128 + 76]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3759] * kernel_shared_1[threadIdx_x // 7 * 128 + 76]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3766] * kernel_shared_1[threadIdx_x // 7 * 128 + 76]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3773] * kernel_shared_1[threadIdx_x // 7 * 128 + 77]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3780] * kernel_shared_1[threadIdx_x // 7 * 128 + 77]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3787] * kernel_shared_1[threadIdx_x // 7 * 128 + 77]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3794] * kernel_shared_1[threadIdx_x // 7 * 128 + 77]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3801] * kernel_shared_1[threadIdx_x // 7 * 128 + 77]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3808] * kernel_shared_1[threadIdx_x // 7 * 128 + 77]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3815] * kernel_shared_1[threadIdx_x // 7 * 128 + 77]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3822] * kernel_shared_1[threadIdx_x // 7 * 128 + 78]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3829] * kernel_shared_1[threadIdx_x // 7 * 128 + 78]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3836] * kernel_shared_1[threadIdx_x // 7 * 128 + 78]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3843] * kernel_shared_1[threadIdx_x // 7 * 128 + 78]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3850] * kernel_shared_1[threadIdx_x // 7 * 128 + 78]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3857] * kernel_shared_1[threadIdx_x // 7 * 128 + 78]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3864] * kernel_shared_1[threadIdx_x // 7 * 128 + 78]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3871] * kernel_shared_1[threadIdx_x // 7 * 128 + 79]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3878] * kernel_shared_1[threadIdx_x // 7 * 128 + 79]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3885] * kernel_shared_1[threadIdx_x // 7 * 128 + 79]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3892] * kernel_shared_1[threadIdx_x // 7 * 128 + 79]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3899] * kernel_shared_1[threadIdx_x // 7 * 128 + 79]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3906] * kernel_shared_1[threadIdx_x // 7 * 128 + 79]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3913] * kernel_shared_1[threadIdx_x // 7 * 128 + 79]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3920] * kernel_shared_1[threadIdx_x // 7 * 128 + 80]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3927] * kernel_shared_1[threadIdx_x // 7 * 128 + 80]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3934] * kernel_shared_1[threadIdx_x // 7 * 128 + 80]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3941] * kernel_shared_1[threadIdx_x // 7 * 128 + 80]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3948] * kernel_shared_1[threadIdx_x // 7 * 128 + 80]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 3955] * kernel_shared_1[threadIdx_x // 7 * 128 + 80]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 3962] * kernel_shared_1[threadIdx_x // 7 * 128 + 80]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 3969] * kernel_shared_1[threadIdx_x // 7 * 128 + 81]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 3976] * kernel_shared_1[threadIdx_x // 7 * 128 + 81]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 3983] * kernel_shared_1[threadIdx_x // 7 * 128 + 81]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 3990] * kernel_shared_1[threadIdx_x // 7 * 128 + 81]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 3997] * kernel_shared_1[threadIdx_x // 7 * 128 + 81]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4004] * kernel_shared_1[threadIdx_x // 7 * 128 + 81]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4011] * kernel_shared_1[threadIdx_x // 7 * 128 + 81]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4018] * kernel_shared_1[threadIdx_x // 7 * 128 + 82]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4025] * kernel_shared_1[threadIdx_x // 7 * 128 + 82]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4032] * kernel_shared_1[threadIdx_x // 7 * 128 + 82]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4039] * kernel_shared_1[threadIdx_x // 7 * 128 + 82]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4046] * kernel_shared_1[threadIdx_x // 7 * 128 + 82]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4053] * kernel_shared_1[threadIdx_x // 7 * 128 + 82]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4060] * kernel_shared_1[threadIdx_x // 7 * 128 + 82]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4067] * kernel_shared_1[threadIdx_x // 7 * 128 + 83]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4074] * kernel_shared_1[threadIdx_x // 7 * 128 + 83]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4081] * kernel_shared_1[threadIdx_x // 7 * 128 + 83]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4088] * kernel_shared_1[threadIdx_x // 7 * 128 + 83]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4095] * kernel_shared_1[threadIdx_x // 7 * 128 + 83]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4102] * kernel_shared_1[threadIdx_x // 7 * 128 + 83]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4109] * kernel_shared_1[threadIdx_x // 7 * 128 + 83]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4116] * kernel_shared_1[threadIdx_x // 7 * 128 + 84]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4123] * kernel_shared_1[threadIdx_x // 7 * 128 + 84]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4130] * kernel_shared_1[threadIdx_x // 7 * 128 + 84]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4137] * kernel_shared_1[threadIdx_x // 7 * 128 + 84]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4144] * kernel_shared_1[threadIdx_x // 7 * 128 + 84]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4151] * kernel_shared_1[threadIdx_x // 7 * 128 + 84]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4158] * kernel_shared_1[threadIdx_x // 7 * 128 + 84]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4165] * kernel_shared_1[threadIdx_x // 7 * 128 + 85]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4172] * kernel_shared_1[threadIdx_x // 7 * 128 + 85]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4179] * kernel_shared_1[threadIdx_x // 7 * 128 + 85]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4186] * kernel_shared_1[threadIdx_x // 7 * 128 + 85]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4193] * kernel_shared_1[threadIdx_x // 7 * 128 + 85]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4200] * kernel_shared_1[threadIdx_x // 7 * 128 + 85]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4207] * kernel_shared_1[threadIdx_x // 7 * 128 + 85]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4214] * kernel_shared_1[threadIdx_x // 7 * 128 + 86]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4221] * kernel_shared_1[threadIdx_x // 7 * 128 + 86]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4228] * kernel_shared_1[threadIdx_x // 7 * 128 + 86]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4235] * kernel_shared_1[threadIdx_x // 7 * 128 + 86]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4242] * kernel_shared_1[threadIdx_x // 7 * 128 + 86]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4249] * kernel_shared_1[threadIdx_x // 7 * 128 + 86]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4256] * kernel_shared_1[threadIdx_x // 7 * 128 + 86]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4263] * kernel_shared_1[threadIdx_x // 7 * 128 + 87]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4270] * kernel_shared_1[threadIdx_x // 7 * 128 + 87]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4277] * kernel_shared_1[threadIdx_x // 7 * 128 + 87]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4284] * kernel_shared_1[threadIdx_x // 7 * 128 + 87]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4291] * kernel_shared_1[threadIdx_x // 7 * 128 + 87]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4298] * kernel_shared_1[threadIdx_x // 7 * 128 + 87]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4305] * kernel_shared_1[threadIdx_x // 7 * 128 + 87]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4312] * kernel_shared_1[threadIdx_x // 7 * 128 + 88]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4319] * kernel_shared_1[threadIdx_x // 7 * 128 + 88]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4326] * kernel_shared_1[threadIdx_x // 7 * 128 + 88]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4333] * kernel_shared_1[threadIdx_x // 7 * 128 + 88]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4340] * kernel_shared_1[threadIdx_x // 7 * 128 + 88]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4347] * kernel_shared_1[threadIdx_x // 7 * 128 + 88]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4354] * kernel_shared_1[threadIdx_x // 7 * 128 + 88]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4361] * kernel_shared_1[threadIdx_x // 7 * 128 + 89]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4368] * kernel_shared_1[threadIdx_x // 7 * 128 + 89]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4375] * kernel_shared_1[threadIdx_x // 7 * 128 + 89]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4382] * kernel_shared_1[threadIdx_x // 7 * 128 + 89]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4389] * kernel_shared_1[threadIdx_x // 7 * 128 + 89]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4396] * kernel_shared_1[threadIdx_x // 7 * 128 + 89]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4403] * kernel_shared_1[threadIdx_x // 7 * 128 + 89]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4410] * kernel_shared_1[threadIdx_x // 7 * 128 + 90]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4417] * kernel_shared_1[threadIdx_x // 7 * 128 + 90]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4424] * kernel_shared_1[threadIdx_x // 7 * 128 + 90]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4431] * kernel_shared_1[threadIdx_x // 7 * 128 + 90]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4438] * kernel_shared_1[threadIdx_x // 7 * 128 + 90]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4445] * kernel_shared_1[threadIdx_x // 7 * 128 + 90]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4452] * kernel_shared_1[threadIdx_x // 7 * 128 + 90]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4459] * kernel_shared_1[threadIdx_x // 7 * 128 + 91]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4466] * kernel_shared_1[threadIdx_x // 7 * 128 + 91]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4473] * kernel_shared_1[threadIdx_x // 7 * 128 + 91]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4480] * kernel_shared_1[threadIdx_x // 7 * 128 + 91]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4487] * kernel_shared_1[threadIdx_x // 7 * 128 + 91]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4494] * kernel_shared_1[threadIdx_x // 7 * 128 + 91]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4501] * kernel_shared_1[threadIdx_x // 7 * 128 + 91]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4508] * kernel_shared_1[threadIdx_x // 7 * 128 + 92]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4515] * kernel_shared_1[threadIdx_x // 7 * 128 + 92]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4522] * kernel_shared_1[threadIdx_x // 7 * 128 + 92]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4529] * kernel_shared_1[threadIdx_x // 7 * 128 + 92]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4536] * kernel_shared_1[threadIdx_x // 7 * 128 + 92]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4543] * kernel_shared_1[threadIdx_x // 7 * 128 + 92]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4550] * kernel_shared_1[threadIdx_x // 7 * 128 + 92]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4557] * kernel_shared_1[threadIdx_x // 7 * 128 + 93]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4564] * kernel_shared_1[threadIdx_x // 7 * 128 + 93]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4571] * kernel_shared_1[threadIdx_x // 7 * 128 + 93]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4578] * kernel_shared_1[threadIdx_x // 7 * 128 + 93]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4585] * kernel_shared_1[threadIdx_x // 7 * 128 + 93]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4592] * kernel_shared_1[threadIdx_x // 7 * 128 + 93]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4599] * kernel_shared_1[threadIdx_x // 7 * 128 + 93]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4606] * kernel_shared_1[threadIdx_x // 7 * 128 + 94]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4613] * kernel_shared_1[threadIdx_x // 7 * 128 + 94]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4620] * kernel_shared_1[threadIdx_x // 7 * 128 + 94]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4627] * kernel_shared_1[threadIdx_x // 7 * 128 + 94]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4634] * kernel_shared_1[threadIdx_x // 7 * 128 + 94]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4641] * kernel_shared_1[threadIdx_x // 7 * 128 + 94]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4648] * kernel_shared_1[threadIdx_x // 7 * 128 + 94]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4655] * kernel_shared_1[threadIdx_x // 7 * 128 + 95]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4662] * kernel_shared_1[threadIdx_x // 7 * 128 + 95]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4669] * kernel_shared_1[threadIdx_x // 7 * 128 + 95]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4676] * kernel_shared_1[threadIdx_x // 7 * 128 + 95]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4683] * kernel_shared_1[threadIdx_x // 7 * 128 + 95]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4690] * kernel_shared_1[threadIdx_x // 7 * 128 + 95]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4697] * kernel_shared_1[threadIdx_x // 7 * 128 + 95]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4704] * kernel_shared_1[threadIdx_x // 7 * 128 + 96]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4711] * kernel_shared_1[threadIdx_x // 7 * 128 + 96]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4718] * kernel_shared_1[threadIdx_x // 7 * 128 + 96]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4725] * kernel_shared_1[threadIdx_x // 7 * 128 + 96]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4732] * kernel_shared_1[threadIdx_x // 7 * 128 + 96]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4739] * kernel_shared_1[threadIdx_x // 7 * 128 + 96]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4746] * kernel_shared_1[threadIdx_x // 7 * 128 + 96]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4753] * kernel_shared_1[threadIdx_x // 7 * 128 + 97]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4760] * kernel_shared_1[threadIdx_x // 7 * 128 + 97]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4767] * kernel_shared_1[threadIdx_x // 7 * 128 + 97]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4774] * kernel_shared_1[threadIdx_x // 7 * 128 + 97]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4781] * kernel_shared_1[threadIdx_x // 7 * 128 + 97]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4788] * kernel_shared_1[threadIdx_x // 7 * 128 + 97]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4795] * kernel_shared_1[threadIdx_x // 7 * 128 + 97]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4802] * kernel_shared_1[threadIdx_x // 7 * 128 + 98]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4809] * kernel_shared_1[threadIdx_x // 7 * 128 + 98]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4816] * kernel_shared_1[threadIdx_x // 7 * 128 + 98]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4823] * kernel_shared_1[threadIdx_x // 7 * 128 + 98]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4830] * kernel_shared_1[threadIdx_x // 7 * 128 + 98]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4837] * kernel_shared_1[threadIdx_x // 7 * 128 + 98]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4844] * kernel_shared_1[threadIdx_x // 7 * 128 + 98]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4851] * kernel_shared_1[threadIdx_x // 7 * 128 + 99]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4858] * kernel_shared_1[threadIdx_x // 7 * 128 + 99]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4865] * kernel_shared_1[threadIdx_x // 7 * 128 + 99]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4872] * kernel_shared_1[threadIdx_x // 7 * 128 + 99]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4879] * kernel_shared_1[threadIdx_x // 7 * 128 + 99]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4886] * kernel_shared_1[threadIdx_x // 7 * 128 + 99]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4893] * kernel_shared_1[threadIdx_x // 7 * 128 + 99]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4900] * kernel_shared_1[threadIdx_x // 7 * 128 + 100]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4907] * kernel_shared_1[threadIdx_x // 7 * 128 + 100]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4914] * kernel_shared_1[threadIdx_x // 7 * 128 + 100]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4921] * kernel_shared_1[threadIdx_x // 7 * 128 + 100]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4928] * kernel_shared_1[threadIdx_x // 7 * 128 + 100]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4935] * kernel_shared_1[threadIdx_x // 7 * 128 + 100]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4942] * kernel_shared_1[threadIdx_x // 7 * 128 + 100]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4949] * kernel_shared_1[threadIdx_x // 7 * 128 + 101]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 4956] * kernel_shared_1[threadIdx_x // 7 * 128 + 101]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 4963] * kernel_shared_1[threadIdx_x // 7 * 128 + 101]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 4970] * kernel_shared_1[threadIdx_x // 7 * 128 + 101]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 4977] * kernel_shared_1[threadIdx_x // 7 * 128 + 101]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 4984] * kernel_shared_1[threadIdx_x // 7 * 128 + 101]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 4991] * kernel_shared_1[threadIdx_x // 7 * 128 + 101]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 4998] * kernel_shared_1[threadIdx_x // 7 * 128 + 102]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5005] * kernel_shared_1[threadIdx_x // 7 * 128 + 102]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5012] * kernel_shared_1[threadIdx_x // 7 * 128 + 102]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5019] * kernel_shared_1[threadIdx_x // 7 * 128 + 102]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5026] * kernel_shared_1[threadIdx_x // 7 * 128 + 102]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5033] * kernel_shared_1[threadIdx_x // 7 * 128 + 102]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5040] * kernel_shared_1[threadIdx_x // 7 * 128 + 102]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5047] * kernel_shared_1[threadIdx_x // 7 * 128 + 103]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5054] * kernel_shared_1[threadIdx_x // 7 * 128 + 103]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5061] * kernel_shared_1[threadIdx_x // 7 * 128 + 103]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5068] * kernel_shared_1[threadIdx_x // 7 * 128 + 103]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5075] * kernel_shared_1[threadIdx_x // 7 * 128 + 103]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5082] * kernel_shared_1[threadIdx_x // 7 * 128 + 103]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5089] * kernel_shared_1[threadIdx_x // 7 * 128 + 103]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5096] * kernel_shared_1[threadIdx_x // 7 * 128 + 104]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5103] * kernel_shared_1[threadIdx_x // 7 * 128 + 104]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5110] * kernel_shared_1[threadIdx_x // 7 * 128 + 104]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5117] * kernel_shared_1[threadIdx_x // 7 * 128 + 104]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5124] * kernel_shared_1[threadIdx_x // 7 * 128 + 104]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5131] * kernel_shared_1[threadIdx_x // 7 * 128 + 104]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5138] * kernel_shared_1[threadIdx_x // 7 * 128 + 104]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5145] * kernel_shared_1[threadIdx_x // 7 * 128 + 105]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5152] * kernel_shared_1[threadIdx_x // 7 * 128 + 105]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5159] * kernel_shared_1[threadIdx_x // 7 * 128 + 105]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5166] * kernel_shared_1[threadIdx_x // 7 * 128 + 105]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5173] * kernel_shared_1[threadIdx_x // 7 * 128 + 105]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5180] * kernel_shared_1[threadIdx_x // 7 * 128 + 105]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5187] * kernel_shared_1[threadIdx_x // 7 * 128 + 105]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5194] * kernel_shared_1[threadIdx_x // 7 * 128 + 106]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5201] * kernel_shared_1[threadIdx_x // 7 * 128 + 106]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5208] * kernel_shared_1[threadIdx_x // 7 * 128 + 106]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5215] * kernel_shared_1[threadIdx_x // 7 * 128 + 106]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5222] * kernel_shared_1[threadIdx_x // 7 * 128 + 106]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5229] * kernel_shared_1[threadIdx_x // 7 * 128 + 106]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5236] * kernel_shared_1[threadIdx_x // 7 * 128 + 106]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5243] * kernel_shared_1[threadIdx_x // 7 * 128 + 107]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5250] * kernel_shared_1[threadIdx_x // 7 * 128 + 107]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5257] * kernel_shared_1[threadIdx_x // 7 * 128 + 107]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5264] * kernel_shared_1[threadIdx_x // 7 * 128 + 107]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5271] * kernel_shared_1[threadIdx_x // 7 * 128 + 107]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5278] * kernel_shared_1[threadIdx_x // 7 * 128 + 107]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5285] * kernel_shared_1[threadIdx_x // 7 * 128 + 107]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5292] * kernel_shared_1[threadIdx_x // 7 * 128 + 108]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5299] * kernel_shared_1[threadIdx_x // 7 * 128 + 108]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5306] * kernel_shared_1[threadIdx_x // 7 * 128 + 108]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5313] * kernel_shared_1[threadIdx_x // 7 * 128 + 108]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5320] * kernel_shared_1[threadIdx_x // 7 * 128 + 108]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5327] * kernel_shared_1[threadIdx_x // 7 * 128 + 108]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5334] * kernel_shared_1[threadIdx_x // 7 * 128 + 108]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5341] * kernel_shared_1[threadIdx_x // 7 * 128 + 109]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5348] * kernel_shared_1[threadIdx_x // 7 * 128 + 109]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5355] * kernel_shared_1[threadIdx_x // 7 * 128 + 109]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5362] * kernel_shared_1[threadIdx_x // 7 * 128 + 109]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5369] * kernel_shared_1[threadIdx_x // 7 * 128 + 109]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5376] * kernel_shared_1[threadIdx_x // 7 * 128 + 109]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5383] * kernel_shared_1[threadIdx_x // 7 * 128 + 109]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5390] * kernel_shared_1[threadIdx_x // 7 * 128 + 110]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5397] * kernel_shared_1[threadIdx_x // 7 * 128 + 110]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5404] * kernel_shared_1[threadIdx_x // 7 * 128 + 110]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5411] * kernel_shared_1[threadIdx_x // 7 * 128 + 110]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5418] * kernel_shared_1[threadIdx_x // 7 * 128 + 110]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5425] * kernel_shared_1[threadIdx_x // 7 * 128 + 110]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5432] * kernel_shared_1[threadIdx_x // 7 * 128 + 110]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5439] * kernel_shared_1[threadIdx_x // 7 * 128 + 111]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5446] * kernel_shared_1[threadIdx_x // 7 * 128 + 111]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5453] * kernel_shared_1[threadIdx_x // 7 * 128 + 111]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5460] * kernel_shared_1[threadIdx_x // 7 * 128 + 111]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5467] * kernel_shared_1[threadIdx_x // 7 * 128 + 111]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5474] * kernel_shared_1[threadIdx_x // 7 * 128 + 111]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5481] * kernel_shared_1[threadIdx_x // 7 * 128 + 111]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5488] * kernel_shared_1[threadIdx_x // 7 * 128 + 112]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5495] * kernel_shared_1[threadIdx_x // 7 * 128 + 112]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5502] * kernel_shared_1[threadIdx_x // 7 * 128 + 112]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5509] * kernel_shared_1[threadIdx_x // 7 * 128 + 112]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5516] * kernel_shared_1[threadIdx_x // 7 * 128 + 112]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5523] * kernel_shared_1[threadIdx_x // 7 * 128 + 112]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5530] * kernel_shared_1[threadIdx_x // 7 * 128 + 112]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5537] * kernel_shared_1[threadIdx_x // 7 * 128 + 113]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5544] * kernel_shared_1[threadIdx_x // 7 * 128 + 113]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5551] * kernel_shared_1[threadIdx_x // 7 * 128 + 113]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5558] * kernel_shared_1[threadIdx_x // 7 * 128 + 113]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5565] * kernel_shared_1[threadIdx_x // 7 * 128 + 113]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5572] * kernel_shared_1[threadIdx_x // 7 * 128 + 113]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5579] * kernel_shared_1[threadIdx_x // 7 * 128 + 113]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5586] * kernel_shared_1[threadIdx_x // 7 * 128 + 114]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5593] * kernel_shared_1[threadIdx_x // 7 * 128 + 114]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5600] * kernel_shared_1[threadIdx_x // 7 * 128 + 114]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5607] * kernel_shared_1[threadIdx_x // 7 * 128 + 114]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5614] * kernel_shared_1[threadIdx_x // 7 * 128 + 114]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5621] * kernel_shared_1[threadIdx_x // 7 * 128 + 114]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5628] * kernel_shared_1[threadIdx_x // 7 * 128 + 114]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5635] * kernel_shared_1[threadIdx_x // 7 * 128 + 115]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5642] * kernel_shared_1[threadIdx_x // 7 * 128 + 115]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5649] * kernel_shared_1[threadIdx_x // 7 * 128 + 115]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5656] * kernel_shared_1[threadIdx_x // 7 * 128 + 115]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5663] * kernel_shared_1[threadIdx_x // 7 * 128 + 115]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5670] * kernel_shared_1[threadIdx_x // 7 * 128 + 115]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5677] * kernel_shared_1[threadIdx_x // 7 * 128 + 115]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5684] * kernel_shared_1[threadIdx_x // 7 * 128 + 116]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5691] * kernel_shared_1[threadIdx_x // 7 * 128 + 116]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5698] * kernel_shared_1[threadIdx_x // 7 * 128 + 116]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5705] * kernel_shared_1[threadIdx_x // 7 * 128 + 116]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5712] * kernel_shared_1[threadIdx_x // 7 * 128 + 116]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5719] * kernel_shared_1[threadIdx_x // 7 * 128 + 116]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5726] * kernel_shared_1[threadIdx_x // 7 * 128 + 116]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5733] * kernel_shared_1[threadIdx_x // 7 * 128 + 117]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5740] * kernel_shared_1[threadIdx_x // 7 * 128 + 117]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5747] * kernel_shared_1[threadIdx_x // 7 * 128 + 117]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5754] * kernel_shared_1[threadIdx_x // 7 * 128 + 117]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5761] * kernel_shared_1[threadIdx_x // 7 * 128 + 117]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5768] * kernel_shared_1[threadIdx_x // 7 * 128 + 117]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5775] * kernel_shared_1[threadIdx_x // 7 * 128 + 117]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5782] * kernel_shared_1[threadIdx_x // 7 * 128 + 118]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5789] * kernel_shared_1[threadIdx_x // 7 * 128 + 118]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5796] * kernel_shared_1[threadIdx_x // 7 * 128 + 118]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5803] * kernel_shared_1[threadIdx_x // 7 * 128 + 118]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5810] * kernel_shared_1[threadIdx_x // 7 * 128 + 118]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5817] * kernel_shared_1[threadIdx_x // 7 * 128 + 118]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5824] * kernel_shared_1[threadIdx_x // 7 * 128 + 118]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5831] * kernel_shared_1[threadIdx_x // 7 * 128 + 119]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5838] * kernel_shared_1[threadIdx_x // 7 * 128 + 119]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5845] * kernel_shared_1[threadIdx_x // 7 * 128 + 119]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5852] * kernel_shared_1[threadIdx_x // 7 * 128 + 119]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5859] * kernel_shared_1[threadIdx_x // 7 * 128 + 119]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5866] * kernel_shared_1[threadIdx_x // 7 * 128 + 119]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5873] * kernel_shared_1[threadIdx_x // 7 * 128 + 119]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5880] * kernel_shared_1[threadIdx_x // 7 * 128 + 120]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5887] * kernel_shared_1[threadIdx_x // 7 * 128 + 120]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5894] * kernel_shared_1[threadIdx_x // 7 * 128 + 120]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5901] * kernel_shared_1[threadIdx_x // 7 * 128 + 120]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5908] * kernel_shared_1[threadIdx_x // 7 * 128 + 120]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5915] * kernel_shared_1[threadIdx_x // 7 * 128 + 120]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5922] * kernel_shared_1[threadIdx_x // 7 * 128 + 120]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5929] * kernel_shared_1[threadIdx_x // 7 * 128 + 121]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5936] * kernel_shared_1[threadIdx_x // 7 * 128 + 121]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5943] * kernel_shared_1[threadIdx_x // 7 * 128 + 121]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5950] * kernel_shared_1[threadIdx_x // 7 * 128 + 121]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 5957] * kernel_shared_1[threadIdx_x // 7 * 128 + 121]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 5964] * kernel_shared_1[threadIdx_x // 7 * 128 + 121]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 5971] * kernel_shared_1[threadIdx_x // 7 * 128 + 121]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 5978] * kernel_shared_1[threadIdx_x // 7 * 128 + 122]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 5985] * kernel_shared_1[threadIdx_x // 7 * 128 + 122]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 5992] * kernel_shared_1[threadIdx_x // 7 * 128 + 122]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 5999] * kernel_shared_1[threadIdx_x // 7 * 128 + 122]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 6006] * kernel_shared_1[threadIdx_x // 7 * 128 + 122]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 6013] * kernel_shared_1[threadIdx_x // 7 * 128 + 122]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 6020] * kernel_shared_1[threadIdx_x // 7 * 128 + 122]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 6027] * kernel_shared_1[threadIdx_x // 7 * 128 + 123]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 6034] * kernel_shared_1[threadIdx_x // 7 * 128 + 123]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 6041] * kernel_shared_1[threadIdx_x // 7 * 128 + 123]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 6048] * kernel_shared_1[threadIdx_x // 7 * 128 + 123]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 6055] * kernel_shared_1[threadIdx_x // 7 * 128 + 123]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 6062] * kernel_shared_1[threadIdx_x // 7 * 128 + 123]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 6069] * kernel_shared_1[threadIdx_x // 7 * 128 + 123]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 6076] * kernel_shared_1[threadIdx_x // 7 * 128 + 124]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 6083] * kernel_shared_1[threadIdx_x // 7 * 128 + 124]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 6090] * kernel_shared_1[threadIdx_x // 7 * 128 + 124]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 6097] * kernel_shared_1[threadIdx_x // 7 * 128 + 124]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 6104] * kernel_shared_1[threadIdx_x // 7 * 128 + 124]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 6111] * kernel_shared_1[threadIdx_x // 7 * 128 + 124]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 6118] * kernel_shared_1[threadIdx_x // 7 * 128 + 124]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 6125] * kernel_shared_1[threadIdx_x // 7 * 128 + 125]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 6132] * kernel_shared_1[threadIdx_x // 7 * 128 + 125]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 6139] * kernel_shared_1[threadIdx_x // 7 * 128 + 125]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 6146] * kernel_shared_1[threadIdx_x // 7 * 128 + 125]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 6153] * kernel_shared_1[threadIdx_x // 7 * 128 + 125]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 6160] * kernel_shared_1[threadIdx_x // 7 * 128 + 125]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 6167] * kernel_shared_1[threadIdx_x // 7 * 128 + 125]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 6174] * kernel_shared_1[threadIdx_x // 7 * 128 + 126]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 6181] * kernel_shared_1[threadIdx_x // 7 * 128 + 126]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 6188] * kernel_shared_1[threadIdx_x // 7 * 128 + 126]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 6195] * kernel_shared_1[threadIdx_x // 7 * 128 + 126]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 6202] * kernel_shared_1[threadIdx_x // 7 * 128 + 126]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 6209] * kernel_shared_1[threadIdx_x // 7 * 128 + 126]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 6216] * kernel_shared_1[threadIdx_x // 7 * 128 + 126]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[threadIdx_x % 7 + 6223] * kernel_shared_1[threadIdx_x // 7 * 128 + 127]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[threadIdx_x % 7 + 6230] * kernel_shared_1[threadIdx_x // 7 * 128 + 127]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[threadIdx_x % 7 + 6237] * kernel_shared_1[threadIdx_x // 7 * 128 + 127]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[threadIdx_x % 7 + 6244] * kernel_shared_1[threadIdx_x // 7 * 128 + 127]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[threadIdx_x % 7 + 6251] * kernel_shared_1[threadIdx_x // 7 * 128 + 127]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[threadIdx_x % 7 + 6258] * kernel_shared_1[threadIdx_x // 7 * 128 + 127]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[threadIdx_x % 7 + 6265] * kernel_shared_1[threadIdx_x // 7 * 128 + 127]
- compute_1 = T.Buffer((25088,), data=compute.data)
- bias_1 = T.Buffer((512,), data=bias.data)
- compute_1[blockIdx_x * 784 + threadIdx_x // 7 * 49 + threadIdx_x % 7] = T.max(conv2d_nchw_1[0] + bias_1[blockIdx_x * 16 + threadIdx_x // 7], T.float32(0))
- compute_1[blockIdx_x * 784 + threadIdx_x // 7 * 49 + threadIdx_x % 7 + 7] = T.max(conv2d_nchw_1[1] + bias_1[blockIdx_x * 16 + threadIdx_x // 7], T.float32(0))
- compute_1[blockIdx_x * 784 + threadIdx_x // 7 * 49 + threadIdx_x % 7 + 14] = T.max(conv2d_nchw_1[2] + bias_1[blockIdx_x * 16 + threadIdx_x // 7], T.float32(0))
- compute_1[blockIdx_x * 784 + threadIdx_x // 7 * 49 + threadIdx_x % 7 + 21] = T.max(conv2d_nchw_1[3] + bias_1[blockIdx_x * 16 + threadIdx_x // 7], T.float32(0))
- compute_1[blockIdx_x * 784 + threadIdx_x // 7 * 49 + threadIdx_x % 7 + 28] = T.max(conv2d_nchw_1[4] + bias_1[blockIdx_x * 16 + threadIdx_x // 7], T.float32(0))
- compute_1[blockIdx_x * 784 + threadIdx_x // 7 * 49 + threadIdx_x % 7 + 35] = T.max(conv2d_nchw_1[5] + bias_1[blockIdx_x * 16 + threadIdx_x // 7], T.float32(0))
- compute_1[blockIdx_x * 784 + threadIdx_x // 7 * 49 + threadIdx_x % 7 + 42] = T.max(conv2d_nchw_1[6] + bias_1[blockIdx_x * 16 + threadIdx_x // 7], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1] = T.if_then_else(1 <= threadIdx_x_1 // 9 + ry_outer_outer and threadIdx_x_1 // 9 + ry_outer_outer < 8 and 1 <= threadIdx_x_1 % 9 and threadIdx_x_1 % 9 < 8, data_1[cse_var_4 + threadIdx_x_1 // 9 * 7 + cse_var_3 + 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(1 <= (threadIdx_x_1 + 56) % 63 // 9 + ry_outer_outer and (threadIdx_x_1 + 56) % 63 // 9 + ry_outer_outer < 8 and 1 <= (threadIdx_x_1 + 2) % 9 and (threadIdx_x_1 + 2) % 9 < 8, data_1[cse_var_4 + (threadIdx_x_1 + 56) // 9 * 7 + cse_var_3 + (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(1 <= (threadIdx_x_1 + 49) % 63 // 9 + ry_outer_outer and (threadIdx_x_1 + 49) % 63 // 9 + ry_outer_outer < 8 and 1 <= (threadIdx_x_1 + 4) % 9 and (threadIdx_x_1 + 4) % 9 < 8, data_1[cse_var_4 + (threadIdx_x_1 + 112) // 9 * 7 + cse_var_3 + (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(1 <= (threadIdx_x_1 + 42) % 63 // 9 + ry_outer_outer and (threadIdx_x_1 + 42) % 63 // 9 + ry_outer_outer < 8 and 1 <= (threadIdx_x_1 + 6) % 9 and (threadIdx_x_1 + 6) % 9 < 8, data_1[cse_var_4 + (threadIdx_x_1 + 168) // 9 * 7 + cse_var_3 + (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(1 <= (threadIdx_x_1 + 35) % 63 // 9 + ry_outer_outer and (threadIdx_x_1 + 35) % 63 // 9 + ry_outer_outer < 8 and 1 <= (threadIdx_x_1 + 8) % 9 and (threadIdx_x_1 + 8) % 9 < 8, data_1[cse_var_4 + (threadIdx_x_1 + 224) // 9 * 7 + cse_var_3 + (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(1 <= (threadIdx_x_1 + 28) % 63 // 9 + ry_outer_outer and (threadIdx_x_1 + 28) % 63 // 9 + ry_outer_outer < 8 and 1 <= (threadIdx_x_1 + 1) % 9 and (threadIdx_x_1 + 1) % 9 < 8, data_1[cse_var_4 + (threadIdx_x_1 + 280) // 9 * 7 + cse_var_3 + (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 + 21) % 63 // 9 + ry_outer_outer and (threadIdx_x_1 + 21) % 63 // 9 + ry_outer_outer < 8 and 1 <= (threadIdx_x_1 + 3) % 9 and (threadIdx_x_1 + 3) % 9 < 8, data_1[cse_var_4 + (threadIdx_x_1 + 336) // 9 * 7 + cse_var_3 + (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(1 <= (threadIdx_x_1 + 14) % 63 // 9 + ry_outer_outer and (threadIdx_x_1 + 14) % 63 // 9 + ry_outer_outer < 8 and 1 <= (threadIdx_x_1 + 5) % 9 and (threadIdx_x_1 + 5) % 9 < 8, data_1[cse_var_4 + (threadIdx_x_1 + 392) // 9 * 7 + cse_var_3 + (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(1 <= (threadIdx_x_1 + 7) // 9 + ry_outer_outer and (threadIdx_x_1 + 7) // 9 + ry_outer_outer < 8 and 1 <= (threadIdx_x_1 + 7) % 9 and (threadIdx_x_1 + 7) % 9 < 8, data_1[cse_var_4 + (threadIdx_x_1 + 448) // 9 * 7 + cse_var_3 + (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(1 <= threadIdx_x_1 // 9 + ry_outer_outer and threadIdx_x_1 // 9 + ry_outer_outer < 8 and 1 <= threadIdx_x_1 % 9 and threadIdx_x_1 % 9 < 8, data_1[cse_var_4 + threadIdx_x_1 // 9 * 7 + cse_var_3 + threadIdx_x_1 % 9 + 384], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 56):
+ pad_temp_shared_1[threadIdx_x_1 + 560] = T.if_then_else(1 <= (threadIdx_x_1 + 56) % 63 // 9 + ry_outer_outer and (threadIdx_x_1 + 56) % 63 // 9 + ry_outer_outer < 8 and 1 <= (threadIdx_x_1 + 2) % 9 and (threadIdx_x_1 + 2) % 9 < 8, data_1[cse_var_4 + (threadIdx_x_1 + 560) // 9 * 7 + cse_var_3 + (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(1 <= (threadIdx_x_1 + 49) % 63 // 9 + ry_outer_outer and (threadIdx_x_1 + 49) % 63 // 9 + ry_outer_outer < 8 and 1 <= (threadIdx_x_1 + 4) % 9 and (threadIdx_x_1 + 4) % 9 < 8, data_1[cse_var_4 + (threadIdx_x_1 + 616) // 9 * 7 + cse_var_3 + (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(1 <= (threadIdx_x_1 + 42) % 63 // 9 + ry_outer_outer and (threadIdx_x_1 + 42) % 63 // 9 + ry_outer_outer < 8 and 1 <= (threadIdx_x_1 + 6) % 9 and (threadIdx_x_1 + 6) % 9 < 8, data_1[cse_var_4 + (threadIdx_x_1 + 672) // 9 * 7 + cse_var_3 + (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(1 <= (threadIdx_x_1 + 35) % 63 // 9 + ry_outer_outer and (threadIdx_x_1 + 35) % 63 // 9 + ry_outer_outer < 8 and 1 <= (threadIdx_x_1 + 8) % 9 and (threadIdx_x_1 + 8) % 9 < 8, data_1[cse_var_4 + (threadIdx_x_1 + 728) // 9 * 7 + cse_var_3 + (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(1 <= (threadIdx_x_1 + 28) % 63 // 9 + ry_outer_outer and (threadIdx_x_1 + 28) % 63 // 9 + ry_outer_outer < 8 and 1 <= (threadIdx_x_1 + 1) % 9 and (threadIdx_x_1 + 1) % 9 < 8, data_1[cse_var_4 + (threadIdx_x_1 + 784) // 9 * 7 + cse_var_3 + (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(1 <= (threadIdx_x_1 + 21) % 63 // 9 + ry_outer_outer and (threadIdx_x_1 + 21) % 63 // 9 + ry_outer_outer < 8 and 1 <= (threadIdx_x_1 + 3) % 9 and (threadIdx_x_1 + 3) % 9 < 8, data_1[cse_var_4 + (threadIdx_x_1 + 840) // 9 * 7 + cse_var_3 + (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(1 <= (threadIdx_x_1 + 14) % 63 // 9 + ry_outer_outer and (threadIdx_x_1 + 14) % 63 // 9 + ry_outer_outer < 8 and 1 <= (threadIdx_x_1 + 5) % 9 and (threadIdx_x_1 + 5) % 9 < 8, data_1[cse_var_4 + (threadIdx_x_1 + 896) // 9 * 7 + cse_var_3 + (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(1 <= (threadIdx_x_1 + 7) // 9 + ry_outer_outer and (threadIdx_x_1 + 7) // 9 + ry_outer_outer < 8 and 1 <= (threadIdx_x_1 + 7) % 9 and (threadIdx_x_1 + 7) % 9 < 8, data_1[cse_var_4 + (threadIdx_x_1 + 952) // 9 * 7 + cse_var_3 + (threadIdx_x_1 + 7) % 9 - 8], T.float32(0))
+ threadIdx_x_2 = T.env_thread("threadIdx.x")
+ kernel_shared_1 = T.Buffer((1536,), 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 * 147456 + threadIdx_x_2 // 48 * 4608 + cse_var_2 + threadIdx_x_2 % 48 // 3 * 9 + cse_var_1 + threadIdx_x_2 % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 56) // 48 * 48 + (threadIdx_x_2 + 8) % 48 // 3 * 3 + (threadIdx_x_2 + 2) % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 56) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 + 8) % 48 // 3 * 9 + cse_var_1 + (threadIdx_x_2 + 2) % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 112) // 48 * 48 + (threadIdx_x_2 + 16) % 48 // 3 * 3 + (threadIdx_x_2 + 1) % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 112) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 + 16) % 48 // 3 * 9 + cse_var_1 + (threadIdx_x_2 + 1) % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 168) // 48 * 48 + (threadIdx_x_2 // 3 + 8) % 16 * 3 + threadIdx_x_2 % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 168) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 // 3 + 8) % 16 * 9 + cse_var_1 + threadIdx_x_2 % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 224) // 48 * 48 + (threadIdx_x_2 + 32) % 48 // 3 * 3 + (threadIdx_x_2 + 2) % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 224) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 + 32) % 48 // 3 * 9 + cse_var_1 + (threadIdx_x_2 + 2) % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 280) // 48 * 48 + (threadIdx_x_2 + 40) % 48 // 3 * 3 + (threadIdx_x_2 + 1) % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 280) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 + 40) % 48 // 3 * 9 + cse_var_1 + (threadIdx_x_2 + 1) % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 336] = kernel_1[blockIdx_x * 147456 + threadIdx_x_2 // 48 * 4608 + cse_var_2 + threadIdx_x_2 % 48 // 3 * 9 + cse_var_1 + threadIdx_x_2 % 3 + 32256]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 392) // 48 * 48 + (threadIdx_x_2 + 8) % 48 // 3 * 3 + (threadIdx_x_2 + 2) % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 392) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 + 8) % 48 // 3 * 9 + cse_var_1 + (threadIdx_x_2 + 2) % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 448) // 48 * 48 + (threadIdx_x_2 + 16) % 48 // 3 * 3 + (threadIdx_x_2 + 1) % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 448) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 + 16) % 48 // 3 * 9 + cse_var_1 + (threadIdx_x_2 + 1) % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 504) // 48 * 48 + (threadIdx_x_2 // 3 + 8) % 16 * 3 + threadIdx_x_2 % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 504) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 // 3 + 8) % 16 * 9 + cse_var_1 + threadIdx_x_2 % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 560) // 48 * 48 + (threadIdx_x_2 + 32) % 48 // 3 * 3 + (threadIdx_x_2 + 2) % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 560) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 + 32) % 48 // 3 * 9 + cse_var_1 + (threadIdx_x_2 + 2) % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 616) // 48 * 48 + (threadIdx_x_2 + 40) % 48 // 3 * 3 + (threadIdx_x_2 + 1) % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 616) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 + 40) % 48 // 3 * 9 + cse_var_1 + (threadIdx_x_2 + 1) % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 672] = kernel_1[blockIdx_x * 147456 + threadIdx_x_2 // 48 * 4608 + cse_var_2 + threadIdx_x_2 % 48 // 3 * 9 + cse_var_1 + threadIdx_x_2 % 3 + 64512]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 728) // 48 * 48 + (threadIdx_x_2 + 8) % 48 // 3 * 3 + (threadIdx_x_2 + 2) % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 728) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 + 8) % 48 // 3 * 9 + cse_var_1 + (threadIdx_x_2 + 2) % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 784) // 48 * 48 + (threadIdx_x_2 + 16) % 48 // 3 * 3 + (threadIdx_x_2 + 1) % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 784) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 + 16) % 48 // 3 * 9 + cse_var_1 + (threadIdx_x_2 + 1) % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 840) // 48 * 48 + (threadIdx_x_2 // 3 + 8) % 16 * 3 + threadIdx_x_2 % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 840) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 // 3 + 8) % 16 * 9 + cse_var_1 + threadIdx_x_2 % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 896) // 48 * 48 + (threadIdx_x_2 + 32) % 48 // 3 * 3 + (threadIdx_x_2 + 2) % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 896) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 + 32) % 48 // 3 * 9 + cse_var_1 + (threadIdx_x_2 + 2) % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 952) // 48 * 48 + (threadIdx_x_2 + 40) % 48 // 3 * 3 + (threadIdx_x_2 + 1) % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 952) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 + 40) % 48 // 3 * 9 + cse_var_1 + (threadIdx_x_2 + 1) % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1008] = kernel_1[blockIdx_x * 147456 + threadIdx_x_2 // 48 * 4608 + cse_var_2 + threadIdx_x_2 % 48 // 3 * 9 + cse_var_1 + threadIdx_x_2 % 3 + 96768]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 1064) // 48 * 48 + (threadIdx_x_2 + 8) % 48 // 3 * 3 + (threadIdx_x_2 + 2) % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 1064) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 + 8) % 48 // 3 * 9 + cse_var_1 + (threadIdx_x_2 + 2) % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 1120) // 48 * 48 + (threadIdx_x_2 + 16) % 48 // 3 * 3 + (threadIdx_x_2 + 1) % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 1120) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 + 16) % 48 // 3 * 9 + cse_var_1 + (threadIdx_x_2 + 1) % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 1176) // 48 * 48 + (threadIdx_x_2 // 3 + 8) % 16 * 3 + threadIdx_x_2 % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 1176) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 // 3 + 8) % 16 * 9 + cse_var_1 + threadIdx_x_2 % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 1232) // 48 * 48 + (threadIdx_x_2 + 32) % 48 // 3 * 3 + (threadIdx_x_2 + 2) % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 1232) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 + 32) % 48 // 3 * 9 + cse_var_1 + (threadIdx_x_2 + 2) % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 1288) // 48 * 48 + (threadIdx_x_2 + 40) % 48 // 3 * 3 + (threadIdx_x_2 + 1) % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 1288) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 + 40) % 48 // 3 * 9 + cse_var_1 + (threadIdx_x_2 + 1) % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[threadIdx_x_2 + 1344] = kernel_1[blockIdx_x * 147456 + threadIdx_x_2 // 48 * 4608 + cse_var_2 + threadIdx_x_2 % 48 // 3 * 9 + cse_var_1 + threadIdx_x_2 % 3 + 129024]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 1400) // 48 * 48 + (threadIdx_x_2 + 8) % 48 // 3 * 3 + (threadIdx_x_2 + 2) % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 1400) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 + 8) % 48 // 3 * 9 + cse_var_1 + (threadIdx_x_2 + 2) % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ kernel_shared_1[(threadIdx_x_2 + 1456) // 48 * 48 + (threadIdx_x_2 + 16) % 48 // 3 * 3 + (threadIdx_x_2 + 1) % 3] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 1456) // 48 * 4608 + cse_var_2 + (threadIdx_x_2 + 16) % 48 // 3 * 9 + cse_var_1 + (threadIdx_x_2 + 1) % 3]
+ with T.launch_thread(threadIdx_x_2, 56):
+ if T.likely(threadIdx_x_2 < 24):
+ kernel_shared_1[(threadIdx_x_2 + 1512) // 48 * 48 + threadIdx_x_2 + 24] = kernel_1[blockIdx_x * 147456 + (threadIdx_x_2 + 1512) // 48 * 4608 + cse_var_2 + threadIdx_x_2 // 3 * 9 + cse_var_1 + threadIdx_x_2 % 3 + 72]
+ for rc_outer_inner, rx_outer_inner in T.grid(2, 3):
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 9] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 18] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 27] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 36] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 45] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 54] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 63] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 3]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 72] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 3]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 81] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 3]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 90] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 3]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 99] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 3]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 108] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 3]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 117] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 3]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 126] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 6]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 135] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 6]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 144] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 6]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 153] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 6]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 162] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 6]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 171] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 6]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 180] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 6]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 189] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 9]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 198] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 9]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 207] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 9]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 216] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 9]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 225] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 9]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 234] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 9]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 243] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 9]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 252] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 12]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 261] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 12]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 270] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 12]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 279] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 12]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 288] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 12]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 297] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 12]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 306] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 12]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 315] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 15]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 324] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 15]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 333] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 15]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 342] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 15]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 351] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 15]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 360] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 15]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 369] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 15]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 378] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 18]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 387] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 18]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 396] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 18]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 405] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 18]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 414] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 18]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 423] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 18]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 432] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 18]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 441] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 21]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 450] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 21]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 459] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 21]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 468] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 21]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 477] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 21]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 486] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 21]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 495] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 21]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 48]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 9] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 48]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 18] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 48]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 27] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 48]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 36] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 48]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 45] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 48]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 54] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 48]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 63] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 51]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 72] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 51]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 81] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 51]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 90] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 51]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 99] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 51]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 108] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 51]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 117] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 51]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 126] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 54]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 135] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 54]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 144] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 54]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 153] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 54]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 162] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 54]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 171] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 54]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 180] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 54]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 189] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 57]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 198] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 57]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 207] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 57]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 216] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 57]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 225] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 57]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 234] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 57]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 243] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 57]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 252] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 60]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 261] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 60]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 270] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 60]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 279] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 60]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 288] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 60]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 297] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 60]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 306] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 60]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 315] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 63]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 324] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 63]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 333] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 63]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 342] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 63]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 351] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 63]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 360] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 63]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 369] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 63]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 378] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 66]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 387] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 66]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 396] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 66]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 405] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 66]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 414] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 66]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 423] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 66]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 432] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 66]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 441] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 69]
+ conv2d_nchw_1[8] = conv2d_nchw_1[8] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 450] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 69]
+ conv2d_nchw_1[9] = conv2d_nchw_1[9] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 459] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 69]
+ conv2d_nchw_1[10] = conv2d_nchw_1[10] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 468] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 69]
+ conv2d_nchw_1[11] = conv2d_nchw_1[11] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 477] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 69]
+ conv2d_nchw_1[12] = conv2d_nchw_1[12] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 486] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 69]
+ conv2d_nchw_1[13] = conv2d_nchw_1[13] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 495] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 69]
+ conv2d_nchw_1[14] = conv2d_nchw_1[14] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 96]
+ conv2d_nchw_1[15] = conv2d_nchw_1[15] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 9] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 96]
+ conv2d_nchw_1[16] = conv2d_nchw_1[16] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 18] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 96]
+ conv2d_nchw_1[17] = conv2d_nchw_1[17] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 27] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 96]
+ conv2d_nchw_1[18] = conv2d_nchw_1[18] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 36] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 96]
+ conv2d_nchw_1[19] = conv2d_nchw_1[19] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 45] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 96]
+ conv2d_nchw_1[20] = conv2d_nchw_1[20] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 54] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 96]
+ conv2d_nchw_1[14] = conv2d_nchw_1[14] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 63] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 99]
+ conv2d_nchw_1[15] = conv2d_nchw_1[15] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 72] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 99]
+ conv2d_nchw_1[16] = conv2d_nchw_1[16] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 81] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 99]
+ conv2d_nchw_1[17] = conv2d_nchw_1[17] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 90] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 99]
+ conv2d_nchw_1[18] = conv2d_nchw_1[18] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 99] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 99]
+ conv2d_nchw_1[19] = conv2d_nchw_1[19] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 108] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 99]
+ conv2d_nchw_1[20] = conv2d_nchw_1[20] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 117] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 99]
+ conv2d_nchw_1[14] = conv2d_nchw_1[14] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 126] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 102]
+ conv2d_nchw_1[15] = conv2d_nchw_1[15] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 135] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 102]
+ conv2d_nchw_1[16] = conv2d_nchw_1[16] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 144] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 102]
+ conv2d_nchw_1[17] = conv2d_nchw_1[17] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 153] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 102]
+ conv2d_nchw_1[18] = conv2d_nchw_1[18] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 162] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 102]
+ conv2d_nchw_1[19] = conv2d_nchw_1[19] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 171] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 102]
+ conv2d_nchw_1[20] = conv2d_nchw_1[20] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 180] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 102]
+ conv2d_nchw_1[14] = conv2d_nchw_1[14] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 189] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 105]
+ conv2d_nchw_1[15] = conv2d_nchw_1[15] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 198] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 105]
+ conv2d_nchw_1[16] = conv2d_nchw_1[16] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 207] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 105]
+ conv2d_nchw_1[17] = conv2d_nchw_1[17] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 216] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 105]
+ conv2d_nchw_1[18] = conv2d_nchw_1[18] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 225] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 105]
+ conv2d_nchw_1[19] = conv2d_nchw_1[19] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 234] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 105]
+ conv2d_nchw_1[20] = conv2d_nchw_1[20] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 243] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 105]
+ conv2d_nchw_1[14] = conv2d_nchw_1[14] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 252] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 108]
+ conv2d_nchw_1[15] = conv2d_nchw_1[15] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 261] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 108]
+ conv2d_nchw_1[16] = conv2d_nchw_1[16] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 270] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 108]
+ conv2d_nchw_1[17] = conv2d_nchw_1[17] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 279] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 108]
+ conv2d_nchw_1[18] = conv2d_nchw_1[18] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 288] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 108]
+ conv2d_nchw_1[19] = conv2d_nchw_1[19] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 297] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 108]
+ conv2d_nchw_1[20] = conv2d_nchw_1[20] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 306] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 108]
+ conv2d_nchw_1[14] = conv2d_nchw_1[14] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 315] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 111]
+ conv2d_nchw_1[15] = conv2d_nchw_1[15] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 324] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 111]
+ conv2d_nchw_1[16] = conv2d_nchw_1[16] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 333] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 111]
+ conv2d_nchw_1[17] = conv2d_nchw_1[17] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 342] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 111]
+ conv2d_nchw_1[18] = conv2d_nchw_1[18] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 351] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 111]
+ conv2d_nchw_1[19] = conv2d_nchw_1[19] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 360] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 111]
+ conv2d_nchw_1[20] = conv2d_nchw_1[20] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 369] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 111]
+ conv2d_nchw_1[14] = conv2d_nchw_1[14] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 378] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 114]
+ conv2d_nchw_1[15] = conv2d_nchw_1[15] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 387] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 114]
+ conv2d_nchw_1[16] = conv2d_nchw_1[16] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 396] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 114]
+ conv2d_nchw_1[17] = conv2d_nchw_1[17] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 405] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 114]
+ conv2d_nchw_1[18] = conv2d_nchw_1[18] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 414] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 114]
+ conv2d_nchw_1[19] = conv2d_nchw_1[19] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 423] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 114]
+ conv2d_nchw_1[20] = conv2d_nchw_1[20] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 432] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 114]
+ conv2d_nchw_1[14] = conv2d_nchw_1[14] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 441] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 117]
+ conv2d_nchw_1[15] = conv2d_nchw_1[15] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 450] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 117]
+ conv2d_nchw_1[16] = conv2d_nchw_1[16] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 459] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 117]
+ conv2d_nchw_1[17] = conv2d_nchw_1[17] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 468] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 117]
+ conv2d_nchw_1[18] = conv2d_nchw_1[18] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 477] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 117]
+ conv2d_nchw_1[19] = conv2d_nchw_1[19] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 486] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 117]
+ conv2d_nchw_1[20] = conv2d_nchw_1[20] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 495] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 117]
+ conv2d_nchw_1[21] = conv2d_nchw_1[21] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 144]
+ conv2d_nchw_1[22] = conv2d_nchw_1[22] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 9] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 144]
+ conv2d_nchw_1[23] = conv2d_nchw_1[23] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 18] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 144]
+ conv2d_nchw_1[24] = conv2d_nchw_1[24] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 27] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 144]
+ conv2d_nchw_1[25] = conv2d_nchw_1[25] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 36] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 144]
+ conv2d_nchw_1[26] = conv2d_nchw_1[26] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 45] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 144]
+ conv2d_nchw_1[27] = conv2d_nchw_1[27] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 54] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 144]
+ conv2d_nchw_1[21] = conv2d_nchw_1[21] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 63] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 147]
+ conv2d_nchw_1[22] = conv2d_nchw_1[22] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 72] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 147]
+ conv2d_nchw_1[23] = conv2d_nchw_1[23] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 81] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 147]
+ conv2d_nchw_1[24] = conv2d_nchw_1[24] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 90] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 147]
+ conv2d_nchw_1[25] = conv2d_nchw_1[25] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 99] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 147]
+ conv2d_nchw_1[26] = conv2d_nchw_1[26] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 108] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 147]
+ conv2d_nchw_1[27] = conv2d_nchw_1[27] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 117] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 147]
+ conv2d_nchw_1[21] = conv2d_nchw_1[21] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 126] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 150]
+ conv2d_nchw_1[22] = conv2d_nchw_1[22] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 135] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 150]
+ conv2d_nchw_1[23] = conv2d_nchw_1[23] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 144] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 150]
+ conv2d_nchw_1[24] = conv2d_nchw_1[24] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 153] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 150]
+ conv2d_nchw_1[25] = conv2d_nchw_1[25] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 162] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 150]
+ conv2d_nchw_1[26] = conv2d_nchw_1[26] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 171] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 150]
+ conv2d_nchw_1[27] = conv2d_nchw_1[27] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 180] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 150]
+ conv2d_nchw_1[21] = conv2d_nchw_1[21] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 189] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 153]
+ conv2d_nchw_1[22] = conv2d_nchw_1[22] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 198] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 153]
+ conv2d_nchw_1[23] = conv2d_nchw_1[23] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 207] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 153]
+ conv2d_nchw_1[24] = conv2d_nchw_1[24] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 216] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 153]
+ conv2d_nchw_1[25] = conv2d_nchw_1[25] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 225] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 153]
+ conv2d_nchw_1[26] = conv2d_nchw_1[26] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 234] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 153]
+ conv2d_nchw_1[27] = conv2d_nchw_1[27] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 243] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 153]
+ conv2d_nchw_1[21] = conv2d_nchw_1[21] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 252] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 156]
+ conv2d_nchw_1[22] = conv2d_nchw_1[22] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 261] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 156]
+ conv2d_nchw_1[23] = conv2d_nchw_1[23] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 270] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 156]
+ conv2d_nchw_1[24] = conv2d_nchw_1[24] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 279] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 156]
+ conv2d_nchw_1[25] = conv2d_nchw_1[25] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 288] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 156]
+ conv2d_nchw_1[26] = conv2d_nchw_1[26] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 297] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 156]
+ conv2d_nchw_1[27] = conv2d_nchw_1[27] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 306] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 156]
+ conv2d_nchw_1[21] = conv2d_nchw_1[21] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 315] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 159]
+ conv2d_nchw_1[22] = conv2d_nchw_1[22] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 324] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 159]
+ conv2d_nchw_1[23] = conv2d_nchw_1[23] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 333] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 159]
+ conv2d_nchw_1[24] = conv2d_nchw_1[24] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 342] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 159]
+ conv2d_nchw_1[25] = conv2d_nchw_1[25] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 351] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 159]
+ conv2d_nchw_1[26] = conv2d_nchw_1[26] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 360] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 159]
+ conv2d_nchw_1[27] = conv2d_nchw_1[27] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 369] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 159]
+ conv2d_nchw_1[21] = conv2d_nchw_1[21] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 378] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 162]
+ conv2d_nchw_1[22] = conv2d_nchw_1[22] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 387] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 162]
+ conv2d_nchw_1[23] = conv2d_nchw_1[23] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 396] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 162]
+ conv2d_nchw_1[24] = conv2d_nchw_1[24] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 405] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 162]
+ conv2d_nchw_1[25] = conv2d_nchw_1[25] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 414] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 162]
+ conv2d_nchw_1[26] = conv2d_nchw_1[26] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 423] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 162]
+ conv2d_nchw_1[27] = conv2d_nchw_1[27] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 432] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 162]
+ conv2d_nchw_1[21] = conv2d_nchw_1[21] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 441] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 165]
+ conv2d_nchw_1[22] = conv2d_nchw_1[22] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 450] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 165]
+ conv2d_nchw_1[23] = conv2d_nchw_1[23] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 459] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 165]
+ conv2d_nchw_1[24] = conv2d_nchw_1[24] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 468] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 165]
+ conv2d_nchw_1[25] = conv2d_nchw_1[25] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 477] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 165]
+ conv2d_nchw_1[26] = conv2d_nchw_1[26] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 486] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 165]
+ conv2d_nchw_1[27] = conv2d_nchw_1[27] + pad_temp_shared_1[rc_outer_inner * 504 + rx_outer_inner + threadIdx_x % 7 + 495] * kernel_shared_1[threadIdx_x // 7 * 192 + rc_outer_inner * 24 + rx_outer_inner + 165]
+ for i1_inner, i2_inner in T.grid(4, 7):
+ compute_1 = T.Buffer((25088,), data=compute.data)
+ bias_1 = T.Buffer((512,), data=bias.data)
+ compute_1[blockIdx_x * 1568 + threadIdx_x // 7 * 196 + i1_inner * 49 + i2_inner * 7 + threadIdx_x % 7] = T.max(conv2d_nchw_1[i1_inner * 7 + i2_inner] + bias_1[blockIdx_x * 32 + threadIdx_x // 7 * 4 + i1_inner], T.float32(0))
</pre></div>
</div>
</div>
@@ -1644,7 +910,7 @@ 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: 0.388 ms
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Execution time of this operator: 0.281 ms
</pre></div>
</div>
</div>
@@ -1674,33 +940,33 @@ conv2d_nchw_nn_o_o_i, conv2d_nchw_nn_o_i = s[conv2d_nchw].split(conv2d_nchw_nn_o
conv2d_nchw_nn_o_o_o_i, conv2d_nchw_nn_o_o_i = s[conv2d_nchw].split(conv2d_nchw_nn_o_o_i, factor=1)
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=1)
-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_i, conv2d_nchw_ff_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_i, factor=4)
+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_i, conv2d_nchw_yy_i = s[conv2d_nchw].split(conv2d_nchw_yy, factor=7)
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=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_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_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=4)
-conv2d_nchw_rc_o_o, conv2d_nchw_rc_o_i = s[conv2d_nchw].split(conv2d_nchw_rc_o_i, factor=32)
+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=2)
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_rx_o_o, conv2d_nchw_rx_o_i = s[conv2d_nchw].split(conv2d_nchw_rx_o_i, factor=1)
+conv2d_nchw_rx_o_o, conv2d_nchw_rx_o_i = s[conv2d_nchw].split(conv2d_nchw_rx_o_i, factor=3)
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=1)
-compute_i1_o_o_i, compute_i1_o_i = s[compute].split(compute_i1_o_i, factor=16)
+compute_i1_o_i, compute_i1_i = s[compute].split(compute_i1, factor=4)
+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_i, compute_i2_i = s[compute].split(compute_i2, factor=7)
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_i2_o_o_o, compute_i2_o_o_i = s[compute].split(compute_i2_o_o_i, factor=1)
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)
@@ -1720,16 +986,16 @@ s[compute].bind(compute_i0_o_o_i_i1_o_o_i_fused_i2_o_o_i_fused_i3_o_o_i_fused, t
compute_i0_o_i_i1_o_i_fused_i2_o_i_fused_i3_o_i_fused = s[compute].fuse(compute_i0_o_i, compute_i1_o_i, compute_i2_o_i, compute_i3_o_i)
s[compute].bind(compute_i0_o_i_i1_o_i_fused_i2_o_i_fused_i3_o_i_fused, te.thread_axis("threadIdx.x"))
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=32)
+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=112)
+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=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=112)
+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", 1024)
+s[conv2d_nchw].pragma(conv2d_nchw_nn_o_o_o_o, "auto_unroll_max_step", 512)
s[conv2d_nchw].pragma(conv2d_nchw_nn_o_o_o_o, "unroll_explicit", True)
CUDA source code:
@@ -1754,10 +1020,10 @@ CUDA source code:
#define int64_t long long
#define uint64_t unsigned long long
#endif
-extern "C" __global__ void __launch_bounds__(112) default_function_kernel0(float* __restrict__ data, float* __restrict__ kernel, float* __restrict__ compute, float* __restrict__ bias) {
- float conv2d_nchw[7];
- __shared__ float pad_temp_shared[6272];
- __shared__ float kernel_shared[2048];
+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[28];
+ __shared__ float pad_temp_shared[1008];
+ __shared__ float kernel_shared[1536];
conv2d_nchw[0] = 0.000000e+00f;
conv2d_nchw[1] = 0.000000e+00f;
conv2d_nchw[2] = 0.000000e+00f;
@@ -1765,1069 +1031,314 @@ extern "C" __global__ void __launch_bounds__(112) default_function_ker
conv2d_nchw[4] = 0.000000e+00f;
conv2d_nchw[5] = 0.000000e+00f;
conv2d_nchw[6] = 0.000000e+00f;
- for (int rc_outer_outer = 0; rc_outer_outer < 4; ++rc_outer_outer) {
+ conv2d_nchw[7] = 0.000000e+00f;
+ conv2d_nchw[8] = 0.000000e+00f;
+ conv2d_nchw[9] = 0.000000e+00f;
+ conv2d_nchw[10] = 0.000000e+00f;
+ conv2d_nchw[11] = 0.000000e+00f;
+ conv2d_nchw[12] = 0.000000e+00f;
+ conv2d_nchw[13] = 0.000000e+00f;
+ conv2d_nchw[14] = 0.000000e+00f;
+ conv2d_nchw[15] = 0.000000e+00f;
+ conv2d_nchw[16] = 0.000000e+00f;
+ conv2d_nchw[17] = 0.000000e+00f;
+ conv2d_nchw[18] = 0.000000e+00f;
+ conv2d_nchw[19] = 0.000000e+00f;
+ conv2d_nchw[20] = 0.000000e+00f;
+ conv2d_nchw[21] = 0.000000e+00f;
+ conv2d_nchw[22] = 0.000000e+00f;
+ conv2d_nchw[23] = 0.000000e+00f;
+ conv2d_nchw[24] = 0.000000e+00f;
+ conv2d_nchw[25] = 0.000000e+00f;
+ conv2d_nchw[26] = 0.000000e+00f;
+ conv2d_nchw[27] = 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();
- pad_temp_shared[((int)threadIdx.x)] = (((((1 <= (((((int)threadIdx.x) % 49) / 7) + ry_outer_outer)) && ((((((int)threadIdx.x) % 49) / 7) + ry_outer_outer) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) - 8)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 112)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 104)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 224)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 216)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 336)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 328)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 448)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 440)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 560)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 552)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 672)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 664)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 784)] = (((((1 <= (((((int)threadIdx.x) % 49) / 7) + ry_outer_outer)) && ((((((int)threadIdx.x) % 49) / 7) + ry_outer_outer) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 776)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 896)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 888)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 1008)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 1000)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 1120)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 1112)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 1232)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 1224)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 1344)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 1336)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 1456)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 1448)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 1568)] = (((((1 <= (((((int)threadIdx.x) % 49) / 7) + ry_outer_outer)) && ((((((int)threadIdx.x) % 49) / 7) + ry_outer_outer) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 1560)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 1680)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 1672)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 1792)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 1784)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 1904)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 1896)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 2016)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 2008)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 2128)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 2120)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 2240)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 2232)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 2352)] = (((((1 <= (((((int)threadIdx.x) % 49) / 7) + ry_outer_outer)) && ((((((int)threadIdx.x) % 49) / 7) + ry_outer_outer) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 2344)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 2464)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 2456)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 2576)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 2568)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 2688)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 2680)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 2800)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 2792)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 2912)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 2904)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 3024)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 3016)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 3136)] = (((((1 <= (((((int)threadIdx.x) % 49) / 7) + ry_outer_outer)) && ((((((int)threadIdx.x) % 49) / 7) + ry_outer_outer) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 3128)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 3248)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 3240)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 3360)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 3352)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 3472)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 3464)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 3584)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 3576)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 3696)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 3688)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 3808)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 3800)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 3920)] = (((((1 <= (((((int)threadIdx.x) % 49) / 7) + ry_outer_outer)) && ((((((int)threadIdx.x) % 49) / 7) + ry_outer_outer) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 3912)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 4032)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 4024)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 4144)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 4136)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 4256)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 4248)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 4368)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 4360)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 4480)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 4472)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 4592)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 4584)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 4704)] = (((((1 <= (((((int)threadIdx.x) % 49) / 7) + ry_outer_outer)) && ((((((int)threadIdx.x) % 49) / 7) + ry_outer_outer) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 4696)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 4816)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 4808)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 4928)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 4920)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 5040)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 5032)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 5152)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 5144)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 5264)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 5256)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 5376)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 5368)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 5488)] = (((((1 <= (((((int)threadIdx.x) % 49) / 7) + ry_outer_outer)) && ((((((int)threadIdx.x) % 49) / 7) + ry_outer_outer) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 5480)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 5600)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 5592)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 5712)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 5704)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 5824)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 5816)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 5936)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 5928)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 6048)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 6040)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 6160)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 6272) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 6152)] : 0.000000e+00f);
- if (((int)threadIdx.x) < 64) {
- kernel_shared[(((int)threadIdx.x) * 32)] = kernel[((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 1)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 9)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 2)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 18)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 3)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 27)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 4)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 36)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 5)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 45)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 6)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 54)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 7)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 63)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 8)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 72)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 9)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 81)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 10)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 90)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 11)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 99)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 12)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 108)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 13)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 117)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 14)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 126)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 15)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 135)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 16)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 144)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 17)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 153)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 18)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 162)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 19)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 171)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 20)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 180)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 21)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 189)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 22)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 198)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 23)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 207)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 24)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 216)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 25)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 225)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 26)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 234)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 27)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 243)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 28)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 252)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 29)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 261)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 30)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 270)];
- }
- if (((int)threadIdx.x) < 64) {
- kernel_shared[((((int)threadIdx.x) * 32) + 31)] = kernel[(((((((((int)blockIdx.x) * 73728) + ((((int)threadIdx.x) >> 2) * 4608)) + (rc_outer_outer * 1152)) + ((((int)threadIdx.x) & 3) * 288)) + (ry_outer_outer * 3)) + rx_outer_outer) + 279)];
+ __syncthreads();
+ pad_temp_shared[((int)threadIdx.x)] = (((((1 <= ((((int)threadIdx.x) / 9) + ry_outer_outer)) && (((((int)threadIdx.x) / 9) + ry_outer_outer) < 8)) && (1 <= (((int)threadIdx.x) % 9))) && ((((int)threadIdx.x) % 9) < 8)) ? data[(((((rc_outer_outer * 784) + ((((int)threadIdx.x) / 9) * 7)) + (ry_outer_outer * 7)) + (((int)threadIdx.x) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 56)] = (((((1 <= ((((((int)threadIdx.x) + 56) % 63) / 9) + ry_outer_outer)) && (((((((int)threadIdx.x) + 56) % 63) / 9) + ry_outer_outer) < 8)) && (1 <= ((((int)threadIdx.x) + 2) % 9))) && (((((int)threadIdx.x) + 2) % 9) < 8)) ? data[(((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 56) / 9) * 7)) + (ry_outer_outer * 7)) + ((((int)threadIdx.x) + 2) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 112)] = (((((1 <= ((((((int)threadIdx.x) + 49) % 63) / 9) + ry_outer_outer)) && (((((((int)threadIdx.x) + 49) % 63) / 9) + ry_outer_outer) < 8)) && (1 <= ((((int)threadIdx.x) + 4) % 9))) && (((((int)threadIdx.x) + 4) % 9) < 8)) ? data[(((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 112) / 9) * 7)) + (ry_outer_outer * 7)) + ((((int)threadIdx.x) + 4) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 168)] = (((((1 <= ((((((int)threadIdx.x) + 42) % 63) / 9) + ry_outer_outer)) && (((((((int)threadIdx.x) + 42) % 63) / 9) + ry_outer_outer) < 8)) && (1 <= ((((int)threadIdx.x) + 6) % 9))) && (((((int)threadIdx.x) + 6) % 9) < 8)) ? data[(((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 168) / 9) * 7)) + (ry_outer_outer * 7)) + ((((int)threadIdx.x) + 6) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 224)] = (((((1 <= ((((((int)threadIdx.x) + 35) % 63) / 9) + ry_outer_outer)) && (((((((int)threadIdx.x) + 35) % 63) / 9) + ry_outer_outer) < 8)) && (1 <= ((((int)threadIdx.x) + 8) % 9))) && (((((int)threadIdx.x) + 8) % 9) < 8)) ? data[(((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 224) / 9) * 7)) + (ry_outer_outer * 7)) + ((((int)threadIdx.x) + 8) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 280)] = (((((1 <= ((((((int)threadIdx.x) + 28) % 63) / 9) + ry_outer_outer)) && (((((((int)threadIdx.x) + 28) % 63) / 9) + ry_outer_outer) < 8)) && (1 <= ((((int)threadIdx.x) + 1) % 9))) && (((((int)threadIdx.x) + 1) % 9) < 8)) ? data[(((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 280) / 9) * 7)) + (ry_outer_outer * 7)) + ((((int)threadIdx.x) + 1) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 336)] = (((((1 <= ((((((int)threadIdx.x) + 21) % 63) / 9) + ry_outer_outer)) && (((((((int)threadIdx.x) + 21) % 63) / 9) + ry_outer_outer) < 8)) && (1 <= ((((int)threadIdx.x) + 3) % 9))) && (((((int)threadIdx.x) + 3) % 9) < 8)) ? data[(((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 336) / 9) * 7)) + (ry_outer_outer * 7)) + ((((int)threadIdx.x) + 3) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 392)] = (((((1 <= ((((((int)threadIdx.x) + 14) % 63) / 9) + ry_outer_outer)) && (((((((int)threadIdx.x) + 14) % 63) / 9) + ry_outer_outer) < 8)) && (1 <= ((((int)threadIdx.x) + 5) % 9))) && (((((int)threadIdx.x) + 5) % 9) < 8)) ? data[(((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 392) / 9) * 7)) + (ry_outer_outer * 7)) + ((((int)threadIdx.x) + 5) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 448)] = (((((1 <= (((((int)threadIdx.x) + 7) / 9) + ry_outer_outer)) && ((((((int)threadIdx.x) + 7) / 9) + ry_outer_outer) < 8)) && (1 <= ((((int)threadIdx.x) + 7) % 9))) && (((((int)threadIdx.x) + 7) % 9) < 8)) ? data[(((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 448) / 9) * 7)) + (ry_outer_outer * 7)) + ((((int)threadIdx.x) + 7) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 504)] = (((((1 <= ((((int)threadIdx.x) / 9) + ry_outer_outer)) && (((((int)threadIdx.x) / 9) + ry_outer_outer) < 8)) && (1 <= (((int)threadIdx.x) % 9))) && ((((int)threadIdx.x) % 9) < 8)) ? data[(((((rc_outer_outer * 784) + ((((int)threadIdx.x) / 9) * 7)) + (ry_outer_outer * 7)) + (((int)threadIdx.x) % 9)) + 384)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 560)] = (((((1 <= ((((((int)threadIdx.x) + 56) % 63) / 9) + ry_outer_outer)) && (((((((int)threadIdx.x) + 56) % 63) / 9) + ry_outer_outer) < 8)) && (1 <= ((((int)threadIdx.x) + 2) % 9))) && (((((int)threadIdx.x) + 2) % 9) < 8)) ? data[(((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 560) / 9) * 7)) + (ry_outer_outer * 7)) + ((((int)threadIdx.x) + 2) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 616)] = (((((1 <= ((((((int)threadIdx.x) + 49) % 63) / 9) + ry_outer_outer)) && (((((((int)threadIdx.x) + 49) % 63) / 9) + ry_outer_outer) < 8)) && (1 <= ((((int)threadIdx.x) + 4) % 9))) && (((((int)threadIdx.x) + 4) % 9) < 8)) ? data[(((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 616) / 9) * 7)) + (ry_outer_outer * 7)) + ((((int)threadIdx.x) + 4) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 672)] = (((((1 <= ((((((int)threadIdx.x) + 42) % 63) / 9) + ry_outer_outer)) && (((((((int)threadIdx.x) + 42) % 63) / 9) + ry_outer_outer) < 8)) && (1 <= ((((int)threadIdx.x) + 6) % 9))) && (((((int)threadIdx.x) + 6) % 9) < 8)) ? data[(((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 672) / 9) * 7)) + (ry_outer_outer * 7)) + ((((int)threadIdx.x) + 6) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 728)] = (((((1 <= ((((((int)threadIdx.x) + 35) % 63) / 9) + ry_outer_outer)) && (((((((int)threadIdx.x) + 35) % 63) / 9) + ry_outer_outer) < 8)) && (1 <= ((((int)threadIdx.x) + 8) % 9))) && (((((int)threadIdx.x) + 8) % 9) < 8)) ? data[(((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 728) / 9) * 7)) + (ry_outer_outer * 7)) + ((((int)threadIdx.x) + 8) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 784)] = (((((1 <= ((((((int)threadIdx.x) + 28) % 63) / 9) + ry_outer_outer)) && (((((((int)threadIdx.x) + 28) % 63) / 9) + ry_outer_outer) < 8)) && (1 <= ((((int)threadIdx.x) + 1) % 9))) && (((((int)threadIdx.x) + 1) % 9) < 8)) ? data[(((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 784) / 9) * 7)) + (ry_outer_outer * 7)) + ((((int)threadIdx.x) + 1) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 840)] = (((((1 <= ((((((int)threadIdx.x) + 21) % 63) / 9) + ry_outer_outer)) && (((((((int)threadIdx.x) + 21) % 63) / 9) + ry_outer_outer) < 8)) && (1 <= ((((int)threadIdx.x) + 3) % 9))) && (((((int)threadIdx.x) + 3) % 9) < 8)) ? data[(((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 840) / 9) * 7)) + (ry_outer_outer * 7)) + ((((int)threadIdx.x) + 3) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 896)] = (((((1 <= ((((((int)threadIdx.x) + 14) % 63) / 9) + ry_outer_outer)) && (((((((int)threadIdx.x) + 14) % 63) / 9) + ry_outer_outer) < 8)) && (1 <= ((((int)threadIdx.x) + 5) % 9))) && (((((int)threadIdx.x) + 5) % 9) < 8)) ? data[(((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 896) / 9) * 7)) + (ry_outer_outer * 7)) + ((((int)threadIdx.x) + 5) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 952)] = (((((1 <= (((((int)threadIdx.x) + 7) / 9) + ry_outer_outer)) && ((((((int)threadIdx.x) + 7) / 9) + ry_outer_outer) < 8)) && (1 <= ((((int)threadIdx.x) + 7) % 9))) && (((((int)threadIdx.x) + 7) % 9) < 8)) ? data[(((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 952) / 9) * 7)) + (ry_outer_outer * 7)) + ((((int)threadIdx.x) + 7) % 9)) - 8)] : 0.000000e+00f);
+ kernel_shared[((int)threadIdx.x)] = kernel[((((((((int)blockIdx.x) * 147456) + ((((int)threadIdx.x) / 48) * 4608)) + (rc_outer_outer * 144)) + (((((int)threadIdx.x) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + (((int)threadIdx.x) % 3))];
+ kernel_shared[(((((((int)threadIdx.x) + 56) / 48) * 48) + ((((((int)threadIdx.x) + 8) % 48) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 56) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 8) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + ((((int)threadIdx.x) + 2) % 3))];
+ kernel_shared[(((((((int)threadIdx.x) + 112) / 48) * 48) + ((((((int)threadIdx.x) + 16) % 48) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 112) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 16) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + ((((int)threadIdx.x) + 1) % 3))];
+ kernel_shared[(((((((int)threadIdx.x) + 168) / 48) * 48) + ((((((int)threadIdx.x) / 3) + 8) & 15) * 3)) + (((int)threadIdx.x) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 168) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) / 3) + 8) & 15) * 9)) + (ry_outer_outer * 3)) + (((int)threadIdx.x) % 3))];
+ kernel_shared[(((((((int)threadIdx.x) + 224) / 48) * 48) + ((((((int)threadIdx.x) + 32) % 48) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 224) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 32) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + ((((int)threadIdx.x) + 2) % 3))];
+ kernel_shared[(((((((int)threadIdx.x) + 280) / 48) * 48) + ((((((int)threadIdx.x) + 40) % 48) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 280) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 40) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + ((((int)threadIdx.x) + 1) % 3))];
+ kernel_shared[(((int)threadIdx.x) + 336)] = kernel[(((((((((int)blockIdx.x) * 147456) + ((((int)threadIdx.x) / 48) * 4608)) + (rc_outer_outer * 144)) + (((((int)threadIdx.x) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + (((int)threadIdx.x) % 3)) + 32256)];
+ kernel_shared[(((((((int)threadIdx.x) + 392) / 48) * 48) + ((((((int)threadIdx.x) + 8) % 48) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 392) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 8) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + ((((int)threadIdx.x) + 2) % 3))];
+ kernel_shared[(((((((int)threadIdx.x) + 448) / 48) * 48) + ((((((int)threadIdx.x) + 16) % 48) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 448) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 16) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + ((((int)threadIdx.x) + 1) % 3))];
+ kernel_shared[(((((((int)threadIdx.x) + 504) / 48) * 48) + ((((((int)threadIdx.x) / 3) + 8) & 15) * 3)) + (((int)threadIdx.x) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 504) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) / 3) + 8) & 15) * 9)) + (ry_outer_outer * 3)) + (((int)threadIdx.x) % 3))];
+ kernel_shared[(((((((int)threadIdx.x) + 560) / 48) * 48) + ((((((int)threadIdx.x) + 32) % 48) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 560) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 32) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + ((((int)threadIdx.x) + 2) % 3))];
+ kernel_shared[(((((((int)threadIdx.x) + 616) / 48) * 48) + ((((((int)threadIdx.x) + 40) % 48) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 616) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 40) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + ((((int)threadIdx.x) + 1) % 3))];
+ kernel_shared[(((int)threadIdx.x) + 672)] = kernel[(((((((((int)blockIdx.x) * 147456) + ((((int)threadIdx.x) / 48) * 4608)) + (rc_outer_outer * 144)) + (((((int)threadIdx.x) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + (((int)threadIdx.x) % 3)) + 64512)];
+ kernel_shared[(((((((int)threadIdx.x) + 728) / 48) * 48) + ((((((int)threadIdx.x) + 8) % 48) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 728) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 8) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + ((((int)threadIdx.x) + 2) % 3))];
+ kernel_shared[(((((((int)threadIdx.x) + 784) / 48) * 48) + ((((((int)threadIdx.x) + 16) % 48) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 784) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 16) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + ((((int)threadIdx.x) + 1) % 3))];
+ kernel_shared[(((((((int)threadIdx.x) + 840) / 48) * 48) + ((((((int)threadIdx.x) / 3) + 8) & 15) * 3)) + (((int)threadIdx.x) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 840) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) / 3) + 8) & 15) * 9)) + (ry_outer_outer * 3)) + (((int)threadIdx.x) % 3))];
+ kernel_shared[(((((((int)threadIdx.x) + 896) / 48) * 48) + ((((((int)threadIdx.x) + 32) % 48) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 896) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 32) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + ((((int)threadIdx.x) + 2) % 3))];
+ kernel_shared[(((((((int)threadIdx.x) + 952) / 48) * 48) + ((((((int)threadIdx.x) + 40) % 48) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 952) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 40) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + ((((int)threadIdx.x) + 1) % 3))];
+ kernel_shared[(((int)threadIdx.x) + 1008)] = kernel[(((((((((int)blockIdx.x) * 147456) + ((((int)threadIdx.x) / 48) * 4608)) + (rc_outer_outer * 144)) + (((((int)threadIdx.x) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + (((int)threadIdx.x) % 3)) + 96768)];
+ kernel_shared[(((((((int)threadIdx.x) + 1064) / 48) * 48) + ((((((int)threadIdx.x) + 8) % 48) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 1064) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 8) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + ((((int)threadIdx.x) + 2) % 3))];
+ kernel_shared[(((((((int)threadIdx.x) + 1120) / 48) * 48) + ((((((int)threadIdx.x) + 16) % 48) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 1120) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 16) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + ((((int)threadIdx.x) + 1) % 3))];
+ kernel_shared[(((((((int)threadIdx.x) + 1176) / 48) * 48) + ((((((int)threadIdx.x) / 3) + 8) & 15) * 3)) + (((int)threadIdx.x) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 1176) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) / 3) + 8) & 15) * 9)) + (ry_outer_outer * 3)) + (((int)threadIdx.x) % 3))];
+ kernel_shared[(((((((int)threadIdx.x) + 1232) / 48) * 48) + ((((((int)threadIdx.x) + 32) % 48) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 1232) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 32) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + ((((int)threadIdx.x) + 2) % 3))];
+ kernel_shared[(((((((int)threadIdx.x) + 1288) / 48) * 48) + ((((((int)threadIdx.x) + 40) % 48) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 1288) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 40) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + ((((int)threadIdx.x) + 1) % 3))];
+ kernel_shared[(((int)threadIdx.x) + 1344)] = kernel[(((((((((int)blockIdx.x) * 147456) + ((((int)threadIdx.x) / 48) * 4608)) + (rc_outer_outer * 144)) + (((((int)threadIdx.x) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + (((int)threadIdx.x) % 3)) + 129024)];
+ kernel_shared[(((((((int)threadIdx.x) + 1400) / 48) * 48) + ((((((int)threadIdx.x) + 8) % 48) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 1400) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 8) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + ((((int)threadIdx.x) + 2) % 3))];
+ kernel_shared[(((((((int)threadIdx.x) + 1456) / 48) * 48) + ((((((int)threadIdx.x) + 16) % 48) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 1456) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 16) % 48) / 3) * 9)) + (ry_outer_outer * 3)) + ((((int)threadIdx.x) + 1) % 3))];
+ if (((int)threadIdx.x) < 24) {
+ kernel_shared[(((((((int)threadIdx.x) + 1512) / 48) * 48) + ((int)threadIdx.x)) + 24)] = kernel[(((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 1512) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((int)threadIdx.x) / 3) * 9)) + (ry_outer_outer * 3)) + (((int)threadIdx.x) % 3)) + 72)];
+ }
+ __syncthreads();
+ for (int rc_outer_inner = 0; rc_outer_inner < 2; ++rc_outer_inner) {
+ for (int rx_outer_inner = 0; rx_outer_inner < 3; ++rx_outer_inner) {
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7))] * kernel_shared[((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 3)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 72)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 3)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 81)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 3)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 3)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 3)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 3)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 3)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 6)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 6)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 144)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 6)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 153)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 6)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 162)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 6)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 6)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 6)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 9)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 9)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 9)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 9)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 225)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 9)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 234)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 9)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 243)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 9)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 12)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 12)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 12)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 12)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 12)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 12)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 306)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 12)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 15)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 324)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 15)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 333)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 15)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 342)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 15)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 351)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 15)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 360)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 15)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 369)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 15)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 18)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 387)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 18)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 396)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 18)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 405)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 18)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 414)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 18)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 423)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 18)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 432)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 18)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 21)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 450)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 21)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 459)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 21)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 468)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 21)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 477)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 21)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 486)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 21)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 495)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 21)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7))] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 48)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 48)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 48)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 48)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 48)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 48)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 48)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 51)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 72)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 51)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 81)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 51)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 51)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 51)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 51)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 51)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 54)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 54)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 144)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 54)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 153)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 54)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 162)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 54)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 54)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 54)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 57)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 57)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 57)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 57)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 225)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 57)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 234)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 57)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 243)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 57)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 60)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 60)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 60)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 60)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 60)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 60)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 306)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 60)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 63)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 324)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 63)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 333)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 63)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 342)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 63)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 351)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 63)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 360)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 63)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 369)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 63)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 66)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 387)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 66)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 396)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 66)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 405)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 66)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 414)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 66)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 423)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 66)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 432)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 66)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 69)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 450)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 69)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 459)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 69)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 468)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 69)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 477)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 69)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 486)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 69)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 495)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 69)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7))] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 96)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 96)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 96)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 96)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 96)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 96)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 96)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 99)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 72)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 99)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 81)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 99)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 99)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 99)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 99)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 99)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 102)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 102)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 144)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 102)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 153)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 102)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 162)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 102)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 102)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 102)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 105)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 105)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 105)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 105)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 225)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 105)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 234)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 105)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 243)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 105)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 108)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 108)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 108)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 108)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 108)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 108)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 306)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 108)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 111)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 324)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 111)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 333)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 111)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 342)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 111)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 351)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 111)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 360)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 111)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 369)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 111)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 114)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 387)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 114)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 396)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 114)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 405)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 114)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 414)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 114)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 423)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 114)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 432)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 114)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 117)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 450)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 117)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 459)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 117)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 468)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 117)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 477)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 117)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 486)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 117)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 495)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 117)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7))] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 144)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 144)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 144)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 144)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 144)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 144)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 144)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 147)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 72)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 147)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 81)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 147)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 147)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 147)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 147)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 147)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 150)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 150)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 144)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 150)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 153)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 150)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 162)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 150)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 150)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 150)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 153)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 153)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 153)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 153)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 225)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 153)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 234)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 153)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 243)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 153)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 156)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 156)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 156)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 156)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 156)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 156)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 306)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 156)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 159)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 324)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 159)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 333)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 159)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 342)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 159)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 351)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 159)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 360)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 159)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 369)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 159)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 162)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 387)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 162)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 396)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 162)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 405)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 162)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 414)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 162)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 423)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 162)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 432)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 162)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 165)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 450)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 165)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 459)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 165)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 468)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 165)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 477)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 165)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 486)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 165)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[((((rc_outer_inner * 504) + rx_outer_inner) + (((int)threadIdx.x) % 7)) + 495)] * kernel_shared[(((((((int)threadIdx.x) / 7) * 192) + (rc_outer_inner * 24)) + rx_outer_inner) + 165)]));
}
- __syncthreads();
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((int)threadIdx.x) % 7)] * kernel_shared[((((int)threadIdx.x) / 7) * 128)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 7)] * kernel_shared[((((int)threadIdx.x) / 7) * 128)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 14)] * kernel_shared[((((int)threadIdx.x) / 7) * 128)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 21)] * kernel_shared[((((int)threadIdx.x) / 7) * 128)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 28)] * kernel_shared[((((int)threadIdx.x) / 7) * 128)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 35)] * kernel_shared[((((int)threadIdx.x) / 7) * 128)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 42)] * kernel_shared[((((int)threadIdx.x) / 7) * 128)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 49)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 1)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 56)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 1)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 63)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 1)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 70)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 1)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 77)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 1)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 84)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 1)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 91)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 1)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 98)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 2)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 105)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 2)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 112)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 2)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 119)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 2)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 126)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 2)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 133)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 2)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 140)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 2)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 147)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 3)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 154)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 3)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 161)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 3)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 168)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 3)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 175)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 3)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 182)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 3)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 189)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 3)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 196)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 4)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 203)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 4)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 210)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 4)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 217)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 4)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 224)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 4)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 231)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 4)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 238)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 4)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 245)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 5)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 252)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 5)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 259)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 5)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 266)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 5)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 273)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 5)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 280)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 5)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 287)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 5)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 294)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 6)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 301)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 6)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 308)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 6)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 315)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 6)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 322)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 6)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 329)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 6)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 336)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 6)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 343)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 7)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 350)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 7)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 357)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 7)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 364)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 7)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 371)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 7)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 378)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 7)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 385)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 7)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 392)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 8)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 399)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 8)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 406)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 8)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 413)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 8)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 420)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 8)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 427)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 8)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 434)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 8)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 441)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 9)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 448)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 9)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 455)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 9)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 462)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 9)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 469)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 9)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 476)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 9)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 483)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 9)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 490)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 10)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 497)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 10)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 504)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 10)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 511)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 10)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 518)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 10)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 525)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 10)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 532)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 10)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 539)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 11)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 546)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 11)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 553)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 11)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 560)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 11)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 567)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 11)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 574)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 11)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 581)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 11)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 588)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 12)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 595)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 12)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 602)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 12)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 609)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 12)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 616)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 12)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 623)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 12)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 630)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 12)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 637)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 13)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 644)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 13)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 651)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 13)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 658)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 13)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 665)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 13)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 672)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 13)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 679)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 13)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 686)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 14)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 693)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 14)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 700)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 14)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 707)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 14)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 714)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 14)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 721)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 14)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 728)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 14)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 735)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 15)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 742)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 15)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 749)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 15)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 756)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 15)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 763)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 15)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 770)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 15)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 777)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 15)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 784)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 16)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 791)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 16)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 798)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 16)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 805)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 16)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 812)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 16)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 819)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 16)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 826)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 16)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 833)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 17)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 840)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 17)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 847)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 17)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 854)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 17)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 861)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 17)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 868)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 17)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 875)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 17)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 882)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 18)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 889)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 18)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 896)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 18)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 903)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 18)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 910)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 18)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 917)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 18)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 924)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 18)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 931)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 19)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 938)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 19)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 945)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 19)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 952)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 19)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 959)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 19)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 966)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 19)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 973)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 19)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 980)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 20)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 987)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 20)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 994)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 20)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1001)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 20)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1008)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 20)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1015)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 20)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1022)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 20)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1029)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 21)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1036)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 21)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1043)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 21)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1050)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 21)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1057)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 21)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1064)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 21)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1071)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 21)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1078)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 22)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1085)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 22)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1092)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 22)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1099)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 22)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1106)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 22)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1113)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 22)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1120)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 22)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1127)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 23)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1134)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 23)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1141)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 23)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1148)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 23)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1155)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 23)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1162)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 23)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1169)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 23)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1176)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 24)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1183)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 24)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1190)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 24)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1197)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 24)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1204)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 24)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1211)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 24)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1218)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 24)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1225)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 25)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1232)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 25)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1239)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 25)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1246)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 25)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1253)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 25)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1260)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 25)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1267)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 25)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1274)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 26)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1281)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 26)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1288)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 26)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1295)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 26)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1302)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 26)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1309)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 26)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1316)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 26)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1323)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 27)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1330)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 27)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1337)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 27)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1344)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 27)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1351)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 27)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1358)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 27)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1365)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 27)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1372)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 28)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1379)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 28)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1386)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 28)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1393)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 28)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1400)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 28)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1407)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 28)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1414)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 28)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1421)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 29)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1428)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 29)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1435)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 29)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1442)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 29)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1449)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 29)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1456)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 29)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1463)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 29)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1470)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 30)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1477)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 30)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1484)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 30)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1491)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 30)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1498)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 30)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1505)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 30)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1512)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 30)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1519)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 31)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1526)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 31)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1533)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 31)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1540)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 31)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1547)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 31)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1554)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 31)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1561)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 31)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1568)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 32)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1575)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 32)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1582)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 32)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1589)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 32)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1596)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 32)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1603)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 32)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1610)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 32)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1617)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 33)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1624)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 33)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1631)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 33)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1638)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 33)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1645)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 33)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1652)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 33)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1659)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 33)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1666)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 34)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1673)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 34)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1680)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 34)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1687)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 34)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1694)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 34)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1701)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 34)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1708)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 34)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1715)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 35)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1722)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 35)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1729)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 35)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1736)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 35)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1743)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 35)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1750)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 35)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1757)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 35)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1764)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 36)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1771)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 36)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1778)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 36)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1785)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 36)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1792)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 36)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1799)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 36)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1806)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 36)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1813)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 37)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1820)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 37)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1827)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 37)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1834)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 37)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1841)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 37)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1848)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 37)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1855)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 37)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1862)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 38)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1869)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 38)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1876)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 38)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1883)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 38)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1890)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 38)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1897)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 38)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1904)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 38)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1911)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 39)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1918)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 39)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1925)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 39)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1932)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 39)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1939)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 39)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1946)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 39)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1953)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 39)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1960)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 40)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1967)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 40)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1974)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 40)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1981)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 40)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1988)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 40)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 1995)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 40)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2002)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 40)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2009)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 41)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2016)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 41)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2023)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 41)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2030)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 41)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2037)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 41)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2044)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 41)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2051)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 41)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2058)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 42)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2065)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 42)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2072)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 42)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2079)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 42)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2086)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 42)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2093)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 42)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2100)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 42)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2107)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 43)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2114)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 43)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2121)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 43)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2128)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 43)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2135)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 43)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2142)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 43)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2149)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 43)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2156)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 44)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2163)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 44)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2170)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 44)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2177)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 44)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2184)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 44)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2191)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 44)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2198)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 44)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2205)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 45)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2212)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 45)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2219)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 45)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2226)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 45)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2233)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 45)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2240)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 45)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2247)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 45)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2254)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 46)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2261)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 46)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2268)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 46)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2275)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 46)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2282)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 46)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2289)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 46)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2296)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 46)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2303)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 47)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2310)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 47)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2317)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 47)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2324)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 47)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2331)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 47)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2338)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 47)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2345)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 47)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2352)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 48)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2359)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 48)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2366)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 48)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2373)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 48)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2380)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 48)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2387)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 48)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2394)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 48)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2401)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 49)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2408)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 49)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2415)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 49)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2422)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 49)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2429)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 49)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2436)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 49)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2443)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 49)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2450)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 50)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2457)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 50)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2464)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 50)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2471)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 50)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2478)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 50)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2485)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 50)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2492)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 50)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2499)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 51)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2506)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 51)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2513)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 51)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2520)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 51)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2527)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 51)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2534)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 51)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2541)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 51)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2548)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 52)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2555)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 52)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2562)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 52)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2569)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 52)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2576)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 52)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2583)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 52)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2590)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 52)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2597)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 53)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2604)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 53)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2611)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 53)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2618)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 53)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2625)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 53)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2632)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 53)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2639)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 53)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2646)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 54)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2653)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 54)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2660)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 54)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2667)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 54)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2674)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 54)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2681)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 54)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2688)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 54)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2695)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 55)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2702)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 55)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2709)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 55)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2716)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 55)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2723)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 55)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2730)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 55)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2737)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 55)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2744)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 56)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2751)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 56)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2758)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 56)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2765)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 56)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2772)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 56)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2779)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 56)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2786)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 56)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2793)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 57)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2800)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 57)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2807)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 57)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2814)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 57)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2821)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 57)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2828)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 57)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2835)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 57)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2842)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 58)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2849)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 58)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2856)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 58)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2863)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 58)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2870)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 58)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2877)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 58)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2884)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 58)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2891)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 59)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2898)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 59)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2905)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 59)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2912)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 59)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2919)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 59)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2926)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 59)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2933)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 59)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2940)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 60)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2947)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 60)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2954)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 60)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2961)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 60)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2968)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 60)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2975)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 60)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2982)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 60)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2989)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 61)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 2996)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 61)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3003)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 61)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3010)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 61)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3017)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 61)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3024)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 61)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3031)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 61)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3038)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 62)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3045)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 62)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3052)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 62)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3059)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 62)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3066)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 62)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3073)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 62)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3080)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 62)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3087)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 63)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3094)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 63)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3101)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 63)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3108)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 63)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3115)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 63)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3122)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 63)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3129)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 63)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3136)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 64)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3143)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 64)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3150)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 64)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3157)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 64)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3164)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 64)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3171)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 64)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3178)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 64)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3185)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 65)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3192)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 65)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3199)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 65)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3206)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 65)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3213)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 65)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3220)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 65)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3227)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 65)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3234)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 66)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3241)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 66)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3248)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 66)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3255)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 66)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3262)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 66)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3269)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 66)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3276)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 66)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3283)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 67)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3290)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 67)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3297)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 67)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3304)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 67)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3311)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 67)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3318)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 67)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3325)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 67)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3332)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 68)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3339)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 68)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3346)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 68)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3353)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 68)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3360)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 68)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3367)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 68)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3374)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 68)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3381)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 69)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3388)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 69)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3395)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 69)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3402)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 69)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3409)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 69)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3416)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 69)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3423)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 69)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3430)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 70)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3437)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 70)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3444)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 70)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3451)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 70)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3458)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 70)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3465)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 70)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3472)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 70)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3479)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 71)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3486)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 71)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3493)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 71)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3500)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 71)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3507)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 71)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3514)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 71)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3521)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 71)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3528)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 72)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3535)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 72)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3542)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 72)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3549)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 72)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3556)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 72)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3563)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 72)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3570)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 72)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3577)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 73)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3584)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 73)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3591)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 73)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3598)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 73)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3605)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 73)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3612)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 73)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3619)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 73)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3626)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 74)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3633)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 74)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3640)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 74)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3647)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 74)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3654)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 74)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3661)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 74)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3668)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 74)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3675)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 75)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3682)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 75)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3689)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 75)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3696)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 75)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3703)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 75)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3710)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 75)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3717)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 75)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3724)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 76)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3731)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 76)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3738)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 76)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3745)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 76)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3752)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 76)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3759)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 76)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3766)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 76)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3773)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 77)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3780)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 77)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3787)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 77)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3794)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 77)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3801)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 77)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3808)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 77)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3815)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 77)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3822)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 78)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3829)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 78)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3836)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 78)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3843)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 78)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3850)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 78)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3857)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 78)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3864)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 78)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3871)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 79)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3878)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 79)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3885)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 79)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3892)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 79)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3899)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 79)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3906)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 79)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3913)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 79)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3920)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 80)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3927)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 80)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3934)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 80)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3941)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 80)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3948)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 80)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3955)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 80)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3962)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 80)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3969)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 81)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3976)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 81)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3983)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 81)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3990)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 81)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 3997)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 81)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4004)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 81)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4011)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 81)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4018)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 82)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4025)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 82)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4032)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 82)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4039)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 82)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4046)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 82)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4053)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 82)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4060)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 82)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4067)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 83)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4074)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 83)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4081)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 83)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4088)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 83)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4095)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 83)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4102)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 83)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4109)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 83)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4116)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 84)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4123)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 84)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4130)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 84)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4137)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 84)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4144)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 84)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4151)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 84)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4158)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 84)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4165)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 85)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4172)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 85)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4179)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 85)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4186)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 85)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4193)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 85)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4200)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 85)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4207)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 85)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4214)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 86)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4221)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 86)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4228)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 86)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4235)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 86)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4242)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 86)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4249)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 86)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4256)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 86)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4263)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 87)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4270)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 87)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4277)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 87)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4284)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 87)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4291)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 87)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4298)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 87)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4305)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 87)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4312)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 88)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4319)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 88)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4326)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 88)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4333)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 88)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4340)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 88)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4347)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 88)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4354)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 88)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4361)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 89)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4368)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 89)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4375)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 89)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4382)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 89)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4389)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 89)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4396)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 89)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4403)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 89)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4410)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 90)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4417)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 90)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4424)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 90)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4431)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 90)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4438)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 90)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4445)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 90)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4452)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 90)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4459)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 91)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4466)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 91)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4473)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 91)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4480)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 91)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4487)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 91)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4494)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 91)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4501)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 91)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4508)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 92)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4515)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 92)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4522)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 92)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4529)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 92)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4536)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 92)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4543)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 92)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4550)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 92)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4557)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 93)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4564)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 93)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4571)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 93)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4578)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 93)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4585)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 93)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4592)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 93)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4599)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 93)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4606)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 94)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4613)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 94)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4620)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 94)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4627)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 94)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4634)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 94)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4641)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 94)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4648)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 94)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4655)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 95)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4662)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 95)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4669)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 95)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4676)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 95)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4683)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 95)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4690)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 95)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4697)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 95)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4704)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 96)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4711)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 96)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4718)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 96)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4725)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 96)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4732)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 96)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4739)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 96)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4746)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 96)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4753)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 97)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4760)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 97)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4767)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 97)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4774)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 97)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4781)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 97)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4788)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 97)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4795)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 97)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4802)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 98)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4809)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 98)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4816)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 98)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4823)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 98)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4830)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 98)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4837)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 98)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4844)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 98)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4851)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 99)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4858)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 99)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4865)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 99)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4872)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 99)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4879)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 99)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4886)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 99)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4893)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 99)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4900)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 100)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4907)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 100)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4914)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 100)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4921)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 100)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4928)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 100)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4935)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 100)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4942)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 100)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4949)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 101)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4956)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 101)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4963)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 101)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4970)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 101)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4977)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 101)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4984)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 101)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4991)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 101)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 4998)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 102)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5005)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 102)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5012)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 102)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5019)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 102)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5026)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 102)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5033)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 102)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5040)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 102)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5047)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 103)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5054)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 103)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5061)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 103)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5068)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 103)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5075)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 103)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5082)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 103)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5089)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 103)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5096)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 104)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5103)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 104)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5110)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 104)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5117)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 104)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5124)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 104)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5131)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 104)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5138)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 104)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5145)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 105)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5152)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 105)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5159)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 105)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5166)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 105)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5173)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 105)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5180)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 105)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5187)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 105)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5194)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 106)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5201)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 106)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5208)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 106)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5215)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 106)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5222)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 106)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5229)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 106)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5236)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 106)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5243)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 107)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5250)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 107)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5257)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 107)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5264)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 107)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5271)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 107)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5278)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 107)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5285)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 107)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5292)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 108)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5299)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 108)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5306)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 108)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5313)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 108)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5320)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 108)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5327)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 108)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5334)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 108)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5341)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 109)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5348)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 109)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5355)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 109)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5362)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 109)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5369)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 109)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5376)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 109)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5383)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 109)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5390)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 110)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5397)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 110)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5404)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 110)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5411)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 110)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5418)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 110)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5425)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 110)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5432)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 110)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5439)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 111)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5446)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 111)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5453)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 111)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5460)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 111)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5467)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 111)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5474)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 111)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5481)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 111)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5488)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 112)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5495)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 112)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5502)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 112)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5509)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 112)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5516)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 112)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5523)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 112)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5530)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 112)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5537)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 113)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5544)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 113)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5551)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 113)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5558)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 113)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5565)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 113)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5572)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 113)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5579)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 113)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5586)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 114)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5593)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 114)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5600)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 114)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5607)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 114)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5614)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 114)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5621)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 114)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5628)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 114)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5635)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 115)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5642)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 115)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5649)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 115)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5656)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 115)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5663)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 115)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5670)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 115)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5677)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 115)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5684)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 116)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5691)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 116)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5698)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 116)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5705)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 116)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5712)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 116)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5719)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 116)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5726)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 116)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5733)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 117)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5740)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 117)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5747)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 117)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5754)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 117)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5761)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 117)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5768)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 117)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5775)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 117)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5782)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 118)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5789)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 118)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5796)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 118)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5803)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 118)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5810)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 118)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5817)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 118)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5824)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 118)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5831)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 119)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5838)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 119)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5845)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 119)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5852)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 119)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5859)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 119)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5866)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 119)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5873)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 119)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5880)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 120)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5887)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 120)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5894)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 120)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5901)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 120)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5908)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 120)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5915)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 120)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5922)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 120)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5929)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 121)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5936)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 121)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5943)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 121)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5950)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 121)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5957)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 121)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5964)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 121)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5971)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 121)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5978)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 122)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5985)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 122)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5992)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 122)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 5999)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 122)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6006)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 122)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6013)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 122)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6020)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 122)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6027)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 123)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6034)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 123)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6041)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 123)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6048)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 123)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6055)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 123)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6062)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 123)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6069)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 123)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6076)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 124)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6083)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 124)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6090)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 124)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6097)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 124)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6104)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 124)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6111)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 124)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6118)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 124)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6125)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 125)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6132)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 125)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6139)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 125)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6146)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 125)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6153)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 125)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6160)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 125)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6167)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 125)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6174)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 126)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6181)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 126)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6188)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 126)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6195)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 126)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6202)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 126)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6209)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 126)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6216)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 126)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6223)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 127)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6230)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 127)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6237)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 127)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6244)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 127)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6251)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 127)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6258)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 127)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((int)threadIdx.x) % 7) + 6265)] * kernel_shared[(((((int)threadIdx.x) / 7) * 128) + 127)]));
}
}
}
- compute[(((((int)blockIdx.x) * 784) + ((((int)threadIdx.x) / 7) * 49)) + (((int)threadIdx.x) % 7))] = max((conv2d_nchw[0] + bias[((((int)blockIdx.x) * 16) + (((int)threadIdx.x) / 7))]), 0.000000e+00f);
- compute[((((((int)blockIdx.x) * 784) + ((((int)threadIdx.x) / 7) * 49)) + (((int)threadIdx.x) % 7)) + 7)] = max((conv2d_nchw[1] + bias[((((int)blockIdx.x) * 16) + (((int)threadIdx.x) / 7))]), 0.000000e+00f);
- compute[((((((int)blockIdx.x) * 784) + ((((int)threadIdx.x) / 7) * 49)) + (((int)threadIdx.x) % 7)) + 14)] = max((conv2d_nchw[2] + bias[((((int)blockIdx.x) * 16) + (((int)threadIdx.x) / 7))]), 0.000000e+00f);
- compute[((((((int)blockIdx.x) * 784) + ((((int)threadIdx.x) / 7) * 49)) + (((int)threadIdx.x) % 7)) + 21)] = max((conv2d_nchw[3] + bias[((((int)blockIdx.x) * 16) + (((int)threadIdx.x) / 7))]), 0.000000e+00f);
- compute[((((((int)blockIdx.x) * 784) + ((((int)threadIdx.x) / 7) * 49)) + (((int)threadIdx.x) % 7)) + 28)] = max((conv2d_nchw[4] + bias[((((int)blockIdx.x) * 16) + (((int)threadIdx.x) / 7))]), 0.000000e+00f);
- compute[((((((int)blockIdx.x) * 784) + ((((int)threadIdx.x) / 7) * 49)) + (((int)threadIdx.x) % 7)) + 35)] = max((conv2d_nchw[5] + bias[((((int)blockIdx.x) * 16) + (((int)threadIdx.x) / 7))]), 0.000000e+00f);
- compute[((((((int)blockIdx.x) * 784) + ((((int)threadIdx.x) / 7) * 49)) + (((int)threadIdx.x) % 7)) + 42)] = max((conv2d_nchw[6] + bias[((((int)blockIdx.x) * 16) + (((int)threadIdx.x) / 7))]), 0.000000e+00f);
+ for (int i1_inner = 0; i1_inner < 4; ++i1_inner) {
+ for (int i2_inner = 0; i2_inner < 7; ++i2_inner) {
+ compute[(((((((int)blockIdx.x) * 1568) + ((((int)threadIdx.x) / 7) * 196)) + (i1_inner * 49)) + (i2_inner * 7)) + (((int)threadIdx.x) % 7))] = max((conv2d_nchw[((i1_inner * 7) + i2_inner)] + bias[(((((int)blockIdx.x) * 32) + ((((int)threadIdx.x) / 7) * 4)) + i1_inner)]), 0.000000e+00f);
+ }
+ }
}
</pre></div>
</div>
@@ -2861,7 +1372,7 @@ 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 30.519 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 6 minutes 17.263 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 446006bb4b..2d395083d6 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)
- 8.0970 8.0959 8.1005 8.0947 0.0025
+ 8.1035 8.1016 8.1151 8.0939 0.0087
</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 9.444 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 12.706 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 32b4e33e45..c6f3643895 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)
- 719.8005 719.7170 720.9380 718.7466 0.8966
+ 759.3156 758.9744 760.6804 758.2919 1.0045
</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 40.103 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 45.435 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 570f370ba8..3e454d6ff6 100644
--- a/docs/how_to/tune_with_autoscheduler/tune_sparse_x86.html
+++ b/docs/how_to/tune_with_autoscheduler/tune_sparse_x86.html
@@ -637,53 +637,26 @@ 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": T.bool(True), "global_symbol": "main", "tir.noalias": T.bool(True)})
- 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 i_outer_inner in range(64):
- cse_var_2: T.int32 = i_outer_inner * 8
- cse_var_1: T.int32 = i0_outer_i1_outer_fused % 64 // 2
- compute_2[cse_var_2] = T.float32(0)
- compute_2[cse_var_2 + 1] = T.float32(0)
- compute_2[cse_var_2 + 2] = T.float32(0)
- compute_2[cse_var_2 + 3] = T.float32(0)
- compute_2[cse_var_2 + 4] = T.float32(0)
- compute_2[cse_var_2 + 5] = T.float32(0)
- compute_2[cse_var_2 + 6] = T.float32(0)
- compute_2[cse_var_2 + 7] = T.float32(0)
- for elem_idx in range(placeholder_5[cse_var_1 + 1] - placeholder_5[cse_var_1]):
+ 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 nb_j_inner in range(2):
+ for i_inner_init, j_init in T.grid(64, 16):
+ compute_2[i_inner_init * 32 + nb_j_inner * 16 + j_init] = T.float32(0)
+ for elem_idx, i_inner, j in T.grid(T.Let(placeholder_5[cse_var_1 + 1] - placeholder_5[cse_var_1], where={cse_var_1: i0_outer_i1_outer_fused % 16 * 2 + nb_j_inner}), 64, 16):
+ cse_var_1 = T.int32()
placeholder_5 = T.Buffer((33,), "int32", data=placeholder_3.data)
+ cse_var_3: T.int32 = i0_outer_i1_outer_fused % 16 * 2 + nb_j_inner
+ cse_var_2: T.int32 = i_inner * 32 + nb_j_inner * 16 + j
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[cse_var_1 + 1] - placeholder_5[cse_var_1]):
- compute_2[cse_var_2] = compute_2[cse_var_2] + placeholder_6[placeholder_5[cse_var_1] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 2 * 8] * T.max(placeholder_7[i0_outer_i1_outer_fused // 64 * 16384 + i_outer_inner * 256 + placeholder_8[placeholder_5[cse_var_1] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[cse_var_1 + 1] - placeholder_5[cse_var_1]):
- cse_var_3: T.int32 = cse_var_2 + 1
- compute_2[cse_var_3] = compute_2[cse_var_3] + placeholder_6[placeholder_5[cse_var_1] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 2 * 8 + 1] * T.max(placeholder_7[i0_outer_i1_outer_fused // 64 * 16384 + i_outer_inner * 256 + placeholder_8[placeholder_5[cse_var_1] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[cse_var_1 + 1] - placeholder_5[cse_var_1]):
- cse_var_4: T.int32 = cse_var_2 + 2
- compute_2[cse_var_4] = compute_2[cse_var_4] + placeholder_6[placeholder_5[cse_var_1] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 2 * 8 + 2] * T.max(placeholder_7[i0_outer_i1_outer_fused // 64 * 16384 + i_outer_inner * 256 + placeholder_8[placeholder_5[cse_var_1] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[cse_var_1 + 1] - placeholder_5[cse_var_1]):
- cse_var_5: T.int32 = cse_var_2 + 3
- compute_2[cse_var_5] = compute_2[cse_var_5] + placeholder_6[placeholder_5[cse_var_1] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 2 * 8 + 3] * T.max(placeholder_7[i0_outer_i1_outer_fused // 64 * 16384 + i_outer_inner * 256 + placeholder_8[placeholder_5[cse_var_1] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[cse_var_1 + 1] - placeholder_5[cse_var_1]):
- cse_var_6: T.int32 = cse_var_2 + 4
- compute_2[cse_var_6] = compute_2[cse_var_6] + placeholder_6[placeholder_5[cse_var_1] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 2 * 8 + 4] * T.max(placeholder_7[i0_outer_i1_outer_fused // 64 * 16384 + i_outer_inner * 256 + placeholder_8[placeholder_5[cse_var_1] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[cse_var_1 + 1] - placeholder_5[cse_var_1]):
- cse_var_7: T.int32 = cse_var_2 + 5
- compute_2[cse_var_7] = compute_2[cse_var_7] + placeholder_6[placeholder_5[cse_var_1] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 2 * 8 + 5] * T.max(placeholder_7[i0_outer_i1_outer_fused // 64 * 16384 + i_outer_inner * 256 + placeholder_8[placeholder_5[cse_var_1] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[cse_var_1 + 1] - placeholder_5[cse_var_1]):
- cse_var_8: T.int32 = cse_var_2 + 6
- compute_2[cse_var_8] = compute_2[cse_var_8] + placeholder_6[placeholder_5[cse_var_1] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 2 * 8 + 6] * T.max(placeholder_7[i0_outer_i1_outer_fused // 64 * 16384 + i_outer_inner * 256 + placeholder_8[placeholder_5[cse_var_1] + elem_idx]], T.float32(0))
- if T.likely(elem_idx < placeholder_5[cse_var_1 + 1] - placeholder_5[cse_var_1]):
- cse_var_9: T.int32 = cse_var_2 + 7
- compute_2[cse_var_9] = compute_2[cse_var_9] + placeholder_6[placeholder_5[cse_var_1] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 2 * 8 + 7] * T.max(placeholder_7[i0_outer_i1_outer_fused // 64 * 16384 + i_outer_inner * 256 + placeholder_8[placeholder_5[cse_var_1] + elem_idx]], T.float32(0))
+ compute_2[cse_var_2] = compute_2[cse_var_2] + placeholder_6[placeholder_5[cse_var_3] * 16 + elem_idx * 16 + j] * T.max(placeholder_7[i0_outer_i1_outer_fused // 16 * 16384 + i_inner * 256 + placeholder_8[placeholder_5[cse_var_3] + elem_idx]], T.float32(0))
for i0_inner in range(64):
- cse_var_10: T.int32 = i0_outer_i1_outer_fused // 64 * 32768 + i0_inner * 512 + i0_outer_i1_outer_fused % 64 * 8
+ cse_var_4: T.int32 = i0_outer_i1_outer_fused // 16 * 32768 + 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_10:cse_var_10 + 8] = T.max(compute_2[i0_inner * 8:i0_inner * 8 + 8] + placeholder_5[cse_var_10:cse_var_10 + 8], T.Broadcast(T.float32(0), 8))
+ compute_3[cse_var_4:cse_var_4 + 32] = T.max(compute_2[i0_inner * 32:i0_inner * 32 + 32] + placeholder_5[cse_var_4:cse_var_4 + 32], T.Broadcast(T.float32(0), 32))
</pre></div>
</div>
</div>
@@ -717,7 +690,7 @@ 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: 2.490 ms
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Execution time of this operator: 1.800 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 ed6711957a..6bcd24826b 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">ΒΆ</a></h1>
-<p><strong>00:39.175</strong> total execution time for <strong>how_to_tune_with_autotvm</strong> files:</p>
+<p><strong>00:45.955</strong> total execution time for <strong>how_to_tune_with_autotvm</strong> files:</p>
<table class="docutils align-default">
<colgroup>
<col style="width: 84%" />
@@ -354,22 +354,22 @@
</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>00:39.137</p></td>
+<td><p>00:45.916</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>00:00.023</p></td>
+<td><p>00:00.024</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>
<td><p>00:00.005</p></td>
<td><p>0.0 MB</p></td>
</tr>
-<tr class="row-even"><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>
+<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>0.0 MB</p></td>
</tr>
-<tr class="row-odd"><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>
+<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>0.0 MB</p></td>
</tr>
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 29707b6b5a..d52afe71cf 100644
--- a/docs/how_to/tune_with_autotvm/tune_conv2d_cuda.html
+++ b/docs/how_to/tune_with_autotvm/tune_conv2d_cuda.html
@@ -573,8 +573,8 @@ for this template</p>
waiting for device...
device available
Get devices for measurement successfully!
-No: 1 GFLOPS: 19.81/19.81 result: MeasureResult(costs=(0.011685259727272728,), error_no=MeasureErrorNo.NO_ERROR, all_cost=2.231394052505493, timestamp=1681665130.3004358) [('tile_f', [-1, 32, 2, 2]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 2, 2]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,5847229
-No: 2 GFLOPS: 0.00/19.81 result: Traceback (most recent call last):
+No: 1 GFLOPS: 108.31/108.31 result: MeasureResult(costs=(0.0021374299215686276,), error_no=MeasureErrorNo.NO_ERROR, all_cost=4.775891065597534, timestamp=1681675917.6439672) [('tile_f', [-1, 4, 1, 1]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 1, 2]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,10103502
+No: 2 GFLOPS: 0.00/108.31 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
@@ -696,8 +696,9 @@ 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, 1, 4, 128]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 1, 64]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,4998835
-No: 3 GFLOPS: 0.00/19.81 result: Traceback (most recent call last):
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 1, 256, 2]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 32, 1]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,4664539
+No: 3 GFLOPS: 81.83/108.31 result: MeasureResult(costs=(0.0028289757358490567,), error_no=MeasureErrorNo.NO_ERROR, all_cost=5.083340644836426, timestamp=1681675920.383326) [('tile_f', [-1, 2, 4, 16]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 8, 1]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,8917656
+No: 4 GFLOPS: 0.00/108.31 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
@@ -819,8 +820,8 @@ 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, 4, 1]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 128, 2]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,7223285
-No: 4 GFLOPS: 0.00/19.81 result: Traceback (most recent call last):
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 2, 8]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 32, 8]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,4762264
+No: 5 GFLOPS: 0.00/108.31 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
@@ -942,8 +943,8 @@ 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, 2, 2]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 1, 512]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,9290890
-No: 5 GFLOPS: 0.00/19.81 result: Traceback (most recent call last):
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 4, 1, 128]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 64, 8]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,7280012
+No: 6 GFLOPS: 0.00/108.31 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
@@ -1065,377 +1066,8 @@ 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, 8, 1, 64]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 32, 8]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,502243
-No: 6 GFLOPS: 0.00/19.81 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:401
- 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:387
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:282
- 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:101
- 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:401
- 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:387
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:282
- 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:101
- 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, 1, 512]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 4, 8]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,492139
-No: 7 GFLOPS: 0.00/19.81 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:401
- 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:387
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:282
- 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:101
- 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:401
- 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:387
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:282
- 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:101
- 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, 8, 16, 4]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 8, 64]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,5785029
-No: 8 GFLOPS: 0.00/19.81 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:401
- 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:387
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:282
- 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:101
- 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:401
- 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:387
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:282
- 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:101
- 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, 4, 4, 16]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 32, 16]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,1299277
-No: 9 GFLOPS: 0.00/19.81 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, 7, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 32, 4]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,1635393
+No: 7 GFLOPS: 0.00/108.31 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
@@ -1523,7 +1155,7 @@ tvm._ffi.base.TVMError: Traceback (most recent call last):
15: _PyEval_EvalFrameDefault
14: 0x0000000000537c30
13: _PyObject_FastCallKeywords
- 12: 0x00007f4651985fa2
+ 12: 0x00007f211f670fa2
11: _ctypes_callproc
10: ffi_call
9: ffi_call_unix64
@@ -1587,132 +1219,8 @@ Traceback (most recent call last):
22: _PyEval_EvalFrameDefault
21: _PyFunction_FastCallKeywords
20: _PyEval_EvalFrameDefault
- 19: _PyFunction_FastCall [('tile_f', [-1, 4, 1, 16]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 1, 16]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,8447946
-No: 10 GFLOPS: 43.92/43.92 result: MeasureResult(costs=(0.005271183210526315,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.4004976749420166, timestamp=1681665144.7520602) [('tile_f', [-1, 2, 4, 1]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 2, 1]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,3294740
-No: 11 GFLOPS: 0.00/43.92 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
-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:401
- 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:387
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:282
- 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:101
- 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
... 3658 lines suppressed ...