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 2022/07/11 10:43:29 UTC
[tvm-site] branch asf-site updated: deploying docs (apache/tvm@fbb7b5d1a0d82acb1f581dd2ec362b4dcad2638e)
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 2838c2cfb deploying docs (apache/tvm@fbb7b5d1a0d82acb1f581dd2ec362b4dcad2638e)
2838c2cfb is described below
commit 2838c2cfb1c426c483979dc37b86264250fc1db9
Author: tvm-bot <95...@users.noreply.github.com>
AuthorDate: Mon Jul 11 10:43:23 2022 +0000
deploying docs (apache/tvm@fbb7b5d1a0d82acb1f581dd2ec362b4dcad2638e)
---
.../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_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 | 18 +-
.../extend_tvm/bring_your_own_datatypes.rst.txt | 4 +-
.../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 | 2073 +++++++++++++-------
.../tune_network_cuda.rst.txt | 2 +-
.../tune_network_x86.rst.txt | 4 +-
.../tune_sparse_x86.rst.txt | 1308 +++++++++++-
.../tune_with_autotvm/sg_execution_times.rst.txt | 4 +-
.../tune_with_autotvm/tune_conv2d_cuda.rst.txt | 34 +-
.../work_with_microtvm/micro_autotune.rst.txt | 16 +-
.../how_to/work_with_microtvm/micro_train.rst.txt | 16 +-
.../work_with_microtvm/sg_execution_times.rst.txt | 8 +-
.../work_with_relay/sg_execution_times.rst.txt | 6 +-
.../how_to/work_with_schedules/intrin_math.rst.txt | 2 +-
.../work_with_schedules/sg_execution_times.rst.txt | 16 +-
.../how_to/work_with_schedules/tensorize.rst.txt | 2 +-
docs/_sources/install/from_source.rst.txt | 4 +-
.../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 | 4 +-
.../tutorial/auto_scheduler_matmul_x86.rst.txt | 2 +-
docs/_sources/tutorial/autotvm_matmul_x86.rst.txt | 20 +-
docs/_sources/tutorial/autotvm_relay_x86.rst.txt | 54 +-
.../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 | 46 +-
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 | 14 +-
docs/how_to/compile_models/from_pytorch.html | 10 +-
docs/how_to/compile_models/from_tensorflow.html | 2 +-
docs/how_to/compile_models/sg_execution_times.html | 22 +-
.../deploy_models/deploy_model_on_android.html | 2 +-
.../deploy_object_detection_pytorch.html | 21 +-
docs/how_to/deploy_models/deploy_prequantized.html | 9 +-
.../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 | 18 +-
.../extend_tvm/bring_your_own_datatypes.html | 4 +-
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 | 2073 +++++++++++++-------
.../tune_with_autoscheduler/tune_network_cuda.html | 2 +-
.../tune_with_autoscheduler/tune_network_x86.html | 4 +-
.../tune_with_autoscheduler/tune_sparse_x86.html | 1308 +++++++++++-
.../tune_with_autotvm/sg_execution_times.html | 4 +-
.../how_to/tune_with_autotvm/tune_conv2d_cuda.html | 34 +-
docs/how_to/work_with_microtvm/micro_autotune.html | 16 +-
docs/how_to/work_with_microtvm/micro_train.html | 16 +-
.../work_with_microtvm/sg_execution_times.html | 8 +-
.../how_to/work_with_relay/sg_execution_times.html | 6 +-
docs/how_to/work_with_schedules/intrin_math.html | 2 +-
.../work_with_schedules/sg_execution_times.html | 16 +-
docs/how_to/work_with_schedules/tensorize.html | 2 +-
docs/install/from_source.html | 4 +-
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 +-
.../api/typedoc/classes/graphexecutor.html | 16 +-
docs/reference/api/typedoc/classes/instance.html | 40 +-
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 +-
docs/reference/api/typedoc/classes/scalar.html | 6 +-
.../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 | 112 +-
.../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 | 4 +-
docs/tutorial/auto_scheduler_matmul_x86.html | 2 +-
docs/tutorial/autotvm_matmul_x86.html | 20 +-
docs/tutorial/autotvm_relay_x86.html | 258 +--
docs/tutorial/cross_compilation_and_rpc.html | 2 +-
docs/tutorial/intro_topi.html | 2 +-
docs/tutorial/sg_execution_times.html | 26 +-
docs/tutorial/tensor_expr_get_started.html | 46 +-
123 files changed, 6202 insertions(+), 2166 deletions(-)
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 cef4327e7..fdfc27bfa 100644
--- a/docs/_sources/how_to/compile_models/from_darknet.rst.txt
+++ b/docs/_sources/how_to/compile_models/from_darknet.rst.txt
@@ -317,7 +317,7 @@ The process is no different from other examples.
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 1.137 seconds)
+ **Total running time of the script:** ( 1 minutes 2.100 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 c8dd2250e..408b75ced 100644
--- a/docs/_sources/how_to/compile_models/from_mxnet.rst.txt
+++ b/docs/_sources/how_to/compile_models/from_mxnet.rst.txt
@@ -115,7 +115,7 @@ In this section, we download a pretrained imagenet model and classify an image.
.. code-block:: none
- Downloading /workspace/.mxnet/models/resnet18_v1-a0666292.zip60b9a966-12eb-4861-83d9-5634ac1241ee from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/resnet18_v1-a0666292.zip...
+ Downloading /workspace/.mxnet/models/resnet18_v1-a0666292.zip1cf8aeaa-377a-4b9f-8caf-ece2bf0475c4 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 7c27720c9..57adfd1ed 100644
--- a/docs/_sources/how_to/compile_models/from_oneflow.rst.txt
+++ b/docs/_sources/how_to/compile_models/from_oneflow.rst.txt
@@ -113,7 +113,7 @@ 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, 60.1MB/s]
35%|###4 | 14.3M/41.5M [00:00<00:00, 55.8MB/s]
47%|####7 | 19.6M/41.5M [00:00<00:00, 55.0MB/s]
60%|#####9 | 24.9M/41.5M [00:00<00:00, 35.0MB/s]
77%|#######7 | 32.1M/41.5M [00:00<00:00, 44.5MB/s]
90%|########9 | 37.2M/41.5M [00:00<00:00, 44.3MB/s]
100%|##########| 41.5M/41.5M [00:00<00:00, 47.3MB/s]
+
0%| | 0.00/41.5M [00:00<?, ?B/s]
15%|#5 | 6.33M/41.5M [00:00<00:01, 32.3MB/s]
24%|##4 | 10.1M/41.5M [00:00<00:00, 35.2MB/s]
39%|###8 | 16.0M/41.5M [00:00<00:00, 42.5MB/s]
58%|#####7 | 24.0M/41.5M [00:00<00:00, 41.9MB/s]
77%|#######7 | 32.0M/41.5M [00:00<00:00, 49.0MB/s]
96%|#########6| 40.0M/41.5M [00:00<00:00, 52.6MB/s]
100%|##########| 41.5M/41.5M [00:00<00:00, 48.1MB/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 ed2521243..ab17717d5 100644
--- a/docs/_sources/how_to/compile_models/from_pytorch.rst.txt
+++ b/docs/_sources/how_to/compile_models/from_pytorch.rst.txt
@@ -94,7 +94,7 @@ Load a pretrained PyTorch model
.. code-block:: none
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]
2%|1 | 760k/44.7M [00:00<00:05, 7.74MB/s]
4%|4 | 1.98M/44.7M [00:00<00:04, 10.8MB/s]
9%|9 | 4.03M/44.7M [00:00<00:02, 15.7MB/s]
17%|#6 | 7.42M/44.7M [00:00<00:01, 23.5MB/s]
29%|##9 | 13.1M/44.7M [00:00<00:00, 36.3MB/s]
51%|#####1 | 22.8M/44.7M [00:00<00:00, 58.6MB/s]
86%|########5 | 38.3M/44.7M [00:00<00:00, 92.5MB/s]
100%|##########| 44.7M/44.7M [00:00<00:00, 63.2MB/s]
+
0%| | 0.00/44.7M [00:00<?, ?B/s]
42%|####2 | 19.0M/44.7M [00:00<00:00, 199MB/s]
100%|##########| 44.7M/44.7M [00:00<00:00, 235MB/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 e44e28d0d..497383ec0 100644
--- a/docs/_sources/how_to/compile_models/from_tensorflow.rst.txt
+++ b/docs/_sources/how_to/compile_models/from_tensorflow.rst.txt
@@ -423,7 +423,7 @@ Run the corresponding model on tensorflow
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 1.537 seconds)
+ **Total running time of the script:** ( 1 minutes 2.920 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 65807820e..400a9b802 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
=================
-**05:13.044** total execution time for **how_to_compile_models** files:
+**05:17.547** total execution time for **how_to_compile_models** files:
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_tensorflow.py` (``from_tensorflow.py``) | 01:01.537 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_tensorflow.py` (``from_tensorflow.py``) | 01:02.920 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_darknet.py` (``from_darknet.py``) | 01:01.137 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_darknet.py` (``from_darknet.py``) | 01:02.100 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_paddle.py` (``from_paddle.py``) | 00:39.852 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_paddle.py` (``from_paddle.py``) | 00:40.181 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_keras.py` (``from_keras.py``) | 00:31.538 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_keras.py` (``from_keras.py``) | 00:32.825 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_oneflow.py` (``from_oneflow.py``) | 00:26.001 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_oneflow.py` (``from_oneflow.py``) | 00:26.237 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_tflite.py` (``from_tflite.py``) | 00:25.569 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_tflite.py` (``from_tflite.py``) | 00:25.090 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_mxnet.py` (``from_mxnet.py``) | 00:23.003 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_mxnet.py` (``from_mxnet.py``) | 00:23.089 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_coreml.py` (``from_coreml.py``) | 00:22.003 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_coreml.py` (``from_coreml.py``) | 00:22.144 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_pytorch.py` (``from_pytorch.py``) | 00:20.045 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_pytorch.py` (``from_pytorch.py``) | 00:20.073 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_onnx.py` (``from_onnx.py``) | 00:02.358 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_onnx.py` (``from_onnx.py``) | 00:02.890 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
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 2629be87e..f07261e21 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
@@ -441,7 +441,7 @@ Execute on TVM
Evaluate inference time cost...
Execution time summary:
mean (ms) median (ms) max (ms) min (ms) std (ms)
- 16.0296 16.1118 16.3556 15.5997 0.2841
+ 16.2902 16.2610 16.6797 16.0264 0.2077
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 afe5f29c0..4ee670030 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
@@ -123,7 +123,7 @@ Load pre-trained maskrcnn from torchvision and do tracing
.. code-block:: none
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]
6%|6 | 10.5M/170M [00:00<00:01, 109MB/s]
18%|#7 | 29.8M/170M [00:00<00:00, 164MB/s]
28%|##8 | 48.3M/170M [00:00<00:00, 173MB/s]
40%|###9 | 67.8M/170M [00:00<00:00, 186MB/s]
51%|#####1 | 87.4M/170M [00:00<00:00, 193MB/s]
63%|######3 | 107M/170M [00:00<00:00, 197MB/s]
75%|#######4 | 127M/170M [00:00<00:00, 200MB/s]
86%|########6 | 146M/170M [00:00<00:00, 201MB/s]
97%|#########7| 165M/170M [00:00<00:00, 200MB/s]
100%|##########| 170M/170M [00:00<00:00, 191MB/s]
+
0%| | 0.00/170M [00:00<?, ?B/s]
3%|2 | 4.45M/170M [00:00<00:03, 46.6MB/s]
5%|5 | 9.01M/170M [00:00<00:03, 47.3MB/s]
21%|## | 35.2M/170M [00:00<00:00, 151MB/s]
37%|###6 | 62.2M/170M [00:00<00:00, 203MB/s]
51%|#####1 | 87.5M/170M [00:00<00:00, 224MB/s]
67%|######7 | 114M/170M [00:00<00:00, 244MB/s]
83%|########2 | 141M/170M [00:00<00:00, 255MB/s]
98%|#########7| 166M/170M [00:00<00:00, 259MB/s]
100%|##########| 170M/170M [00:00<00:00, 218MB/s]
/usr/local/lib/python3.7/dist-packages/torch/nn/functional.py:3878: 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)
/usr/local/lib/python3.7/dist-packages/torchvision/models/detection/anchor_utils.py:127: 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').
@@ -292,7 +292,7 @@ Get boxes with score larger than 0.9
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 2 minutes 55.692 seconds)
+ **Total running time of the script:** ( 2 minutes 57.631 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 b81be0ba5..3adde9f17 100644
--- a/docs/_sources/how_to/deploy_models/deploy_prequantized.rst.txt
+++ b/docs/_sources/how_to/deploy_models/deploy_prequantized.rst.txt
@@ -232,7 +232,7 @@ training. Other models require a full post training calibration.
.. code-block:: none
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]
27%|##7 | 3.69M/13.6M [00:00<00:00, 38.5MB/s]
60%|###### | 8.19M/13.6M [00:00<00:00, 43.4MB/s]
91%|######### | 12.3M/13.6M [00:00<00:00, 34.7MB/s]
100%|##########| 13.6M/13.6M [00:00<00:00, 34.2MB/s]
+
0%| | 0.00/13.6M [00:00<?, ?B/s]
100%|##########| 13.6M/13.6M [00:00<00:00, 147MB/s]
@@ -412,7 +412,7 @@ Here we give an example of how to measure performance of TVM compiled models.
Execution time summary:
mean (ms) median (ms) max (ms) min (ms) std (ms)
- 90.5425 90.3991 97.1210 90.0839 0.8348
+ 90.4966 90.4039 94.7650 90.2492 0.5064
@@ -461,7 +461,7 @@ TODO
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 8.038 seconds)
+ **Total running time of the script:** ( 1 minutes 7.701 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 08f891d00..69647a4e4 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
@@ -439,7 +439,7 @@ 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)
- 120.0127 119.9937 122.4806 119.2379 0.4334
+ 119.7802 119.7182 121.4246 119.0629 0.4287
@@ -476,7 +476,7 @@ 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 1.639 seconds)
+ **Total running time of the script:** ( 1 minutes 59.854 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 0d23cdad9..65c077095 100644
--- a/docs/_sources/how_to/deploy_models/deploy_quantized.rst.txt
+++ b/docs/_sources/how_to/deploy_models/deploy_quantized.rst.txt
@@ -255,7 +255,7 @@ We create a Relay VM to build and execute the model.
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 25.372 seconds)
+ **Total running time of the script:** ( 1 minutes 19.144 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 dd070a1a1..fc48ee16e 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
@@ -158,7 +158,7 @@ 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]
1%|1 | 1952/132723 [00:00<00:06, 19399.67KB/s]
4%|4 | 5802/132723 [00:00<00:04, 30604.28KB/s]
10%|9 | 12820/132723 [00:00<00:02, 48644.83KB/s]
16%|#5 | 21041/132723 [00:00<00:01, 61882.38KB/s]
22%|##2 | 29257/132723 [00:00<00:01, 69187.03KB/s]
28%|##7 | 36556/132723 [00:00<00:01, 70478.37KB/s]
33%|###2 | 43767/132723 [00:00<00:01, 71008.77KB/s]
39%|###9 | 51860/132723 [00:00<00:01, 74163.86KB/s]
45%|####5 | 60034/132723 [00:00<00:00, 76530.44KB/s]
51%|#####1 | 68161/132723 [00:01<00:00, 77991.85KB/s]
58%|#####7 | 76409/132723 [00:01<00:00, 79363.71KB/s]
64%|######3 | 84697/132723 [00:01<00:00, 80428.89KB/s]
70%|####### | 93021/132723 [00:01<00:00, 81276.60KB/s]
76%|#######6 | 101315/132723 [00:01<00:00, 81776.29KB/s]
83%|########2 | 109562/132723 [00:01<00:00, 81983.22KB/s]
89%|########8
| 117845/132723 [00:01<00:00, 82235.45KB/s]
95%|#########5| 126231/132723 [00:01<00:00, 82713.67KB/s]
100%|##########| 132723/132723 [00:01<00:00, 74562.70KB/s]
+
0%| | 0/132723 [00:00<?, ?KB/s]
4%|3 | 4752/132723 [00:00<00:02, 47515.17KB/s]
10%|9 | 12617/132723 [00:00<00:01, 65826.37KB/s]
15%|#5 | 20499/132723 [00:00<00:01, 71755.79KB/s]
21%|##1 | 28397/132723 [00:00<00:01, 74605.24KB/s]
27%|##7 | 36289/132723 [00:00<00:01, 76160.00KB/s]
33%|###3 | 44319/132723 [00:00<00:01, 77565.61KB/s]
39%|###9 | 52308/132723 [00:00<00:01, 78322.63KB/s]
45%|####5 | 60195/132723 [00:00<00:00, 78491.36KB/s]
51%|#####1 | 68045/132723 [00:00<00:00, 78400.42KB/s]
57%|#####7 | 76004/132723 [00:01<00:00, 78766.15KB/s]
63%|######3 | 84004/132723 [00:01<00:00, 79139.23KB/s]
69%|######9 | 91918/132723 [00:01<00:00, 79138.11KB/s]
75%|#######5 | 99832/132723 [00:01<00:00, 79045.28KB/s]
81%|########1 | 107816/132723 [00:01<00:00, 79281.95KB/s]
87%|########7 | 115806/132723 [00:01<00:00, 79465.39KB/s]
93%|#########
3| 123832/132723 [00:01<00:00, 79698.80KB/s]
99%|#########9| 131837/132723 [00:01<00:00, 79803.01KB/s]
100%|##########| 132723/132723 [00:01<00:00, 77466.29KB/s]
@@ -241,7 +241,7 @@ Display result
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 2 minutes 24.100 seconds)
+ **Total running time of the script:** ( 2 minutes 24.274 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 970d1085d..49a0f8df4 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
@@ -5,22 +5,22 @@
Computation times
=================
-**10:47.156** total execution time for **how_to_deploy_models** files:
+**10:41.384** total execution time for **how_to_deploy_models** files:
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_object_detection_pytorch.py` (``deploy_object_detection_pytorch.py``) | 02:55.692 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_object_detection_pytorch.py` (``deploy_object_detection_pytorch.py``) | 02:57.631 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_ssd_gluoncv.py` (``deploy_ssd_gluoncv.py``) | 02:24.100 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_ssd_gluoncv.py` (``deploy_ssd_gluoncv.py``) | 02:24.274 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_prequantized_tflite.py` (``deploy_prequantized_tflite.py``) | 02:01.639 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_prequantized_tflite.py` (``deploy_prequantized_tflite.py``) | 01:59.854 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_quantized.py` (``deploy_quantized.py``) | 01:25.372 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_quantized.py` (``deploy_quantized.py``) | 01:19.144 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_prequantized.py` (``deploy_prequantized.py``) | 01:08.038 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_prequantized.py` (``deploy_prequantized.py``) | 01:07.701 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_android.py` (``deploy_model_on_android.py``) | 00:29.707 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_android.py` (``deploy_model_on_android.py``) | 00:29.762 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_rasp.py` (``deploy_model_on_rasp.py``) | 00:22.603 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_rasp.py` (``deploy_model_on_rasp.py``) | 00:23.011 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_sparse.py` (``deploy_sparse.py``) | 00:00.007 | 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 6b325c694..788280836 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
@@ -476,7 +476,7 @@ 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.zip38f87827-2bbd-487d-90d7-9352b529a175 from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/mobilenet0.25-9f83e440.zip...
+ Downloading /workspace/.mxnet/models/mobilenet0.25-9f83e440.zip1a99afe2-ef49-4e65-b085-29d2f003d48b from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/mobilenet0.25-9f83e440.zip...
@@ -590,7 +590,7 @@ Now, to actually convert the entire network, we have written `a pass in Relay <h
/workspace/python/tvm/driver/build_module.py:268: UserWarning: target_host parameter is going to be deprecated. Please pass in tvm.target.Target(target, host=target_host) instead.
"target_host parameter is going to be deprecated. "
- Check failed: (lower) is false: Intrinsic lowering function for target llvm, intrinsic name tir.sqrt, type 150 not found
+ Check failed: (lower) is false: FloatImm lowering function for target llvm type 150 not found
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 5d550e568..b75003fb9 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
=================
-**00:40.080** total execution time for **how_to_extend_tvm** files:
+**00:38.584** 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``) | 00:36.847 | 0.0 MB |
+| :ref:`sphx_glr_how_to_extend_tvm_bring_your_own_datatypes.py` (``bring_your_own_datatypes.py``) | 00:35.580 | 0.0 MB |
+-------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_extend_tvm_use_pass_instrument.py` (``use_pass_instrument.py``) | 00:02.276 | 0.0 MB |
+| :ref:`sphx_glr_how_to_extend_tvm_use_pass_instrument.py` (``use_pass_instrument.py``) | 00:02.109 | 0.0 MB |
+-------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_extend_tvm_use_pass_infra.py` (``use_pass_infra.py``) | 00:00.949 | 0.0 MB |
+| :ref:`sphx_glr_how_to_extend_tvm_use_pass_infra.py` (``use_pass_infra.py``) | 00:00.887 | 0.0 MB |
+-------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_extend_tvm_low_level_custom_pass.py` (``low_level_custom_pass.py``) | 00:00.008 | 0.0 MB |
+| :ref:`sphx_glr_how_to_extend_tvm_low_level_custom_pass.py` (``low_level_custom_pass.py``) | 00:00.007 | 0.0 MB |
+-------------------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/how_to/extend_tvm/use_pass_instrument.rst.txt b/docs/_sources/how_to/extend_tvm/use_pass_instrument.rst.txt
index 371d98de1..06d35d843 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
@@ -216,10 +216,10 @@ profile the execution time of each passes.
.. code-block:: none
Printing results of timing profile...
- InferType: 6840us [6840us] (45.60%; 45.60%)
- FoldScaleAxis: 8161us [6us] (54.40%; 54.40%)
- FoldConstant: 8156us [1652us] (54.36%; 99.93%)
- InferType: 6503us [6503us] (43.35%; 79.74%)
+ InferType: 6354us [6354us] (45.75%; 45.75%)
+ FoldScaleAxis: 7534us [5us] (54.25%; 54.25%)
+ FoldConstant: 7529us [1590us] (54.21%; 99.93%)
+ InferType: 5939us [5939us] (42.76%; 78.88%)
@@ -258,10 +258,10 @@ Refer to following sections and :py:func:`tvm.instrument.pass_instrument` for th
.. code-block:: none
Printing results of timing profile...
- InferType: 6597us [6597us] (44.00%; 44.00%)
- FoldScaleAxis: 8397us [6us] (56.00%; 56.00%)
- FoldConstant: 8391us [1711us] (55.96%; 99.93%)
- InferType: 6680us [6680us] (44.55%; 79.61%)
+ InferType: 6062us [6062us] (44.88%; 44.88%)
+ FoldScaleAxis: 7445us [5us] (55.12%; 55.12%)
+ FoldConstant: 7440us [1526us] (55.08%; 99.94%)
+ InferType: 5914us [5914us] (43.79%; 79.49%)
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 7403a4a8f..26ddc12dd 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
@@ -340,7 +340,7 @@ latency of convolution.
.. code-block:: none
- Convolution: 54.203512 ms
+ Convolution: 54.167578 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 4fd3c9a16..de546af91 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
@@ -671,7 +671,7 @@ be able to run on our build server
.. code-block:: none
- conv2d with tensor core: 6.885543 ms
+ conv2d with tensor core: 6.995651 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 0cf29348c..55a82f07b 100644
--- a/docs/_sources/how_to/optimize_operators/opt_gemm.rst.txt
+++ b/docs/_sources/how_to/optimize_operators/opt_gemm.rst.txt
@@ -143,8 +143,8 @@ Then we write a baseline implementation, the simplest way to write a matrix mult
.. code-block:: none
- Numpy running time: 0.018863
- Baseline: 3.335876
+ Numpy running time: 0.019033
+ Baseline: 3.534831
@@ -239,7 +239,7 @@ fill 32 * 32 * sizeof(float) which is 4KB in the cache whose total size is 32KB
.. code-block:: none
- Opt1: 0.306134
+ Opt1: 0.305472
@@ -342,7 +342,7 @@ In this tutorial, we chose to vectorize the inner loop row data since it is cach
.. code-block:: none
- Opt2: 0.339910
+ Opt2: 0.345996
@@ -438,7 +438,7 @@ the access pattern for A matrix is more cache friendly.
.. code-block:: none
- Opt3: 0.115891
+ Opt3: 0.121514
@@ -563,7 +563,7 @@ flattening.
.. code-block:: none
- Opt4: 0.110579
+ Opt4: 0.110981
@@ -685,7 +685,7 @@ write to C when all the block results are ready.
.. code-block:: none
- Opt5: 0.111304
+ Opt5: 0.111529
@@ -810,7 +810,7 @@ Futhermore, we can also utilize multi-core processors to do the thread-level par
.. code-block:: none
- Opt6: 0.145312
+ Opt6: 0.145572
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 fa65ff2a2..7c72f2339 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
=================
-**00:34.430** total execution time for **how_to_optimize_operators** files:
+**00:35.226** total execution time for **how_to_optimize_operators** files:
+-----------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_optimize_operators_opt_gemm.py` (``opt_gemm.py``) | 00:32.091 | 0.0 MB |
+| :ref:`sphx_glr_how_to_optimize_operators_opt_gemm.py` (``opt_gemm.py``) | 00:32.887 | 0.0 MB |
+-----------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_optimize_operators_opt_conv_tensorcore.py` (``opt_conv_tensorcore.py``) | 00:01.297 | 0.0 MB |
+| :ref:`sphx_glr_how_to_optimize_operators_opt_conv_tensorcore.py` (``opt_conv_tensorcore.py``) | 00:01.293 | 0.0 MB |
+-----------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_optimize_operators_opt_conv_cuda.py` (``opt_conv_cuda.py``) | 00:01.042 | 0.0 MB |
+| :ref:`sphx_glr_how_to_optimize_operators_opt_conv_cuda.py` (``opt_conv_cuda.py``) | 00:01.046 | 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 f41a3eb6d..e7b5a9c0b 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
=================
-**05:08.137** total execution time for **how_to_tune_with_autoscheduler** files:
+**05:16.608** 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``) | 02:29.143 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_conv2d_layer_cuda.py` (``tune_conv2d_layer_cuda.py``) | 02:35.377 | 0.0 MB |
+----------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_x86.py` (``tune_network_x86.py``) | 01:20.297 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_x86.py` (``tune_network_x86.py``) | 01:19.283 | 0.0 MB |
+----------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_cuda.py` (``tune_network_cuda.py``) | 00:43.367 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_cuda.py` (``tune_network_cuda.py``) | 00:42.637 | 0.0 MB |
+----------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_sparse_x86.py` (``tune_sparse_x86.py``) | 00:17.854 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_sparse_x86.py` (``tune_sparse_x86.py``) | 00:22.459 | 0.0 MB |
+----------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_mali.py` (``tune_network_mali.py``) | 00:08.906 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_mali.py` (``tune_network_mali.py``) | 00:08.478 | 0.0 MB |
+----------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_arm.py` (``tune_network_arm.py``) | 00:08.570 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_arm.py` (``tune_network_arm.py``) | 00:08.374 | 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 fab34f6bf..82e7074e3 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
@@ -240,355 +240,762 @@ cooperative fetching, unrolling and operator fusion.
compute: Buffer(compute_2: Pointer(float32), float32, [25088], [])}
buffer_map = {data_1: data, kernel_1: kernel, bias_1: bias, compute_1: compute}
preflattened_buffer_map = {data_1: data_3: Buffer(data_2, float32, [1, 512, 7, 7], []), kernel_1: kernel_3: Buffer(kernel_2, float32, [512, 512, 3, 3], []), bias_1: bias_3: Buffer(bias_2, float32, [1, 512, 1, 1], []), compute_1: compute_3: Buffer(compute_2, float32, [1, 512, 7, 7], [])} {
- attr [IterVar(blockIdx.x: int32, (nullptr), "ThreadIndex", "blockIdx.x")] "thread_extent" = 64;
- allocate(conv2d_nchw: Pointer(local float32), float32, [8]), storage_scope = local;
- allocate(pad_temp.shared: Pointer(shared float32), float32, [324]), storage_scope = shared;
- allocate(kernel.shared: Pointer(shared float32), float32, [288]), storage_scope = shared;
+ attr [IterVar(blockIdx.x: int32, (nullptr), "ThreadIndex", "blockIdx.x")] "thread_extent" = 128;
+ allocate(conv2d_nchw: Pointer(local float32), float32, [4]), storage_scope = local;
+ allocate(pad_temp.shared: Pointer(shared float32), float32, [1008]), storage_scope = shared;
+ allocate(kernel.shared: Pointer(shared float32), float32, [192]), storage_scope = shared;
attr [IterVar(threadIdx.x: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- conv2d_nchw_1: Buffer(conv2d_nchw, float32, [1], [], scope="local", align=4)[0] = 0f32
+ conv2d_nchw_1: Buffer(conv2d_nchw, float32, [4], [], scope="local", align=16)[0] = 0f32
conv2d_nchw_1[1] = 0f32
conv2d_nchw_1[2] = 0f32
conv2d_nchw_1[3] = 0f32
- conv2d_nchw_1[4] = 0f32
- conv2d_nchw_1[5] = 0f32
- conv2d_nchw_1[6] = 0f32
- conv2d_nchw_1[7] = 0f32
- for (rc.outer.outer: int32, 0, 128) {
- let cse_var_2: int32 = (rc.outer.outer*196)
- let cse_var_1: int32 = (rc.outer.outer*36)
+ for (rc.outer.outer: int32, 0, 32) {
+ let cse_var_2: int32 = (rc.outer.outer*784)
+ let cse_var_1: int32 = (rc.outer.outer*144)
{
- attr [IterVar(threadIdx.x_1: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1: Buffer(pad_temp.shared, float32, [324], [], scope="shared")[(threadIdx.x_1*2)] = @tir.if_then_else(((((9 <= floormod((threadIdx.x_1*2), 81)) && (floormod((threadIdx.x_1*2), 81) < 72)) && (1 <= floormod((threadIdx.x_1*2), 9))) && (floormod((threadIdx.x_1*2), 9) < 8)), data[((((cse_var_2 + (floordiv((threadIdx.x_1*2), 81)*49)) + (floordiv(floormod((threadIdx.x_1*2), 81), 9)*7)) + floormod((threadIdx.x_1*2), 9)) - 8)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*2) + 1)] = @tir.if_then_else(((((9 <= floormod(((threadIdx.x_1*2) + 1), 81)) && (floormod(((threadIdx.x_1*2) + 1), 81) < 72)) && (1 <= floormod(((threadIdx.x_1*2) + 1), 9))) && (floormod(((threadIdx.x_1*2) + 1), 9) < 8)), data[((((cse_var_2 + (floordiv(((threadIdx.x_1*2) + 1), 81)*49)) + (floordiv(floormod(((threadIdx.x_1*2) + 1), 81), 9)*7)) + floormod(((threadIdx.x_1*2) + 1), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1: Buffer(pad_temp.shared, float32, [1008], [], scope="shared")[threadIdx.x_1] = @tir.if_then_else((((9 <= threadIdx.x_1) && (1 <= floormod(threadIdx.x_1, 9))) && (floormod(threadIdx.x_1, 9) < 8)), data[(((cse_var_2 + (floordiv(threadIdx.x_1, 9)*7)) + floormod(threadIdx.x_1, 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 49)] = @tir.if_then_else((((9 <= floormod((threadIdx.x_1 + 49), 63)) && (1 <= floormod((threadIdx.x_1 + 4), 9))) && (floormod((threadIdx.x_1 + 4), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 49), 9)*7)) + floormod((threadIdx.x_1 + 4), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 98)] = @tir.if_then_else((((9 <= floormod((threadIdx.x_1 + 35), 63)) && (1 <= floormod((threadIdx.x_1 + 8), 9))) && (floormod((threadIdx.x_1 + 8), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 98), 9)*7)) + floormod((threadIdx.x_1 + 8), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 147)] = @tir.if_then_else((((9 <= floormod((threadIdx.x_1 + 21), 63)) && (1 <= floormod((threadIdx.x_1 + 3), 9))) && (floormod((threadIdx.x_1 + 3), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 147), 9)*7)) + floormod((threadIdx.x_1 + 3), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 196)] = @tir.if_then_else((((9 <= floormod((threadIdx.x_1 + 7), 63)) && (1 <= floormod((threadIdx.x_1 + 7), 9))) && (floormod((threadIdx.x_1 + 7), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 196), 9)*7)) + floormod((threadIdx.x_1 + 7), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 245)] = @tir.if_then_else((((9 <= floormod((threadIdx.x_1 + 56), 63)) && (1 <= floormod((threadIdx.x_1 + 2), 9))) && (floormod((threadIdx.x_1 + 2), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 245), 9)*7)) + floormod((threadIdx.x_1 + 2), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 294)] = @tir.if_then_else((((9 <= floormod((threadIdx.x_1 + 42), 63)) && (1 <= floormod((threadIdx.x_1 + 6), 9))) && (floormod((threadIdx.x_1 + 6), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 294), 9)*7)) + floormod((threadIdx.x_1 + 6), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 343)] = @tir.if_then_else((((9 <= floormod((threadIdx.x_1 + 28), 63)) && (1 <= floormod((threadIdx.x_1 + 1), 9))) && (floormod((threadIdx.x_1 + 1), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 343), 9)*7)) + floormod((threadIdx.x_1 + 1), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 392)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 5), 9)) && (floormod((threadIdx.x_1 + 5), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 392), 9)*7)) + floormod((threadIdx.x_1 + 5), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 441)] = @tir.if_then_else((((9 <= threadIdx.x_1) && (1 <= floormod(threadIdx.x_1, 9))) && (floormod(threadIdx.x_1, 9) < 8)), data[(((cse_var_2 + (floordiv(threadIdx.x_1, 9)*7)) + floormod(threadIdx.x_1, 9)) + 335)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 490)] = @tir.if_then_else((((9 <= floormod((threadIdx.x_1 + 49), 63)) && (1 <= floormod((threadIdx.x_1 + 4), 9))) && (floormod((threadIdx.x_1 + 4), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 490), 9)*7)) + floormod((threadIdx.x_1 + 4), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 539)] = @tir.if_then_else((((9 <= floormod((threadIdx.x_1 + 35), 63)) && (1 <= floormod((threadIdx.x_1 + 8), 9))) && (floormod((threadIdx.x_1 + 8), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 539), 9)*7)) + floormod((threadIdx.x_1 + 8), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 588)] = @tir.if_then_else((((9 <= floormod((threadIdx.x_1 + 21), 63)) && (1 <= floormod((threadIdx.x_1 + 3), 9))) && (floormod((threadIdx.x_1 + 3), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 588), 9)*7)) + floormod((threadIdx.x_1 + 3), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 637)] = @tir.if_then_else((((9 <= floormod((threadIdx.x_1 + 7), 63)) && (1 <= floormod((threadIdx.x_1 + 7), 9))) && (floormod((threadIdx.x_1 + 7), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 637), 9)*7)) + floormod((threadIdx.x_1 + 7), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 686)] = @tir.if_then_else((((9 <= floormod((threadIdx.x_1 + 56), 63)) && (1 <= floormod((threadIdx.x_1 + 2), 9))) && (floormod((threadIdx.x_1 + 2), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 686), 9)*7)) + floormod((threadIdx.x_1 + 2), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 735)] = @tir.if_then_else((((9 <= floormod((threadIdx.x_1 + 42), 63)) && (1 <= floormod((threadIdx.x_1 + 6), 9))) && (floormod((threadIdx.x_1 + 6), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 735), 9)*7)) + floormod((threadIdx.x_1 + 6), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 784)] = @tir.if_then_else((((9 <= floormod((threadIdx.x_1 + 28), 63)) && (1 <= floormod((threadIdx.x_1 + 1), 9))) && (floormod((threadIdx.x_1 + 1), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 784), 9)*7)) + floormod((threadIdx.x_1 + 1), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 833)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 5), 9)) && (floormod((threadIdx.x_1 + 5), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 833), 9)*7)) + floormod((threadIdx.x_1 + 5), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 882)] = @tir.if_then_else((((9 <= threadIdx.x_1) && (1 <= floormod(threadIdx.x_1, 9))) && (floormod(threadIdx.x_1, 9) < 8)), data[(((cse_var_2 + (floordiv(threadIdx.x_1, 9)*7)) + floormod(threadIdx.x_1, 9)) + 678)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 931)] = @tir.if_then_else((((9 <= floormod((threadIdx.x_1 + 49), 63)) && (1 <= floormod((threadIdx.x_1 + 4), 9))) && (floormod((threadIdx.x_1 + 4), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 931), 9)*7)) + floormod((threadIdx.x_1 + 4), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ if @tir.likely((threadIdx.x_1 < 28), dtype=bool) {
+ pad_temp.shared_1[(threadIdx.x_1 + 980)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 8), 9)) && (floormod((threadIdx.x_1 + 8), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 980), 9)*7)) + floormod((threadIdx.x_1 + 8), 9)) - 8)], 0f32, dtype=float32)
}
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[((threadIdx.x_1*2) + 98)] = @tir.if_then_else(((((9 <= floormod(((threadIdx.x_1*2) + 17), 81)) && (floormod(((threadIdx.x_1*2) + 17), 81) < 72)) && (1 <= floormod(((threadIdx.x_1*2) + 8), 9))) && (floormod(((threadIdx.x_1*2) + 8), 9) < 8)), data[((((cse_var_2 + (floordiv(((threadIdx.x_1*2) + 98), 81)*49)) + (floordiv(floormod(((threadIdx.x_1*2) + 17), 81), 9)*7)) + floormod(((threadIdx.x_1*2) + 8), 9)) - 8)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*2) + 99)] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*2), 9) + 2), 9)) && (floormod(((threadIdx.x_1*2) + 18), 81) < 72)) && (1 <= floormod((threadIdx.x_1*2), 9))) && (floormod((threadIdx.x_1*2), 9) < 8)), data[((((cse_var_2 + (floordiv(((threadIdx.x_1*2) + 99), 81)*49)) + (floormod((floordiv((threadIdx.x_1*2), 9) + 2), 9)*7)) + floormod((threadIdx.x_1*2), 9)) - 8)], 0f32, dtype=float32)
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[((threadIdx.x_1*2) + 196)] = @tir.if_then_else(((((9 <= floormod(((threadIdx.x_1*2) + 34), 81)) && (floormod(((threadIdx.x_1*2) + 34), 81) < 72)) && (1 <= floormod(((threadIdx.x_1*2) + 7), 9))) && (floormod(((threadIdx.x_1*2) + 7), 9) < 8)), data[((((cse_var_2 + (floordiv(((threadIdx.x_1*2) + 196), 81)*49)) + (floordiv(floormod(((threadIdx.x_1*2) + 34), 81), 9)*7)) + floormod(((threadIdx.x_1*2) + 7), 9)) - 8)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*2) + 197)] = @tir.if_then_else(((((9 <= floormod(((threadIdx.x_1*2) + 35), 81)) && (floormod(((threadIdx.x_1*2) + 35), 81) < 72)) && (1 <= floormod(((threadIdx.x_1*2) + 8), 9))) && (floormod(((threadIdx.x_1*2) + 8), 9) < 8)), data[((((cse_var_2 + (floordiv(((threadIdx.x_1*2) + 197), 81)*49)) + (floordiv(floormod(((threadIdx.x_1*2) + 35), 81), 9)*7)) + floormod(((threadIdx.x_1*2) + 8), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_2: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ kernel.shared_1: Buffer(kernel.shared, float32, [192], [], scope="shared")[threadIdx.x_2] = kernel[(((((blockIdx.x*18432) + (floordiv(threadIdx.x_2, 48)*4608)) + cse_var_1) + (floordiv(floormod(threadIdx.x_2, 48), 3)*9)) + floormod(threadIdx.x_2, 3))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ kernel.shared_1[(threadIdx.x_2 + 49)] = kernel[(((((blockIdx.x*18432) + (floordiv((threadIdx.x_2 + 49), 48)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 1), 48), 3)*9)) + floormod((threadIdx.x_2 + 1), 3))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ kernel.shared_1[(threadIdx.x_2 + 98)] = kernel[(((((blockIdx.x*18432) + (floordiv((threadIdx.x_2 + 98), 48)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 2), 48), 3)*9)) + floormod((threadIdx.x_2 + 2), 3))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ if @tir.likely((threadIdx.x_2 < 45), dtype=bool) {
+ kernel.shared_1[(threadIdx.x_2 + 147)] = kernel[((((((blockIdx.x*18432) + (floordiv((threadIdx.x_2 + 147), 48)*4608)) + cse_var_1) + (floordiv(threadIdx.x_2, 3)*9)) + floormod(threadIdx.x_2, 3)) + 9)]
}
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- if @tir.likely((threadIdx.x_1 < 15), dtype=bool) {
- pad_temp.shared_1[((threadIdx.x_1*2) + 294)] = @tir.if_then_else((((threadIdx.x_1 < 11) && (1 <= floormod(((threadIdx.x_1*2) + 6), 9))) && (floormod(((threadIdx.x_1*2) + 6), 9) < 8)), data[(((cse_var_2 + (floordiv(floormod(((threadIdx.x_1*2) + 51), 81), 9)*7)) + floormod(((threadIdx.x_1*2) + 6), 9)) + 139)], 0f32, dtype=float32)
- }
- if @tir.likely((threadIdx.x_1 < 15), dtype=bool) {
- pad_temp.shared_1[((threadIdx.x_1*2) + 295)] = @tir.if_then_else((((threadIdx.x_1 < 10) && (1 <= floormod(((threadIdx.x_1*2) + 7), 9))) && (floormod(((threadIdx.x_1*2) + 7), 9) < 8)), data[(((cse_var_2 + (floordiv(floormod(((threadIdx.x_1*2) + 52), 81), 9)*7)) + floormod(((threadIdx.x_1*2) + 7), 9)) + 139)], 0f32, dtype=float32)
- }
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[0]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[48]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[96]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[144]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[1]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[49]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[97]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[145]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[2]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[50]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[98]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[146]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[3]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[51]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[99]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[147]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[4]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[52]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[100]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[148]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[5]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[53]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[101]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[149]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[6]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[54]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[102]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[150]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[7]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[55]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[103]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[151]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[8]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[56]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[104]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[152]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[9]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[57]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[105]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[153]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[10]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[58]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[106]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[154]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[11]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[59]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[107]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[155]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[12]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[60]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[108]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[156]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[13]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[61]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[109]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[157]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[14]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[62]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[110]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[158]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 315)]*kernel.shared_1[15]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 315)]*kernel.shared_1[63]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 315)]*kernel.shared_1[111]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 315)]*kernel.shared_1[159]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 316)]*kernel.shared_1[16]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 316)]*kernel.shared_1[64]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 316)]*kernel.shared_1[112]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 316)]*kernel.shared_1[160]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 317)]*kernel.shared_1[17]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 317)]*kernel.shared_1[65]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 317)]*kernel.shared_1[113]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 317)]*kernel.shared_1[161]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 378)]*kernel.shared_1[18]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 378)]*kernel.shared_1[66]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 378)]*kernel.shared_1[114]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 378)]*kernel.shared_1[162]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 379)]*kernel.shared_1[19]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 379)]*kernel.shared_1[67]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 379)]*kernel.shared_1[115]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 379)]*kernel.shared_1[163]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 380)]*kernel.shared_1[20]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 380)]*kernel.shared_1[68]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 380)]*kernel.shared_1[116]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 380)]*kernel.shared_1[164]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 441)]*kernel.shared_1[21]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 441)]*kernel.shared_1[69]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 441)]*kernel.shared_1[117]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 441)]*kernel.shared_1[165]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 442)]*kernel.shared_1[22]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 442)]*kernel.shared_1[70]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 442)]*kernel.shared_1[118]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 442)]*kernel.shared_1[166]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 443)]*kernel.shared_1[23]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 443)]*kernel.shared_1[71]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 443)]*kernel.shared_1[119]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 443)]*kernel.shared_1[167]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 504)]*kernel.shared_1[24]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 504)]*kernel.shared_1[72]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 504)]*kernel.shared_1[120]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 504)]*kernel.shared_1[168]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 505)]*kernel.shared_1[25]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 505)]*kernel.shared_1[73]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 505)]*kernel.shared_1[121]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 505)]*kernel.shared_1[169]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 506)]*kernel.shared_1[26]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 506)]*kernel.shared_1[74]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 506)]*kernel.shared_1[122]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 506)]*kernel.shared_1[170]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 567)]*kernel.shared_1[27]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 567)]*kernel.shared_1[75]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 567)]*kernel.shared_1[123]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 567)]*kernel.shared_1[171]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 568)]*kernel.shared_1[28]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 568)]*kernel.shared_1[76]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 568)]*kernel.shared_1[124]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 568)]*kernel.shared_1[172]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 569)]*kernel.shared_1[29]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 569)]*kernel.shared_1[77]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 569)]*kernel.shared_1[125]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 569)]*kernel.shared_1[173]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 630)]*kernel.shared_1[30]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 630)]*kernel.shared_1[78]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 630)]*kernel.shared_1[126]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 630)]*kernel.shared_1[174]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 631)]*kernel.shared_1[31]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 631)]*kernel.shared_1[79]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 631)]*kernel.shared_1[127]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 631)]*kernel.shared_1[175]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 632)]*kernel.shared_1[32]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 632)]*kernel.shared_1[80]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 632)]*kernel.shared_1[128]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 632)]*kernel.shared_1[176]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 693)]*kernel.shared_1[33]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 693)]*kernel.shared_1[81]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 693)]*kernel.shared_1[129]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 693)]*kernel.shared_1[177]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 694)]*kernel.shared_1[34]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 694)]*kernel.shared_1[82]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 694)]*kernel.shared_1[130]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 694)]*kernel.shared_1[178]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 695)]*kernel.shared_1[35]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 695)]*kernel.shared_1[83]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 695)]*kernel.shared_1[131]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 695)]*kernel.shared_1[179]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 756)]*kernel.shared_1[36]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 756)]*kernel.shared_1[84]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 756)]*kernel.shared_1[132]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 756)]*kernel.shared_1[180]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 757)]*kernel.shared_1[37]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 757)]*kernel.shared_1[85]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 757)]*kernel.shared_1[133]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 757)]*kernel.shared_1[181]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 758)]*kernel.shared_1[38]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 758)]*kernel.shared_1[86]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 758)]*kernel.shared_1[134]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 758)]*kernel.shared_1[182]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 819)]*kernel.shared_1[39]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 819)]*kernel.shared_1[87]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 819)]*kernel.shared_1[135]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 819)]*kernel.shared_1[183]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 820)]*kernel.shared_1[40]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 820)]*kernel.shared_1[88]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 820)]*kernel.shared_1[136]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 820)]*kernel.shared_1[184]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 821)]*kernel.shared_1[41]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 821)]*kernel.shared_1[89]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 821)]*kernel.shared_1[137]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 821)]*kernel.shared_1[185]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 882)]*kernel.shared_1[42]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 882)]*kernel.shared_1[90]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 882)]*kernel.shared_1[138]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 882)]*kernel.shared_1[186]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 883)]*kernel.shared_1[43]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 883)]*kernel.shared_1[91]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 883)]*kernel.shared_1[139]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 883)]*kernel.shared_1[187]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 884)]*kernel.shared_1[44]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 884)]*kernel.shared_1[92]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 884)]*kernel.shared_1[140]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 884)]*kernel.shared_1[188]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 945)]*kernel.shared_1[45]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 945)]*kernel.shared_1[93]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 945)]*kernel.shared_1[141]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 945)]*kernel.shared_1[189]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 946)]*kernel.shared_1[46]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 946)]*kernel.shared_1[94]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 946)]*kernel.shared_1[142]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 946)]*kernel.shared_1[190]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 947)]*kernel.shared_1[47]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 947)]*kernel.shared_1[95]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 947)]*kernel.shared_1[143]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 947)]*kernel.shared_1[191]))
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[threadIdx.x_1] = @tir.if_then_else(((1 <= floormod(threadIdx.x_1, 9)) && (floormod(threadIdx.x_1, 9) < 8)), data[(((cse_var_2 + (floordiv(threadIdx.x_1, 9)*7)) + floormod(threadIdx.x_1, 9)) - 1)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 49)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 4), 9)) && (floormod((threadIdx.x_1 + 4), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 49), 9)*7)) + floormod((threadIdx.x_1 + 4), 9)) - 1)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 98)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 8), 9)) && (floormod((threadIdx.x_1 + 8), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 98), 9)*7)) + floormod((threadIdx.x_1 + 8), 9)) - 1)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 147)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 3), 9)) && (floormod((threadIdx.x_1 + 3), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 147), 9)*7)) + floormod((threadIdx.x_1 + 3), 9)) - 1)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 196)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 7), 9)) && (floormod((threadIdx.x_1 + 7), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 196), 9)*7)) + floormod((threadIdx.x_1 + 7), 9)) - 1)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 245)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 2), 9)) && (floormod((threadIdx.x_1 + 2), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 245), 9)*7)) + floormod((threadIdx.x_1 + 2), 9)) - 1)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 294)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 6), 9)) && (floormod((threadIdx.x_1 + 6), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 294), 9)*7)) + floormod((threadIdx.x_1 + 6), 9)) - 1)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 343)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 1), 9)) && (floormod((threadIdx.x_1 + 1), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 343), 9)*7)) + floormod((threadIdx.x_1 + 1), 9)) - 1)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 392)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 5), 9)) && (floormod((threadIdx.x_1 + 5), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 392), 9)*7)) + floormod((threadIdx.x_1 + 5), 9)) - 1)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 441)] = @tir.if_then_else(((1 <= floormod(threadIdx.x_1, 9)) && (floormod(threadIdx.x_1, 9) < 8)), data[(((cse_var_2 + (floordiv(threadIdx.x_1, 9)*7)) + floormod(threadIdx.x_1, 9)) + 342)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 490)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 4), 9)) && (floormod((threadIdx.x_1 + 4), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 490), 9)*7)) + floormod((threadIdx.x_1 + 4), 9)) - 1)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 539)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 8), 9)) && (floormod((threadIdx.x_1 + 8), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 539), 9)*7)) + floormod((threadIdx.x_1 + 8), 9)) - 1)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 588)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 3), 9)) && (floormod((threadIdx.x_1 + 3), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 588), 9)*7)) + floormod((threadIdx.x_1 + 3), 9)) - 1)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 637)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 7), 9)) && (floormod((threadIdx.x_1 + 7), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 637), 9)*7)) + floormod((threadIdx.x_1 + 7), 9)) - 1)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 686)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 2), 9)) && (floormod((threadIdx.x_1 + 2), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 686), 9)*7)) + floormod((threadIdx.x_1 + 2), 9)) - 1)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 735)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 6), 9)) && (floormod((threadIdx.x_1 + 6), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 735), 9)*7)) + floormod((threadIdx.x_1 + 6), 9)) - 1)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 784)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 1), 9)) && (floormod((threadIdx.x_1 + 1), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 784), 9)*7)) + floormod((threadIdx.x_1 + 1), 9)) - 1)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 833)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 5), 9)) && (floormod((threadIdx.x_1 + 5), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 833), 9)*7)) + floormod((threadIdx.x_1 + 5), 9)) - 1)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 882)] = @tir.if_then_else(((1 <= floormod(threadIdx.x_1, 9)) && (floormod(threadIdx.x_1, 9) < 8)), data[(((cse_var_2 + (floordiv(threadIdx.x_1, 9)*7)) + floormod(threadIdx.x_1, 9)) + 685)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 931)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 4), 9)) && (floormod((threadIdx.x_1 + 4), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 931), 9)*7)) + floormod((threadIdx.x_1 + 4), 9)) - 1)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ if @tir.likely((threadIdx.x_1 < 28), dtype=bool) {
+ pad_temp.shared_1[(threadIdx.x_1 + 980)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 8), 9)) && (floormod((threadIdx.x_1 + 8), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 980), 9)*7)) + floormod((threadIdx.x_1 + 8), 9)) - 1)], 0f32, dtype=float32)
}
- attr [IterVar(threadIdx.x_2: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1: Buffer(kernel.shared, float32, [288], [], scope="shared")[threadIdx.x_2] = kernel[((((blockIdx.x*36864) + (floordiv(threadIdx.x_2, 36)*4608)) + cse_var_1) + floormod(threadIdx.x_2, 36))]
attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 49)] = kernel[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 49), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 13), 36), 9)*9)) + (floordiv(floormod((threadIdx.x_2 + 4), 9), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
+ kernel.shared_1[threadIdx.x_2] = kernel[((((((blockIdx.x*18432) + (floordiv(threadIdx.x_2, 48)*4608)) + cse_var_1) + (floordiv(floormod(threadIdx.x_2, 48), 3)*9)) + floormod(threadIdx.x_2, 3)) + 3)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ kernel.shared_1[(threadIdx.x_2 + 49)] = kernel[((((((blockIdx.x*18432) + (floordiv((threadIdx.x_2 + 49), 48)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 1), 48), 3)*9)) + floormod((threadIdx.x_2 + 1), 3)) + 3)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ kernel.shared_1[(threadIdx.x_2 + 98)] = kernel[((((((blockIdx.x*18432) + (floordiv((threadIdx.x_2 + 98), 48)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 2), 48), 3)*9)) + floormod((threadIdx.x_2 + 2), 3)) + 3)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ if @tir.likely((threadIdx.x_2 < 45), dtype=bool) {
+ kernel.shared_1[(threadIdx.x_2 + 147)] = kernel[((((((blockIdx.x*18432) + (floordiv((threadIdx.x_2 + 147), 48)*4608)) + cse_var_1) + (floordiv(threadIdx.x_2, 3)*9)) + floormod(threadIdx.x_2, 3)) + 12)]
+ }
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[0]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[48]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[96]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[144]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[1]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[49]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[97]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[145]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[2]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[50]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[98]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[146]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[3]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[51]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[99]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[147]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[4]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[52]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[100]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[148]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[5]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[53]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[101]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[149]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[6]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[54]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[102]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[150]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[7]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[55]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[103]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[151]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[8]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[56]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[104]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[152]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[9]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[57]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[105]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[153]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[10]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[58]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[106]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[154]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[11]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[59]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[107]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[155]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[12]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[60]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[108]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[156]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[13]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[61]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[109]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[157]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[14]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[62]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[110]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[158]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 315)]*kernel.shared_1[15]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 315)]*kernel.shared_1[63]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 315)]*kernel.shared_1[111]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 315)]*kernel.shared_1[159]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 316)]*kernel.shared_1[16]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 316)]*kernel.shared_1[64]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 316)]*kernel.shared_1[112]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 316)]*kernel.shared_1[160]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 317)]*kernel.shared_1[17]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 317)]*kernel.shared_1[65]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 317)]*kernel.shared_1[113]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 317)]*kernel.shared_1[161]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 378)]*kernel.shared_1[18]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 378)]*kernel.shared_1[66]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 378)]*kernel.shared_1[114]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 378)]*kernel.shared_1[162]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 379)]*kernel.shared_1[19]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 379)]*kernel.shared_1[67]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 379)]*kernel.shared_1[115]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 379)]*kernel.shared_1[163]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 380)]*kernel.shared_1[20]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 380)]*kernel.shared_1[68]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 380)]*kernel.shared_1[116]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 380)]*kernel.shared_1[164]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 441)]*kernel.shared_1[21]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 441)]*kernel.shared_1[69]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 441)]*kernel.shared_1[117]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 441)]*kernel.shared_1[165]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 442)]*kernel.shared_1[22]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 442)]*kernel.shared_1[70]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 442)]*kernel.shared_1[118]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 442)]*kernel.shared_1[166]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 443)]*kernel.shared_1[23]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 443)]*kernel.shared_1[71]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 443)]*kernel.shared_1[119]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 443)]*kernel.shared_1[167]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 504)]*kernel.shared_1[24]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 504)]*kernel.shared_1[72]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 504)]*kernel.shared_1[120]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 504)]*kernel.shared_1[168]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 505)]*kernel.shared_1[25]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 505)]*kernel.shared_1[73]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 505)]*kernel.shared_1[121]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 505)]*kernel.shared_1[169]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 506)]*kernel.shared_1[26]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 506)]*kernel.shared_1[74]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 506)]*kernel.shared_1[122]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 506)]*kernel.shared_1[170]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 567)]*kernel.shared_1[27]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 567)]*kernel.shared_1[75]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 567)]*kernel.shared_1[123]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 567)]*kernel.shared_1[171]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 568)]*kernel.shared_1[28]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 568)]*kernel.shared_1[76]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 568)]*kernel.shared_1[124]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 568)]*kernel.shared_1[172]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 569)]*kernel.shared_1[29]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 569)]*kernel.shared_1[77]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 569)]*kernel.shared_1[125]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 569)]*kernel.shared_1[173]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 630)]*kernel.shared_1[30]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 630)]*kernel.shared_1[78]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 630)]*kernel.shared_1[126]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 630)]*kernel.shared_1[174]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 631)]*kernel.shared_1[31]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 631)]*kernel.shared_1[79]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 631)]*kernel.shared_1[127]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 631)]*kernel.shared_1[175]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 632)]*kernel.shared_1[32]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 632)]*kernel.shared_1[80]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 632)]*kernel.shared_1[128]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 632)]*kernel.shared_1[176]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 693)]*kernel.shared_1[33]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 693)]*kernel.shared_1[81]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 693)]*kernel.shared_1[129]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 693)]*kernel.shared_1[177]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 694)]*kernel.shared_1[34]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 694)]*kernel.shared_1[82]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 694)]*kernel.shared_1[130]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 694)]*kernel.shared_1[178]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 695)]*kernel.shared_1[35]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 695)]*kernel.shared_1[83]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 695)]*kernel.shared_1[131]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 695)]*kernel.shared_1[179]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 756)]*kernel.shared_1[36]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 756)]*kernel.shared_1[84]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 756)]*kernel.shared_1[132]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 756)]*kernel.shared_1[180]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 757)]*kernel.shared_1[37]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 757)]*kernel.shared_1[85]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 757)]*kernel.shared_1[133]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 757)]*kernel.shared_1[181]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 758)]*kernel.shared_1[38]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 758)]*kernel.shared_1[86]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 758)]*kernel.shared_1[134]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 758)]*kernel.shared_1[182]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 819)]*kernel.shared_1[39]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 819)]*kernel.shared_1[87]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 819)]*kernel.shared_1[135]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 819)]*kernel.shared_1[183]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 820)]*kernel.shared_1[40]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 820)]*kernel.shared_1[88]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 820)]*kernel.shared_1[136]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 820)]*kernel.shared_1[184]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 821)]*kernel.shared_1[41]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 821)]*kernel.shared_1[89]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 821)]*kernel.shared_1[137]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 821)]*kernel.shared_1[185]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 882)]*kernel.shared_1[42]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 882)]*kernel.shared_1[90]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 882)]*kernel.shared_1[138]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 882)]*kernel.shared_1[186]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 883)]*kernel.shared_1[43]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 883)]*kernel.shared_1[91]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 883)]*kernel.shared_1[139]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 883)]*kernel.shared_1[187]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 884)]*kernel.shared_1[44]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 884)]*kernel.shared_1[92]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 884)]*kernel.shared_1[140]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 884)]*kernel.shared_1[188]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 945)]*kernel.shared_1[45]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 945)]*kernel.shared_1[93]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 945)]*kernel.shared_1[141]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 945)]*kernel.shared_1[189]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 946)]*kernel.shared_1[46]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 946)]*kernel.shared_1[94]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 946)]*kernel.shared_1[142]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 946)]*kernel.shared_1[190]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 947)]*kernel.shared_1[47]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 947)]*kernel.shared_1[95]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 947)]*kernel.shared_1[143]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 947)]*kernel.shared_1[191]))
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[threadIdx.x_1] = @tir.if_then_else(((1 <= floormod(threadIdx.x_1, 9)) && (floormod(threadIdx.x_1, 9) < 8)), data[(((cse_var_2 + (floordiv(threadIdx.x_1, 9)*7)) + floormod(threadIdx.x_1, 9)) + 6)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 49)] = @tir.if_then_else((((floormod((threadIdx.x_1 + 49), 63) < 54) && (1 <= floormod((threadIdx.x_1 + 4), 9))) && (floormod((threadIdx.x_1 + 4), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 49), 9)*7)) + floormod((threadIdx.x_1 + 4), 9)) + 6)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 98)] = @tir.if_then_else((((floormod((threadIdx.x_1 + 35), 63) < 54) && (1 <= floormod((threadIdx.x_1 + 8), 9))) && (floormod((threadIdx.x_1 + 8), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 98), 9)*7)) + floormod((threadIdx.x_1 + 8), 9)) + 6)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 147)] = @tir.if_then_else((((floormod((threadIdx.x_1 + 21), 63) < 54) && (1 <= floormod((threadIdx.x_1 + 3), 9))) && (floormod((threadIdx.x_1 + 3), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 147), 9)*7)) + floormod((threadIdx.x_1 + 3), 9)) + 6)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 196)] = @tir.if_then_else((((floormod((threadIdx.x_1 + 7), 63) < 54) && (1 <= floormod((threadIdx.x_1 + 7), 9))) && (floormod((threadIdx.x_1 + 7), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 196), 9)*7)) + floormod((threadIdx.x_1 + 7), 9)) + 6)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 245)] = @tir.if_then_else((((floormod((threadIdx.x_1 + 56), 63) < 54) && (1 <= floormod((threadIdx.x_1 + 2), 9))) && (floormod((threadIdx.x_1 + 2), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 245), 9)*7)) + floormod((threadIdx.x_1 + 2), 9)) + 6)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 294)] = @tir.if_then_else((((floormod((threadIdx.x_1 + 42), 63) < 54) && (1 <= floormod((threadIdx.x_1 + 6), 9))) && (floormod((threadIdx.x_1 + 6), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 294), 9)*7)) + floormod((threadIdx.x_1 + 6), 9)) + 6)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 343)] = @tir.if_then_else((((floormod((threadIdx.x_1 + 28), 63) < 54) && (1 <= floormod((threadIdx.x_1 + 1), 9))) && (floormod((threadIdx.x_1 + 1), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 343), 9)*7)) + floormod((threadIdx.x_1 + 1), 9)) + 6)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 392)] = @tir.if_then_else((((floormod((threadIdx.x_1 + 14), 63) < 54) && (1 <= floormod((threadIdx.x_1 + 5), 9))) && (floormod((threadIdx.x_1 + 5), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 392), 9)*7)) + floormod((threadIdx.x_1 + 5), 9)) + 6)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 441)] = @tir.if_then_else(((1 <= floormod(threadIdx.x_1, 9)) && (floormod(threadIdx.x_1, 9) < 8)), data[(((cse_var_2 + (floordiv(threadIdx.x_1, 9)*7)) + floormod(threadIdx.x_1, 9)) + 349)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 490)] = @tir.if_then_else((((floormod((threadIdx.x_1 + 49), 63) < 54) && (1 <= floormod((threadIdx.x_1 + 4), 9))) && (floormod((threadIdx.x_1 + 4), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 490), 9)*7)) + floormod((threadIdx.x_1 + 4), 9)) + 6)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 539)] = @tir.if_then_else((((floormod((threadIdx.x_1 + 35), 63) < 54) && (1 <= floormod((threadIdx.x_1 + 8), 9))) && (floormod((threadIdx.x_1 + 8), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 539), 9)*7)) + floormod((threadIdx.x_1 + 8), 9)) + 6)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 588)] = @tir.if_then_else((((floormod((threadIdx.x_1 + 21), 63) < 54) && (1 <= floormod((threadIdx.x_1 + 3), 9))) && (floormod((threadIdx.x_1 + 3), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 588), 9)*7)) + floormod((threadIdx.x_1 + 3), 9)) + 6)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 637)] = @tir.if_then_else((((floormod((threadIdx.x_1 + 7), 63) < 54) && (1 <= floormod((threadIdx.x_1 + 7), 9))) && (floormod((threadIdx.x_1 + 7), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 637), 9)*7)) + floormod((threadIdx.x_1 + 7), 9)) + 6)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 686)] = @tir.if_then_else((((floormod((threadIdx.x_1 + 56), 63) < 54) && (1 <= floormod((threadIdx.x_1 + 2), 9))) && (floormod((threadIdx.x_1 + 2), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 686), 9)*7)) + floormod((threadIdx.x_1 + 2), 9)) + 6)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 735)] = @tir.if_then_else((((floormod((threadIdx.x_1 + 42), 63) < 54) && (1 <= floormod((threadIdx.x_1 + 6), 9))) && (floormod((threadIdx.x_1 + 6), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 735), 9)*7)) + floormod((threadIdx.x_1 + 6), 9)) + 6)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 784)] = @tir.if_then_else((((floormod((threadIdx.x_1 + 28), 63) < 54) && (1 <= floormod((threadIdx.x_1 + 1), 9))) && (floormod((threadIdx.x_1 + 1), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 784), 9)*7)) + floormod((threadIdx.x_1 + 1), 9)) + 6)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 833)] = @tir.if_then_else((((floormod((threadIdx.x_1 + 14), 63) < 54) && (1 <= floormod((threadIdx.x_1 + 5), 9))) && (floormod((threadIdx.x_1 + 5), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 833), 9)*7)) + floormod((threadIdx.x_1 + 5), 9)) + 6)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 882)] = @tir.if_then_else(((1 <= floormod(threadIdx.x_1, 9)) && (floormod(threadIdx.x_1, 9) < 8)), data[(((cse_var_2 + (floordiv(threadIdx.x_1, 9)*7)) + floormod(threadIdx.x_1, 9)) + 692)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 931)] = @tir.if_then_else((((floormod((threadIdx.x_1 + 49), 63) < 54) && (1 <= floormod((threadIdx.x_1 + 4), 9))) && (floormod((threadIdx.x_1 + 4), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 931), 9)*7)) + floormod((threadIdx.x_1 + 4), 9)) + 6)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ if @tir.likely((threadIdx.x_1 < 28), dtype=bool) {
+ pad_temp.shared_1[(threadIdx.x_1 + 980)] = @tir.if_then_else((((floormod((threadIdx.x_1 + 35), 63) < 54) && (1 <= floormod((threadIdx.x_1 + 8), 9))) && (floormod((threadIdx.x_1 + 8), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 980), 9)*7)) + floormod((threadIdx.x_1 + 8), 9)) + 6)], 0f32, dtype=float32)
+ }
attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 98)] = kernel[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 98), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 26), 36), 9)*9)) + (floordiv(floormod((threadIdx.x_2 + 8), 9), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
+ kernel.shared_1[threadIdx.x_2] = kernel[((((((blockIdx.x*18432) + (floordiv(threadIdx.x_2, 48)*4608)) + cse_var_1) + (floordiv(floormod(threadIdx.x_2, 48), 3)*9)) + floormod(threadIdx.x_2, 3)) + 6)]
attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 147)] = kernel[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 147), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 3), 36), 9)*9)) + (floormod((floordiv(threadIdx.x_2, 3) + 1), 3)*3)) + floormod(threadIdx.x_2, 3))]
+ kernel.shared_1[(threadIdx.x_2 + 49)] = kernel[((((((blockIdx.x*18432) + (floordiv((threadIdx.x_2 + 49), 48)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 1), 48), 3)*9)) + floormod((threadIdx.x_2 + 1), 3)) + 6)]
attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 196)] = kernel[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 196), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 16), 36), 9)*9)) + (floordiv(floormod((threadIdx.x_2 + 7), 9), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
+ kernel.shared_1[(threadIdx.x_2 + 98)] = kernel[((((((blockIdx.x*18432) + (floordiv((threadIdx.x_2 + 98), 48)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 2), 48), 3)*9)) + floormod((threadIdx.x_2 + 2), 3)) + 6)]
attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- if @tir.likely((threadIdx.x_2 < 43), dtype=bool) {
- kernel.shared_1[(threadIdx.x_2 + 245)] = kernel[(((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 245), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 29), 36), 9)*9)) + floormod((threadIdx.x_2 + 2), 9))]
+ if @tir.likely((threadIdx.x_2 < 45), dtype=bool) {
+ kernel.shared_1[(threadIdx.x_2 + 147)] = kernel[((((((blockIdx.x*18432) + (floordiv((threadIdx.x_2 + 147), 48)*4608)) + cse_var_1) + (floordiv(threadIdx.x_2, 3)*9)) + floormod(threadIdx.x_2, 3)) + 15)]
}
conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[0]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[36]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[72]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[108]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[144]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[180]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[216]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[252]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[48]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[96]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[144]))
conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[1]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[37]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[73]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[109]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[145]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[181]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[217]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[253]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[49]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[97]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[145]))
conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[2]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[38]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[74]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[110]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[146]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[182]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[218]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[254]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 9)]*kernel.shared_1[3]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 9)]*kernel.shared_1[39]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 9)]*kernel.shared_1[75]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 9)]*kernel.shared_1[111]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 9)]*kernel.shared_1[147]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 9)]*kernel.shared_1[183]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 9)]*kernel.shared_1[219]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 9)]*kernel.shared_1[255]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 10)]*kernel.shared_1[4]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 10)]*kernel.shared_1[40]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 10)]*kernel.shared_1[76]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 10)]*kernel.shared_1[112]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 10)]*kernel.shared_1[148]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 10)]*kernel.shared_1[184]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 10)]*kernel.shared_1[220]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 10)]*kernel.shared_1[256]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 11)]*kernel.shared_1[5]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 11)]*kernel.shared_1[41]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 11)]*kernel.shared_1[77]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 11)]*kernel.shared_1[113]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 11)]*kernel.shared_1[149]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 11)]*kernel.shared_1[185]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 11)]*kernel.shared_1[221]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 11)]*kernel.shared_1[257]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[6]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[42]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[78]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[114]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[150]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[186]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[222]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[258]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[7]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[43]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[79]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[115]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[151]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[187]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[223]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[259]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[8]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[44]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[80]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[116]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[152]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[188]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[224]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[260]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 81)]*kernel.shared_1[9]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 81)]*kernel.shared_1[45]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 81)]*kernel.shared_1[81]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 81)]*kernel.shared_1[117]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 81)]*kernel.shared_1[153]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 81)]*kernel.shared_1[189]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 81)]*kernel.shared_1[225]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 81)]*kernel.shared_1[261]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 82)]*kernel.shared_1[10]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 82)]*kernel.shared_1[46]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 82)]*kernel.shared_1[82]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 82)]*kernel.shared_1[118]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 82)]*kernel.shared_1[154]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 82)]*kernel.shared_1[190]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 82)]*kernel.shared_1[226]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 82)]*kernel.shared_1[262]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 83)]*kernel.shared_1[11]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 83)]*kernel.shared_1[47]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 83)]*kernel.shared_1[83]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 83)]*kernel.shared_1[119]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 83)]*kernel.shared_1[155]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 83)]*kernel.shared_1[191]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 83)]*kernel.shared_1[227]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 83)]*kernel.shared_1[263]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 90)]*kernel.shared_1[12]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 90)]*kernel.shared_1[48]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 90)]*kernel.shared_1[84]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 90)]*kernel.shared_1[120]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 90)]*kernel.shared_1[156]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 90)]*kernel.shared_1[192]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 90)]*kernel.shared_1[228]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 90)]*kernel.shared_1[264]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 91)]*kernel.shared_1[13]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 91)]*kernel.shared_1[49]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 91)]*kernel.shared_1[85]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 91)]*kernel.shared_1[121]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 91)]*kernel.shared_1[157]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 91)]*kernel.shared_1[193]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 91)]*kernel.shared_1[229]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 91)]*kernel.shared_1[265]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 92)]*kernel.shared_1[14]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 92)]*kernel.shared_1[50]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 92)]*kernel.shared_1[86]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 92)]*kernel.shared_1[122]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 92)]*kernel.shared_1[158]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 92)]*kernel.shared_1[194]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 92)]*kernel.shared_1[230]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 92)]*kernel.shared_1[266]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[15]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[51]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[87]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[123]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[159]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[195]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[231]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[267]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[16]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[52]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[88]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[124]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[160]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[196]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[232]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[268]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[17]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[53]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[89]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[125]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[161]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[197]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[233]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[269]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 162)]*kernel.shared_1[18]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 162)]*kernel.shared_1[54]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 162)]*kernel.shared_1[90]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 162)]*kernel.shared_1[126]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 162)]*kernel.shared_1[162]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 162)]*kernel.shared_1[198]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 162)]*kernel.shared_1[234]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 162)]*kernel.shared_1[270]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 163)]*kernel.shared_1[19]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 163)]*kernel.shared_1[55]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 163)]*kernel.shared_1[91]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 163)]*kernel.shared_1[127]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 163)]*kernel.shared_1[163]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 163)]*kernel.shared_1[199]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 163)]*kernel.shared_1[235]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 163)]*kernel.shared_1[271]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 164)]*kernel.shared_1[20]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 164)]*kernel.shared_1[56]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 164)]*kernel.shared_1[92]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 164)]*kernel.shared_1[128]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 164)]*kernel.shared_1[164]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 164)]*kernel.shared_1[200]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 164)]*kernel.shared_1[236]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 164)]*kernel.shared_1[272]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 171)]*kernel.shared_1[21]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 171)]*kernel.shared_1[57]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 171)]*kernel.shared_1[93]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 171)]*kernel.shared_1[129]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 171)]*kernel.shared_1[165]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 171)]*kernel.shared_1[201]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 171)]*kernel.shared_1[237]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 171)]*kernel.shared_1[273]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 172)]*kernel.shared_1[22]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 172)]*kernel.shared_1[58]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 172)]*kernel.shared_1[94]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 172)]*kernel.shared_1[130]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 172)]*kernel.shared_1[166]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 172)]*kernel.shared_1[202]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 172)]*kernel.shared_1[238]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 172)]*kernel.shared_1[274]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 173)]*kernel.shared_1[23]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 173)]*kernel.shared_1[59]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 173)]*kernel.shared_1[95]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 173)]*kernel.shared_1[131]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 173)]*kernel.shared_1[167]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 173)]*kernel.shared_1[203]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 173)]*kernel.shared_1[239]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 173)]*kernel.shared_1[275]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[24]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[60]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[96]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[132]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[168]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[204]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[240]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[276]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[25]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[61]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[97]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[133]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[169]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[205]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[241]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[277]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[26]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[62]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[98]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[134]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[170]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[206]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[242]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[278]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 243)]*kernel.shared_1[27]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 243)]*kernel.shared_1[63]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 243)]*kernel.shared_1[99]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 243)]*kernel.shared_1[135]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 243)]*kernel.shared_1[171]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 243)]*kernel.shared_1[207]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 243)]*kernel.shared_1[243]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 243)]*kernel.shared_1[279]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 244)]*kernel.shared_1[28]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 244)]*kernel.shared_1[64]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 244)]*kernel.shared_1[100]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 244)]*kernel.shared_1[136]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 244)]*kernel.shared_1[172]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 244)]*kernel.shared_1[208]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 244)]*kernel.shared_1[244]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 244)]*kernel.shared_1[280]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 245)]*kernel.shared_1[29]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 245)]*kernel.shared_1[65]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 245)]*kernel.shared_1[101]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 245)]*kernel.shared_1[137]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 245)]*kernel.shared_1[173]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 245)]*kernel.shared_1[209]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 245)]*kernel.shared_1[245]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 245)]*kernel.shared_1[281]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[30]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[66]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[102]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[138]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[174]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[210]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[246]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[282]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[31]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[67]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[103]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[139]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[175]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[211]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[247]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[283]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[32]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[68]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[104]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[140]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[176]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[212]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[248]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[284]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[33]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[69]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[105]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[141]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[177]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[213]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[249]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[285]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[34]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[70]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[106]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[142]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[178]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[214]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[250]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[286]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[35]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[71]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[107]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[143]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[179]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[215]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[251]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[287]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[50]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[98]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[146]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[3]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[51]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[99]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[147]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[4]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[52]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[100]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[148]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[5]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[53]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[101]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[149]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[6]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[54]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[102]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[150]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[7]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[55]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[103]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[151]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[8]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[56]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[104]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[152]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[9]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[57]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[105]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[153]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[10]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[58]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[106]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[154]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[11]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[59]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[107]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[155]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[12]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[60]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[108]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[156]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[13]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[61]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[109]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[157]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[14]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[62]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[110]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[158]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 315)]*kernel.shared_1[15]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 315)]*kernel.shared_1[63]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 315)]*kernel.shared_1[111]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 315)]*kernel.shared_1[159]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 316)]*kernel.shared_1[16]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 316)]*kernel.shared_1[64]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 316)]*kernel.shared_1[112]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 316)]*kernel.shared_1[160]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 317)]*kernel.shared_1[17]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 317)]*kernel.shared_1[65]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 317)]*kernel.shared_1[113]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 317)]*kernel.shared_1[161]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 378)]*kernel.shared_1[18]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 378)]*kernel.shared_1[66]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 378)]*kernel.shared_1[114]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 378)]*kernel.shared_1[162]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 379)]*kernel.shared_1[19]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 379)]*kernel.shared_1[67]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 379)]*kernel.shared_1[115]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 379)]*kernel.shared_1[163]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 380)]*kernel.shared_1[20]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 380)]*kernel.shared_1[68]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 380)]*kernel.shared_1[116]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 380)]*kernel.shared_1[164]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 441)]*kernel.shared_1[21]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 441)]*kernel.shared_1[69]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 441)]*kernel.shared_1[117]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 441)]*kernel.shared_1[165]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 442)]*kernel.shared_1[22]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 442)]*kernel.shared_1[70]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 442)]*kernel.shared_1[118]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 442)]*kernel.shared_1[166]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 443)]*kernel.shared_1[23]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 443)]*kernel.shared_1[71]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 443)]*kernel.shared_1[119]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 443)]*kernel.shared_1[167]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 504)]*kernel.shared_1[24]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 504)]*kernel.shared_1[72]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 504)]*kernel.shared_1[120]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 504)]*kernel.shared_1[168]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 505)]*kernel.shared_1[25]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 505)]*kernel.shared_1[73]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 505)]*kernel.shared_1[121]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 505)]*kernel.shared_1[169]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 506)]*kernel.shared_1[26]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 506)]*kernel.shared_1[74]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 506)]*kernel.shared_1[122]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 506)]*kernel.shared_1[170]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 567)]*kernel.shared_1[27]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 567)]*kernel.shared_1[75]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 567)]*kernel.shared_1[123]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 567)]*kernel.shared_1[171]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 568)]*kernel.shared_1[28]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 568)]*kernel.shared_1[76]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 568)]*kernel.shared_1[124]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 568)]*kernel.shared_1[172]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 569)]*kernel.shared_1[29]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 569)]*kernel.shared_1[77]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 569)]*kernel.shared_1[125]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 569)]*kernel.shared_1[173]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 630)]*kernel.shared_1[30]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 630)]*kernel.shared_1[78]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 630)]*kernel.shared_1[126]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 630)]*kernel.shared_1[174]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 631)]*kernel.shared_1[31]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 631)]*kernel.shared_1[79]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 631)]*kernel.shared_1[127]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 631)]*kernel.shared_1[175]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 632)]*kernel.shared_1[32]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 632)]*kernel.shared_1[80]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 632)]*kernel.shared_1[128]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 632)]*kernel.shared_1[176]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 693)]*kernel.shared_1[33]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 693)]*kernel.shared_1[81]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 693)]*kernel.shared_1[129]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 693)]*kernel.shared_1[177]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 694)]*kernel.shared_1[34]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 694)]*kernel.shared_1[82]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 694)]*kernel.shared_1[130]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 694)]*kernel.shared_1[178]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 695)]*kernel.shared_1[35]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 695)]*kernel.shared_1[83]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 695)]*kernel.shared_1[131]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 695)]*kernel.shared_1[179]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 756)]*kernel.shared_1[36]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 756)]*kernel.shared_1[84]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 756)]*kernel.shared_1[132]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 756)]*kernel.shared_1[180]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 757)]*kernel.shared_1[37]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 757)]*kernel.shared_1[85]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 757)]*kernel.shared_1[133]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 757)]*kernel.shared_1[181]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 758)]*kernel.shared_1[38]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 758)]*kernel.shared_1[86]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 758)]*kernel.shared_1[134]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 758)]*kernel.shared_1[182]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 819)]*kernel.shared_1[39]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 819)]*kernel.shared_1[87]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 819)]*kernel.shared_1[135]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 819)]*kernel.shared_1[183]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 820)]*kernel.shared_1[40]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 820)]*kernel.shared_1[88]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 820)]*kernel.shared_1[136]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 820)]*kernel.shared_1[184]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 821)]*kernel.shared_1[41]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 821)]*kernel.shared_1[89]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 821)]*kernel.shared_1[137]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 821)]*kernel.shared_1[185]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 882)]*kernel.shared_1[42]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 882)]*kernel.shared_1[90]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 882)]*kernel.shared_1[138]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 882)]*kernel.shared_1[186]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 883)]*kernel.shared_1[43]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 883)]*kernel.shared_1[91]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 883)]*kernel.shared_1[139]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 883)]*kernel.shared_1[187]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 884)]*kernel.shared_1[44]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 884)]*kernel.shared_1[92]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 884)]*kernel.shared_1[140]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 884)]*kernel.shared_1[188]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 945)]*kernel.shared_1[45]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 945)]*kernel.shared_1[93]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 945)]*kernel.shared_1[141]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 945)]*kernel.shared_1[189]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 946)]*kernel.shared_1[46]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 946)]*kernel.shared_1[94]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 946)]*kernel.shared_1[142]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 946)]*kernel.shared_1[190]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 947)]*kernel.shared_1[47]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 947)]*kernel.shared_1[95]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 947)]*kernel.shared_1[143]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 947)]*kernel.shared_1[191]))
}
}
- compute[((blockIdx.x*392) + threadIdx.x)] = max((conv2d_nchw_1[0] + bias[(blockIdx.x*8)]), 0f32)
- compute[(((blockIdx.x*392) + threadIdx.x) + 49)] = max((conv2d_nchw_1[1] + bias[((blockIdx.x*8) + 1)]), 0f32)
- compute[(((blockIdx.x*392) + threadIdx.x) + 98)] = max((conv2d_nchw_1[2] + bias[((blockIdx.x*8) + 2)]), 0f32)
- compute[(((blockIdx.x*392) + threadIdx.x) + 147)] = max((conv2d_nchw_1[3] + bias[((blockIdx.x*8) + 3)]), 0f32)
- compute[(((blockIdx.x*392) + threadIdx.x) + 196)] = max((conv2d_nchw_1[4] + bias[((blockIdx.x*8) + 4)]), 0f32)
- compute[(((blockIdx.x*392) + threadIdx.x) + 245)] = max((conv2d_nchw_1[5] + bias[((blockIdx.x*8) + 5)]), 0f32)
- compute[(((blockIdx.x*392) + threadIdx.x) + 294)] = max((conv2d_nchw_1[6] + bias[((blockIdx.x*8) + 6)]), 0f32)
- compute[(((blockIdx.x*392) + threadIdx.x) + 343)] = max((conv2d_nchw_1[7] + bias[((blockIdx.x*8) + 7)]), 0f32)
+ for (i1.inner: int32, 0, 4) {
+ compute[(((blockIdx.x*196) + (i1.inner*49)) + threadIdx.x)] = max((conv2d_nchw_1[i1.inner] + bias[((blockIdx.x*4) + i1.inner)]), 0f32)
+ }
}
}
@@ -642,7 +1049,7 @@ We build the binary and check its correctness and performance.
.. code-block:: none
- Execution time of this operator: 0.223 ms
+ Execution time of this operator: 0.268 ms
@@ -690,10 +1097,10 @@ They can be used for debugging and learning the behavior of the auto-scheduler.
conv2d_nchw_nn_o_o_i, conv2d_nchw_nn_o_i = s[conv2d_nchw].split(conv2d_nchw_nn_o_i, factor=1)
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_i, conv2d_nchw_ff_i = s[conv2d_nchw].split(conv2d_nchw_ff, factor=4)
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=1)
- 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=8)
+ conv2d_nchw_ff_o_o_o_o, conv2d_nchw_ff_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_o_o_i, factor=1)
conv2d_nchw_yy_o_i, conv2d_nchw_yy_i = s[conv2d_nchw].split(conv2d_nchw_yy, factor=1)
conv2d_nchw_yy_o_o_i, conv2d_nchw_yy_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_i, factor=1)
conv2d_nchw_yy_o_o_o_i, conv2d_nchw_yy_o_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_o_i, factor=7)
@@ -702,9 +1109,9 @@ They can be used for debugging and learning the behavior of the auto-scheduler.
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_i, conv2d_nchw_rc_i = s[conv2d_nchw].split(conv2d_nchw_rc, factor=16)
conv2d_nchw_rc_o_o, conv2d_nchw_rc_o_i = s[conv2d_nchw].split(conv2d_nchw_rc_o_i, factor=1)
- conv2d_nchw_ry_o_i, conv2d_nchw_ry_i = s[conv2d_nchw].split(conv2d_nchw_ry, factor=3)
+ 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=3)
conv2d_nchw_rx_o_o, conv2d_nchw_rx_o_i = s[conv2d_nchw].split(conv2d_nchw_rx_o_i, factor=1)
@@ -712,9 +1119,9 @@ They can be used for debugging and learning the behavior of the auto-scheduler.
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_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=1)
- compute_i1_o_o_o, compute_i1_o_o_i = s[compute].split(compute_i1_o_o_i, factor=8)
+ compute_i1_o_o_o, compute_i1_o_o_i = s[compute].split(compute_i1_o_o_i, factor=1)
compute_i2_o_i, compute_i2_i = s[compute].split(compute_i2, factor=1)
compute_i2_o_o_i, compute_i2_o_i = s[compute].split(compute_i2_o_i, factor=7)
compute_i2_o_o_o, compute_i2_o_o_i = s[compute].split(compute_i2_o_o_i, factor=1)
@@ -742,11 +1149,11 @@ They can be used for debugging and learning the behavior of the auto-scheduler.
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=49)
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=2)
+ 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=49)
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", 512)
+ s[conv2d_nchw].pragma(conv2d_nchw_nn_o_o_o_o, "auto_unroll_max_step", 1024)
s[conv2d_nchw].pragma(conv2d_nchw_nn_o_o_o_o, "unroll_explicit", True)
CUDA source code:
@@ -765,337 +1172,687 @@ They can be used for debugging and learning the behavior of the auto-scheduler.
#define uint64_t unsigned long long
#endif
extern "C" __global__ void __launch_bounds__(49) default_function_kernel0(float* __restrict__ data, float* __restrict__ kernel, float* __restrict__ compute, float* __restrict__ bias) {
- float conv2d_nchw[8];
- __shared__ float pad_temp_shared[324];
- __shared__ float kernel_shared[288];
+ float conv2d_nchw[4];
+ __shared__ float pad_temp_shared[1008];
+ __shared__ float kernel_shared[192];
conv2d_nchw[0] = 0.000000e+00f;
conv2d_nchw[1] = 0.000000e+00f;
conv2d_nchw[2] = 0.000000e+00f;
conv2d_nchw[3] = 0.000000e+00f;
- conv2d_nchw[4] = 0.000000e+00f;
- conv2d_nchw[5] = 0.000000e+00f;
- conv2d_nchw[6] = 0.000000e+00f;
- conv2d_nchw[7] = 0.000000e+00f;
- for (int rc_outer_outer = 0; rc_outer_outer < 128; ++rc_outer_outer) {
+ for (int rc_outer_outer = 0; rc_outer_outer < 32; ++rc_outer_outer) {
+ __syncthreads();
+ pad_temp_shared[((int)threadIdx.x)] = ((((9 <= ((int)threadIdx.x)) && (1 <= (((int)threadIdx.x) % 9))) && ((((int)threadIdx.x) % 9) < 8)) ? data[((((rc_outer_outer * 784) + ((((int)threadIdx.x) / 9) * 7)) + (((int)threadIdx.x) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 49)] = ((((9 <= ((((int)threadIdx.x) + 49) % 63)) && (1 <= ((((int)threadIdx.x) + 4) % 9))) && (((((int)threadIdx.x) + 4) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 49) / 9) * 7)) + ((((int)threadIdx.x) + 4) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 98)] = ((((9 <= ((((int)threadIdx.x) + 35) % 63)) && (1 <= ((((int)threadIdx.x) + 8) % 9))) && (((((int)threadIdx.x) + 8) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 98) / 9) * 7)) + ((((int)threadIdx.x) + 8) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 147)] = ((((9 <= ((((int)threadIdx.x) + 21) % 63)) && (1 <= ((((int)threadIdx.x) + 3) % 9))) && (((((int)threadIdx.x) + 3) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 147) / 9) * 7)) + ((((int)threadIdx.x) + 3) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 196)] = ((((2 <= ((int)threadIdx.x)) && (1 <= ((((int)threadIdx.x) + 7) % 9))) && (((((int)threadIdx.x) + 7) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 196) / 9) * 7)) + ((((int)threadIdx.x) + 7) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 245)] = ((((9 <= ((((int)threadIdx.x) + 56) % 63)) && (1 <= ((((int)threadIdx.x) + 2) % 9))) && (((((int)threadIdx.x) + 2) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 245) / 9) * 7)) + ((((int)threadIdx.x) + 2) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 294)] = ((((9 <= ((((int)threadIdx.x) + 42) % 63)) && (1 <= ((((int)threadIdx.x) + 6) % 9))) && (((((int)threadIdx.x) + 6) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 294) / 9) * 7)) + ((((int)threadIdx.x) + 6) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 343)] = ((((9 <= ((((int)threadIdx.x) + 28) % 63)) && (1 <= ((((int)threadIdx.x) + 1) % 9))) && (((((int)threadIdx.x) + 1) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 343) / 9) * 7)) + ((((int)threadIdx.x) + 1) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 392)] = (((1 <= ((((int)threadIdx.x) + 5) % 9)) && (((((int)threadIdx.x) + 5) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 392) / 9) * 7)) + ((((int)threadIdx.x) + 5) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 441)] = ((((9 <= ((int)threadIdx.x)) && (1 <= (((int)threadIdx.x) % 9))) && ((((int)threadIdx.x) % 9) < 8)) ? data[((((rc_outer_outer * 784) + ((((int)threadIdx.x) / 9) * 7)) + (((int)threadIdx.x) % 9)) + 335)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 490)] = ((((9 <= ((((int)threadIdx.x) + 49) % 63)) && (1 <= ((((int)threadIdx.x) + 4) % 9))) && (((((int)threadIdx.x) + 4) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 490) / 9) * 7)) + ((((int)threadIdx.x) + 4) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 539)] = ((((9 <= ((((int)threadIdx.x) + 35) % 63)) && (1 <= ((((int)threadIdx.x) + 8) % 9))) && (((((int)threadIdx.x) + 8) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 539) / 9) * 7)) + ((((int)threadIdx.x) + 8) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 588)] = ((((9 <= ((((int)threadIdx.x) + 21) % 63)) && (1 <= ((((int)threadIdx.x) + 3) % 9))) && (((((int)threadIdx.x) + 3) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 588) / 9) * 7)) + ((((int)threadIdx.x) + 3) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 637)] = ((((2 <= ((int)threadIdx.x)) && (1 <= ((((int)threadIdx.x) + 7) % 9))) && (((((int)threadIdx.x) + 7) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 637) / 9) * 7)) + ((((int)threadIdx.x) + 7) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 686)] = ((((9 <= ((((int)threadIdx.x) + 56) % 63)) && (1 <= ((((int)threadIdx.x) + 2) % 9))) && (((((int)threadIdx.x) + 2) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 686) / 9) * 7)) + ((((int)threadIdx.x) + 2) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 735)] = ((((9 <= ((((int)threadIdx.x) + 42) % 63)) && (1 <= ((((int)threadIdx.x) + 6) % 9))) && (((((int)threadIdx.x) + 6) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 735) / 9) * 7)) + ((((int)threadIdx.x) + 6) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 784)] = ((((9 <= ((((int)threadIdx.x) + 28) % 63)) && (1 <= ((((int)threadIdx.x) + 1) % 9))) && (((((int)threadIdx.x) + 1) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 784) / 9) * 7)) + ((((int)threadIdx.x) + 1) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 833)] = (((1 <= ((((int)threadIdx.x) + 5) % 9)) && (((((int)threadIdx.x) + 5) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 833) / 9) * 7)) + ((((int)threadIdx.x) + 5) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 882)] = ((((9 <= ((int)threadIdx.x)) && (1 <= (((int)threadIdx.x) % 9))) && ((((int)threadIdx.x) % 9) < 8)) ? data[((((rc_outer_outer * 784) + ((((int)threadIdx.x) / 9) * 7)) + (((int)threadIdx.x) % 9)) + 678)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 931)] = ((((9 <= ((((int)threadIdx.x) + 49) % 63)) && (1 <= ((((int)threadIdx.x) + 4) % 9))) && (((((int)threadIdx.x) + 4) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 931) / 9) * 7)) + ((((int)threadIdx.x) + 4) % 9)) - 8)] : 0.000000e+00f);
+ if (((int)threadIdx.x) < 28) {
+ pad_temp_shared[(((int)threadIdx.x) + 980)] = (((1 <= ((((int)threadIdx.x) + 8) % 9)) && (((((int)threadIdx.x) + 8) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 980) / 9) * 7)) + ((((int)threadIdx.x) + 8) % 9)) - 8)] : 0.000000e+00f);
+ }
+ kernel_shared[((int)threadIdx.x)] = kernel[(((((((int)blockIdx.x) * 18432) + ((((int)threadIdx.x) / 48) * 4608)) + (rc_outer_outer * 144)) + (((((int)threadIdx.x) % 48) / 3) * 9)) + (((int)threadIdx.x) % 3))];
+ kernel_shared[(((int)threadIdx.x) + 49)] = kernel[(((((((int)blockIdx.x) * 18432) + (((((int)threadIdx.x) + 49) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 1) % 48) / 3) * 9)) + ((((int)threadIdx.x) + 1) % 3))];
+ kernel_shared[(((int)threadIdx.x) + 98)] = kernel[(((((((int)blockIdx.x) * 18432) + (((((int)threadIdx.x) + 98) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 2) % 48) / 3) * 9)) + ((((int)threadIdx.x) + 2) % 3))];
+ if (((int)threadIdx.x) < 45) {
+ kernel_shared[(((int)threadIdx.x) + 147)] = kernel[((((((((int)blockIdx.x) * 18432) + (((((int)threadIdx.x) + 147) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((int)threadIdx.x) / 3) * 9)) + (((int)threadIdx.x) % 3)) + 9)];
+ }
+ __syncthreads();
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[0]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[48]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[96]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[144]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[1]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[49]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[97]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[145]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[2]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[50]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[98]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[146]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[3]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[51]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[99]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[147]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[4]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[52]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[100]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[148]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[5]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[53]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[101]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[149]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[6]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[54]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[102]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[150]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[7]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[55]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[103]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[151]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[8]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[56]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[104]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[152]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[9]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[57]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[105]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[153]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[10]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[58]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[106]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[154]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[11]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[59]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[107]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[155]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[12]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[60]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[108]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[156]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[13]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[61]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[109]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[157]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[14]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[62]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[110]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[158]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[15]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[63]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[111]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[159]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 316)] * kernel_shared[16]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 316)] * kernel_shared[64]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 316)] * kernel_shared[112]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 316)] * kernel_shared[160]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 317)] * kernel_shared[17]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 317)] * kernel_shared[65]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 317)] * kernel_shared[113]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 317)] * kernel_shared[161]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[18]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[66]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[114]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[162]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 379)] * kernel_shared[19]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 379)] * kernel_shared[67]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 379)] * kernel_shared[115]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 379)] * kernel_shared[163]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 380)] * kernel_shared[20]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 380)] * kernel_shared[68]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 380)] * kernel_shared[116]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 380)] * kernel_shared[164]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[21]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[69]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[117]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[165]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 442)] * kernel_shared[22]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 442)] * kernel_shared[70]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 442)] * kernel_shared[118]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 442)] * kernel_shared[166]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 443)] * kernel_shared[23]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 443)] * kernel_shared[71]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 443)] * kernel_shared[119]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 443)] * kernel_shared[167]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 504)] * kernel_shared[24]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 504)] * kernel_shared[72]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 504)] * kernel_shared[120]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 504)] * kernel_shared[168]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 505)] * kernel_shared[25]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 505)] * kernel_shared[73]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 505)] * kernel_shared[121]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 505)] * kernel_shared[169]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 506)] * kernel_shared[26]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 506)] * kernel_shared[74]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 506)] * kernel_shared[122]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 506)] * kernel_shared[170]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 567)] * kernel_shared[27]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 567)] * kernel_shared[75]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 567)] * kernel_shared[123]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 567)] * kernel_shared[171]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 568)] * kernel_shared[28]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 568)] * kernel_shared[76]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 568)] * kernel_shared[124]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 568)] * kernel_shared[172]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 569)] * kernel_shared[29]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 569)] * kernel_shared[77]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 569)] * kernel_shared[125]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 569)] * kernel_shared[173]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 630)] * kernel_shared[30]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 630)] * kernel_shared[78]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 630)] * kernel_shared[126]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 630)] * kernel_shared[174]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 631)] * kernel_shared[31]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 631)] * kernel_shared[79]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 631)] * kernel_shared[127]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 631)] * kernel_shared[175]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 632)] * kernel_shared[32]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 632)] * kernel_shared[80]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 632)] * kernel_shared[128]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 632)] * kernel_shared[176]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 693)] * kernel_shared[33]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 693)] * kernel_shared[81]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 693)] * kernel_shared[129]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 693)] * kernel_shared[177]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 694)] * kernel_shared[34]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 694)] * kernel_shared[82]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 694)] * kernel_shared[130]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 694)] * kernel_shared[178]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 695)] * kernel_shared[35]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 695)] * kernel_shared[83]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 695)] * kernel_shared[131]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 695)] * kernel_shared[179]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 756)] * kernel_shared[36]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 756)] * kernel_shared[84]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 756)] * kernel_shared[132]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 756)] * kernel_shared[180]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 757)] * kernel_shared[37]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 757)] * kernel_shared[85]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 757)] * kernel_shared[133]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 757)] * kernel_shared[181]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 758)] * kernel_shared[38]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 758)] * kernel_shared[86]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 758)] * kernel_shared[134]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 758)] * kernel_shared[182]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 819)] * kernel_shared[39]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 819)] * kernel_shared[87]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 819)] * kernel_shared[135]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 819)] * kernel_shared[183]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 820)] * kernel_shared[40]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 820)] * kernel_shared[88]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 820)] * kernel_shared[136]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 820)] * kernel_shared[184]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 821)] * kernel_shared[41]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 821)] * kernel_shared[89]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 821)] * kernel_shared[137]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 821)] * kernel_shared[185]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 882)] * kernel_shared[42]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 882)] * kernel_shared[90]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 882)] * kernel_shared[138]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 882)] * kernel_shared[186]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 883)] * kernel_shared[43]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 883)] * kernel_shared[91]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 883)] * kernel_shared[139]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 883)] * kernel_shared[187]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 884)] * kernel_shared[44]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 884)] * kernel_shared[92]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 884)] * kernel_shared[140]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 884)] * kernel_shared[188]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 945)] * kernel_shared[45]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 945)] * kernel_shared[93]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 945)] * kernel_shared[141]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 945)] * kernel_shared[189]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 946)] * kernel_shared[46]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 946)] * kernel_shared[94]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 946)] * kernel_shared[142]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 946)] * kernel_shared[190]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 947)] * kernel_shared[47]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 947)] * kernel_shared[95]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 947)] * kernel_shared[143]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 947)] * kernel_shared[191]));
__syncthreads();
- pad_temp_shared[(((int)threadIdx.x) * 2)] = (((((9 <= ((((int)threadIdx.x) * 2) % 81)) && (((((int)threadIdx.x) * 2) % 81) < 72)) && (1 <= ((((int)threadIdx.x) * 2) % 9))) && (((((int)threadIdx.x) * 2) % 9) < 8)) ? data[(((((rc_outer_outer * 196) + (((((int)threadIdx.x) * 2) / 81) * 49)) + ((((((int)threadIdx.x) * 2) % 81) / 9) * 7)) + ((((int)threadIdx.x) * 2) % 9)) - 8)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 2) + 1)] = (((((9 <= (((((int)threadIdx.x) * 2) + 1) % 81)) && ((((((int)threadIdx.x) * 2) + 1) % 81) < 72)) && (1 <= (((((int)threadIdx.x) * 2) + 1) % 9))) && ((((((int)threadIdx.x) * 2) + 1) % 9) < 8)) ? data[(((((rc_outer_outer * 196) + ((((((int)threadIdx.x) * 2) + 1) / 81) * 49)) + (((((((int)threadIdx.x) * 2) + 1) % 81) / 9) * 7)) + (((((int)threadIdx.x) * 2) + 1) % 9)) - 8)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 2) + 98)] = (((((9 <= (((((int)threadIdx.x) * 2) + 17) % 81)) && ((((((int)threadIdx.x) * 2) + 17) % 81) < 72)) && (1 <= (((((int)threadIdx.x) * 2) + 8) % 9))) && ((((((int)threadIdx.x) * 2) + 8) % 9) < 8)) ? data[(((((rc_outer_outer * 196) + ((((((int)threadIdx.x) * 2) + 98) / 81) * 49)) + (((((((int)threadIdx.x) * 2) + 17) % 81) / 9) * 7)) + (((((int)threadIdx.x) * 2) + 8) % 9)) - 8)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 2) + 99)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 9) + 2) % 9)) && ((((((int)threadIdx.x) * 2) + 18) % 81) < 72)) && (1 <= ((((int)threadIdx.x) * 2) % 9))) && (((((int)threadIdx.x) * 2) % 9) < 8)) ? data[(((((rc_outer_outer * 196) + ((((((int)threadIdx.x) * 2) + 99) / 81) * 49)) + (((((((int)threadIdx.x) * 2) / 9) + 2) % 9) * 7)) + ((((int)threadIdx.x) * 2) % 9)) - 8)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 2) + 196)] = (((((9 <= (((((int)threadIdx.x) * 2) + 34) % 81)) && ((((((int)threadIdx.x) * 2) + 34) % 81) < 72)) && (1 <= (((((int)threadIdx.x) * 2) + 7) % 9))) && ((((((int)threadIdx.x) * 2) + 7) % 9) < 8)) ? data[(((((rc_outer_outer * 196) + ((((((int)threadIdx.x) * 2) + 196) / 81) * 49)) + (((((((int)threadIdx.x) * 2) + 34) % 81) / 9) * 7)) + (((((int)threadIdx.x) * 2) + 7) % 9)) - 8)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 2) + 197)] = (((((9 <= (((((int)threadIdx.x) * 2) + 35) % 81)) && ((((((int)threadIdx.x) * 2) + 35) % 81) < 72)) && (1 <= (((((int)threadIdx.x) * 2) + 8) % 9))) && ((((((int)threadIdx.x) * 2) + 8) % 9) < 8)) ? data[(((((rc_outer_outer * 196) + ((((((int)threadIdx.x) * 2) + 197) / 81) * 49)) + (((((((int)threadIdx.x) * 2) + 35) % 81) / 9) * 7)) + (((((int)threadIdx.x) * 2) + 8) % 9)) - 8)] : 0.000000e+00f);
- if (((int)threadIdx.x) < 15) {
- pad_temp_shared[((((int)threadIdx.x) * 2) + 294)] = ((((((int)threadIdx.x) < 11) && (1 <= (((((int)threadIdx.x) * 2) + 6) % 9))) && ((((((int)threadIdx.x) * 2) + 6) % 9) < 8)) ? data[((((rc_outer_outer * 196) + ((((((int)threadIdx.x) * 2) + 51) / 9) * 7)) + (((((int)threadIdx.x) * 2) + 6) % 9)) + 139)] : 0.000000e+00f);
+ pad_temp_shared[((int)threadIdx.x)] = (((1 <= (((int)threadIdx.x) % 9)) && ((((int)threadIdx.x) % 9) < 8)) ? data[((((rc_outer_outer * 784) + ((((int)threadIdx.x) / 9) * 7)) + (((int)threadIdx.x) % 9)) - 1)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 49)] = (((1 <= ((((int)threadIdx.x) + 4) % 9)) && (((((int)threadIdx.x) + 4) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 49) / 9) * 7)) + ((((int)threadIdx.x) + 4) % 9)) - 1)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 98)] = (((1 <= ((((int)threadIdx.x) + 8) % 9)) && (((((int)threadIdx.x) + 8) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 98) / 9) * 7)) + ((((int)threadIdx.x) + 8) % 9)) - 1)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 147)] = (((1 <= ((((int)threadIdx.x) + 3) % 9)) && (((((int)threadIdx.x) + 3) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 147) / 9) * 7)) + ((((int)threadIdx.x) + 3) % 9)) - 1)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 196)] = (((1 <= ((((int)threadIdx.x) + 7) % 9)) && (((((int)threadIdx.x) + 7) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 196) / 9) * 7)) + ((((int)threadIdx.x) + 7) % 9)) - 1)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 245)] = (((1 <= ((((int)threadIdx.x) + 2) % 9)) && (((((int)threadIdx.x) + 2) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 245) / 9) * 7)) + ((((int)threadIdx.x) + 2) % 9)) - 1)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 294)] = (((1 <= ((((int)threadIdx.x) + 6) % 9)) && (((((int)threadIdx.x) + 6) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 294) / 9) * 7)) + ((((int)threadIdx.x) + 6) % 9)) - 1)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 343)] = (((1 <= ((((int)threadIdx.x) + 1) % 9)) && (((((int)threadIdx.x) + 1) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 343) / 9) * 7)) + ((((int)threadIdx.x) + 1) % 9)) - 1)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 392)] = (((1 <= ((((int)threadIdx.x) + 5) % 9)) && (((((int)threadIdx.x) + 5) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 392) / 9) * 7)) + ((((int)threadIdx.x) + 5) % 9)) - 1)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 441)] = (((1 <= (((int)threadIdx.x) % 9)) && ((((int)threadIdx.x) % 9) < 8)) ? data[((((rc_outer_outer * 784) + ((((int)threadIdx.x) / 9) * 7)) + (((int)threadIdx.x) % 9)) + 342)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 490)] = (((1 <= ((((int)threadIdx.x) + 4) % 9)) && (((((int)threadIdx.x) + 4) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 490) / 9) * 7)) + ((((int)threadIdx.x) + 4) % 9)) - 1)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 539)] = (((1 <= ((((int)threadIdx.x) + 8) % 9)) && (((((int)threadIdx.x) + 8) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 539) / 9) * 7)) + ((((int)threadIdx.x) + 8) % 9)) - 1)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 588)] = (((1 <= ((((int)threadIdx.x) + 3) % 9)) && (((((int)threadIdx.x) + 3) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 588) / 9) * 7)) + ((((int)threadIdx.x) + 3) % 9)) - 1)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 637)] = (((1 <= ((((int)threadIdx.x) + 7) % 9)) && (((((int)threadIdx.x) + 7) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 637) / 9) * 7)) + ((((int)threadIdx.x) + 7) % 9)) - 1)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 686)] = (((1 <= ((((int)threadIdx.x) + 2) % 9)) && (((((int)threadIdx.x) + 2) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 686) / 9) * 7)) + ((((int)threadIdx.x) + 2) % 9)) - 1)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 735)] = (((1 <= ((((int)threadIdx.x) + 6) % 9)) && (((((int)threadIdx.x) + 6) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 735) / 9) * 7)) + ((((int)threadIdx.x) + 6) % 9)) - 1)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 784)] = (((1 <= ((((int)threadIdx.x) + 1) % 9)) && (((((int)threadIdx.x) + 1) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 784) / 9) * 7)) + ((((int)threadIdx.x) + 1) % 9)) - 1)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 833)] = (((1 <= ((((int)threadIdx.x) + 5) % 9)) && (((((int)threadIdx.x) + 5) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 833) / 9) * 7)) + ((((int)threadIdx.x) + 5) % 9)) - 1)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 882)] = (((1 <= (((int)threadIdx.x) % 9)) && ((((int)threadIdx.x) % 9) < 8)) ? data[((((rc_outer_outer * 784) + ((((int)threadIdx.x) / 9) * 7)) + (((int)threadIdx.x) % 9)) + 685)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 931)] = (((1 <= ((((int)threadIdx.x) + 4) % 9)) && (((((int)threadIdx.x) + 4) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 931) / 9) * 7)) + ((((int)threadIdx.x) + 4) % 9)) - 1)] : 0.000000e+00f);
+ if (((int)threadIdx.x) < 28) {
+ pad_temp_shared[(((int)threadIdx.x) + 980)] = (((1 <= ((((int)threadIdx.x) + 8) % 9)) && (((((int)threadIdx.x) + 8) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 980) / 9) * 7)) + ((((int)threadIdx.x) + 8) % 9)) - 1)] : 0.000000e+00f);
+ }
+ kernel_shared[((int)threadIdx.x)] = kernel[((((((((int)blockIdx.x) * 18432) + ((((int)threadIdx.x) / 48) * 4608)) + (rc_outer_outer * 144)) + (((((int)threadIdx.x) % 48) / 3) * 9)) + (((int)threadIdx.x) % 3)) + 3)];
+ kernel_shared[(((int)threadIdx.x) + 49)] = kernel[((((((((int)blockIdx.x) * 18432) + (((((int)threadIdx.x) + 49) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 1) % 48) / 3) * 9)) + ((((int)threadIdx.x) + 1) % 3)) + 3)];
+ kernel_shared[(((int)threadIdx.x) + 98)] = kernel[((((((((int)blockIdx.x) * 18432) + (((((int)threadIdx.x) + 98) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 2) % 48) / 3) * 9)) + ((((int)threadIdx.x) + 2) % 3)) + 3)];
+ if (((int)threadIdx.x) < 45) {
+ kernel_shared[(((int)threadIdx.x) + 147)] = kernel[((((((((int)blockIdx.x) * 18432) + (((((int)threadIdx.x) + 147) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((int)threadIdx.x) / 3) * 9)) + (((int)threadIdx.x) % 3)) + 12)];
}
- if (((int)threadIdx.x) < 15) {
- pad_temp_shared[((((int)threadIdx.x) * 2) + 295)] = ((((((int)threadIdx.x) < 10) && (1 <= (((((int)threadIdx.x) * 2) + 7) % 9))) && ((((((int)threadIdx.x) * 2) + 7) % 9) < 8)) ? data[((((rc_outer_outer * 196) + ((((((int)threadIdx.x) * 2) + 52) / 9) * 7)) + (((((int)threadIdx.x) * 2) + 7) % 9)) + 139)] : 0.000000e+00f);
+ __syncthreads();
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[0]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[48]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[96]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[144]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[1]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[49]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[97]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[145]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[2]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[50]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[98]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[146]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[3]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[51]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[99]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[147]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[4]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[52]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[100]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[148]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[5]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[53]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[101]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[149]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[6]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[54]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[102]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[150]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[7]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[55]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[103]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[151]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[8]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[56]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[104]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[152]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[9]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[57]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[105]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[153]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[10]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[58]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[106]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[154]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[11]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[59]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[107]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[155]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[12]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[60]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[108]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[156]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[13]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[61]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[109]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[157]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[14]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[62]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[110]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[158]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[15]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[63]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[111]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[159]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 316)] * kernel_shared[16]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 316)] * kernel_shared[64]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 316)] * kernel_shared[112]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 316)] * kernel_shared[160]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 317)] * kernel_shared[17]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 317)] * kernel_shared[65]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 317)] * kernel_shared[113]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 317)] * kernel_shared[161]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[18]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[66]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[114]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[162]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 379)] * kernel_shared[19]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 379)] * kernel_shared[67]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 379)] * kernel_shared[115]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 379)] * kernel_shared[163]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 380)] * kernel_shared[20]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 380)] * kernel_shared[68]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 380)] * kernel_shared[116]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 380)] * kernel_shared[164]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[21]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[69]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[117]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[165]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 442)] * kernel_shared[22]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 442)] * kernel_shared[70]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 442)] * kernel_shared[118]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 442)] * kernel_shared[166]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 443)] * kernel_shared[23]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 443)] * kernel_shared[71]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 443)] * kernel_shared[119]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 443)] * kernel_shared[167]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 504)] * kernel_shared[24]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 504)] * kernel_shared[72]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 504)] * kernel_shared[120]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 504)] * kernel_shared[168]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 505)] * kernel_shared[25]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 505)] * kernel_shared[73]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 505)] * kernel_shared[121]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 505)] * kernel_shared[169]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 506)] * kernel_shared[26]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 506)] * kernel_shared[74]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 506)] * kernel_shared[122]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 506)] * kernel_shared[170]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 567)] * kernel_shared[27]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 567)] * kernel_shared[75]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 567)] * kernel_shared[123]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 567)] * kernel_shared[171]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 568)] * kernel_shared[28]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 568)] * kernel_shared[76]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 568)] * kernel_shared[124]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 568)] * kernel_shared[172]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 569)] * kernel_shared[29]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 569)] * kernel_shared[77]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 569)] * kernel_shared[125]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 569)] * kernel_shared[173]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 630)] * kernel_shared[30]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 630)] * kernel_shared[78]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 630)] * kernel_shared[126]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 630)] * kernel_shared[174]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 631)] * kernel_shared[31]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 631)] * kernel_shared[79]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 631)] * kernel_shared[127]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 631)] * kernel_shared[175]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 632)] * kernel_shared[32]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 632)] * kernel_shared[80]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 632)] * kernel_shared[128]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 632)] * kernel_shared[176]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 693)] * kernel_shared[33]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 693)] * kernel_shared[81]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 693)] * kernel_shared[129]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 693)] * kernel_shared[177]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 694)] * kernel_shared[34]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 694)] * kernel_shared[82]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 694)] * kernel_shared[130]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 694)] * kernel_shared[178]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 695)] * kernel_shared[35]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 695)] * kernel_shared[83]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 695)] * kernel_shared[131]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 695)] * kernel_shared[179]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 756)] * kernel_shared[36]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 756)] * kernel_shared[84]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 756)] * kernel_shared[132]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 756)] * kernel_shared[180]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 757)] * kernel_shared[37]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 757)] * kernel_shared[85]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 757)] * kernel_shared[133]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 757)] * kernel_shared[181]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 758)] * kernel_shared[38]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 758)] * kernel_shared[86]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 758)] * kernel_shared[134]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 758)] * kernel_shared[182]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 819)] * kernel_shared[39]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 819)] * kernel_shared[87]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 819)] * kernel_shared[135]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 819)] * kernel_shared[183]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 820)] * kernel_shared[40]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 820)] * kernel_shared[88]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 820)] * kernel_shared[136]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 820)] * kernel_shared[184]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 821)] * kernel_shared[41]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 821)] * kernel_shared[89]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 821)] * kernel_shared[137]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 821)] * kernel_shared[185]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 882)] * kernel_shared[42]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 882)] * kernel_shared[90]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 882)] * kernel_shared[138]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 882)] * kernel_shared[186]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 883)] * kernel_shared[43]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 883)] * kernel_shared[91]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 883)] * kernel_shared[139]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 883)] * kernel_shared[187]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 884)] * kernel_shared[44]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 884)] * kernel_shared[92]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 884)] * kernel_shared[140]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 884)] * kernel_shared[188]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 945)] * kernel_shared[45]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 945)] * kernel_shared[93]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 945)] * kernel_shared[141]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 945)] * kernel_shared[189]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 946)] * kernel_shared[46]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 946)] * kernel_shared[94]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 946)] * kernel_shared[142]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 946)] * kernel_shared[190]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 947)] * kernel_shared[47]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 947)] * kernel_shared[95]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 947)] * kernel_shared[143]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 947)] * kernel_shared[191]));
+ __syncthreads();
+ pad_temp_shared[((int)threadIdx.x)] = (((1 <= (((int)threadIdx.x) % 9)) && ((((int)threadIdx.x) % 9) < 8)) ? data[((((rc_outer_outer * 784) + ((((int)threadIdx.x) / 9) * 7)) + (((int)threadIdx.x) % 9)) + 6)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 49)] = ((((((((int)threadIdx.x) + 49) % 63) < 54) && (1 <= ((((int)threadIdx.x) + 4) % 9))) && (((((int)threadIdx.x) + 4) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 49) / 9) * 7)) + ((((int)threadIdx.x) + 4) % 9)) + 6)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 98)] = ((((((((int)threadIdx.x) + 35) % 63) < 54) && (1 <= ((((int)threadIdx.x) + 8) % 9))) && (((((int)threadIdx.x) + 8) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 98) / 9) * 7)) + ((((int)threadIdx.x) + 8) % 9)) + 6)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 147)] = ((((((((int)threadIdx.x) + 21) % 63) < 54) && (1 <= ((((int)threadIdx.x) + 3) % 9))) && (((((int)threadIdx.x) + 3) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 147) / 9) * 7)) + ((((int)threadIdx.x) + 3) % 9)) + 6)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 196)] = ((((((int)threadIdx.x) < 47) && (1 <= ((((int)threadIdx.x) + 7) % 9))) && (((((int)threadIdx.x) + 7) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 196) / 9) * 7)) + ((((int)threadIdx.x) + 7) % 9)) + 6)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 245)] = ((((((((int)threadIdx.x) + 56) % 63) < 54) && (1 <= ((((int)threadIdx.x) + 2) % 9))) && (((((int)threadIdx.x) + 2) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 245) / 9) * 7)) + ((((int)threadIdx.x) + 2) % 9)) + 6)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 294)] = ((((((((int)threadIdx.x) + 42) % 63) < 54) && (1 <= ((((int)threadIdx.x) + 6) % 9))) && (((((int)threadIdx.x) + 6) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 294) / 9) * 7)) + ((((int)threadIdx.x) + 6) % 9)) + 6)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 343)] = ((((((((int)threadIdx.x) + 28) % 63) < 54) && (1 <= ((((int)threadIdx.x) + 1) % 9))) && (((((int)threadIdx.x) + 1) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 343) / 9) * 7)) + ((((int)threadIdx.x) + 1) % 9)) + 6)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 392)] = ((((((int)threadIdx.x) < 40) && (1 <= ((((int)threadIdx.x) + 5) % 9))) && (((((int)threadIdx.x) + 5) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 392) / 9) * 7)) + ((((int)threadIdx.x) + 5) % 9)) + 6)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 441)] = (((1 <= (((int)threadIdx.x) % 9)) && ((((int)threadIdx.x) % 9) < 8)) ? data[((((rc_outer_outer * 784) + ((((int)threadIdx.x) / 9) * 7)) + (((int)threadIdx.x) % 9)) + 349)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 490)] = ((((((((int)threadIdx.x) + 49) % 63) < 54) && (1 <= ((((int)threadIdx.x) + 4) % 9))) && (((((int)threadIdx.x) + 4) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 490) / 9) * 7)) + ((((int)threadIdx.x) + 4) % 9)) + 6)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 539)] = ((((((((int)threadIdx.x) + 35) % 63) < 54) && (1 <= ((((int)threadIdx.x) + 8) % 9))) && (((((int)threadIdx.x) + 8) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 539) / 9) * 7)) + ((((int)threadIdx.x) + 8) % 9)) + 6)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 588)] = ((((((((int)threadIdx.x) + 21) % 63) < 54) && (1 <= ((((int)threadIdx.x) + 3) % 9))) && (((((int)threadIdx.x) + 3) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 588) / 9) * 7)) + ((((int)threadIdx.x) + 3) % 9)) + 6)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 637)] = ((((((int)threadIdx.x) < 47) && (1 <= ((((int)threadIdx.x) + 7) % 9))) && (((((int)threadIdx.x) + 7) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 637) / 9) * 7)) + ((((int)threadIdx.x) + 7) % 9)) + 6)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 686)] = ((((((((int)threadIdx.x) + 56) % 63) < 54) && (1 <= ((((int)threadIdx.x) + 2) % 9))) && (((((int)threadIdx.x) + 2) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 686) / 9) * 7)) + ((((int)threadIdx.x) + 2) % 9)) + 6)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 735)] = ((((((((int)threadIdx.x) + 42) % 63) < 54) && (1 <= ((((int)threadIdx.x) + 6) % 9))) && (((((int)threadIdx.x) + 6) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 735) / 9) * 7)) + ((((int)threadIdx.x) + 6) % 9)) + 6)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 784)] = ((((((((int)threadIdx.x) + 28) % 63) < 54) && (1 <= ((((int)threadIdx.x) + 1) % 9))) && (((((int)threadIdx.x) + 1) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 784) / 9) * 7)) + ((((int)threadIdx.x) + 1) % 9)) + 6)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 833)] = ((((((int)threadIdx.x) < 40) && (1 <= ((((int)threadIdx.x) + 5) % 9))) && (((((int)threadIdx.x) + 5) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 833) / 9) * 7)) + ((((int)threadIdx.x) + 5) % 9)) + 6)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 882)] = (((1 <= (((int)threadIdx.x) % 9)) && ((((int)threadIdx.x) % 9) < 8)) ? data[((((rc_outer_outer * 784) + ((((int)threadIdx.x) / 9) * 7)) + (((int)threadIdx.x) % 9)) + 692)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 931)] = ((((((((int)threadIdx.x) + 49) % 63) < 54) && (1 <= ((((int)threadIdx.x) + 4) % 9))) && (((((int)threadIdx.x) + 4) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 931) / 9) * 7)) + ((((int)threadIdx.x) + 4) % 9)) + 6)] : 0.000000e+00f);
+ if (((int)threadIdx.x) < 28) {
+ pad_temp_shared[(((int)threadIdx.x) + 980)] = ((((((int)threadIdx.x) < 19) && (1 <= ((((int)threadIdx.x) + 8) % 9))) && (((((int)threadIdx.x) + 8) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 980) / 9) * 7)) + ((((int)threadIdx.x) + 8) % 9)) + 6)] : 0.000000e+00f);
}
- kernel_shared[((int)threadIdx.x)] = kernel[((((((int)blockIdx.x) * 36864) + ((((int)threadIdx.x) / 36) * 4608)) + (rc_outer_outer * 36)) + (((int)threadIdx.x) % 36))];
- kernel_shared[(((int)threadIdx.x) + 49)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 49) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 13) % 36) / 9) * 9)) + ((((((int)threadIdx.x) + 4) % 9) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
- kernel_shared[(((int)threadIdx.x) + 98)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 98) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 26) % 36) / 9) * 9)) + ((((((int)threadIdx.x) + 8) % 9) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
- kernel_shared[(((int)threadIdx.x) + 147)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 147) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 3) % 36) / 9) * 9)) + ((((((int)threadIdx.x) / 3) + 1) % 3) * 3)) + (((int)threadIdx.x) % 3))];
- kernel_shared[(((int)threadIdx.x) + 196)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 196) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 16) % 36) / 9) * 9)) + ((((((int)threadIdx.x) + 7) % 9) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
- if (((int)threadIdx.x) < 43) {
- kernel_shared[(((int)threadIdx.x) + 245)] = kernel[(((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 245) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 29) % 36) / 9) * 9)) + ((((int)threadIdx.x) + 2) % 9))];
+ kernel_shared[((int)threadIdx.x)] = kernel[((((((((int)blockIdx.x) * 18432) + ((((int)threadIdx.x) / 48) * 4608)) + (rc_outer_outer * 144)) + (((((int)threadIdx.x) % 48) / 3) * 9)) + (((int)threadIdx.x) % 3)) + 6)];
+ kernel_shared[(((int)threadIdx.x) + 49)] = kernel[((((((((int)blockIdx.x) * 18432) + (((((int)threadIdx.x) + 49) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 1) % 48) / 3) * 9)) + ((((int)threadIdx.x) + 1) % 3)) + 6)];
+ kernel_shared[(((int)threadIdx.x) + 98)] = kernel[((((((((int)blockIdx.x) * 18432) + (((((int)threadIdx.x) + 98) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 2) % 48) / 3) * 9)) + ((((int)threadIdx.x) + 2) % 3)) + 6)];
+ if (((int)threadIdx.x) < 45) {
+ kernel_shared[(((int)threadIdx.x) + 147)] = kernel[((((((((int)blockIdx.x) * 18432) + (((((int)threadIdx.x) + 147) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((int)threadIdx.x) / 3) * 9)) + (((int)threadIdx.x) % 3)) + 15)];
}
__syncthreads();
conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[0]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[36]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[72]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[108]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[144]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[180]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[216]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[252]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[48]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[96]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[144]));
conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[1]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[37]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[73]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[109]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[145]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[181]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[217]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[253]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[49]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[97]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[145]));
conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[2]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[38]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[74]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[110]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[146]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[182]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[218]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[254]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[3]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[39]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[75]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[111]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[147]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[183]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[219]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[255]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[4]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[40]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[76]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[112]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[148]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[184]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[220]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[256]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[5]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[41]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[77]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[113]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[149]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[185]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[221]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[257]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[6]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[42]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[78]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[114]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[150]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[186]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[222]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[258]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[7]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[43]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[79]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[115]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[151]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[187]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[223]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[259]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[8]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[44]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[80]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[116]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[152]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[188]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[224]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[260]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 81)] * kernel_shared[9]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 81)] * kernel_shared[45]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 81)] * kernel_shared[81]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 81)] * kernel_shared[117]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 81)] * kernel_shared[153]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 81)] * kernel_shared[189]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 81)] * kernel_shared[225]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 81)] * kernel_shared[261]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 82)] * kernel_shared[10]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 82)] * kernel_shared[46]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 82)] * kernel_shared[82]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 82)] * kernel_shared[118]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 82)] * kernel_shared[154]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 82)] * kernel_shared[190]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 82)] * kernel_shared[226]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 82)] * kernel_shared[262]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 83)] * kernel_shared[11]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 83)] * kernel_shared[47]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 83)] * kernel_shared[83]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 83)] * kernel_shared[119]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 83)] * kernel_shared[155]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 83)] * kernel_shared[191]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 83)] * kernel_shared[227]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 83)] * kernel_shared[263]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[12]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[48]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[84]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[120]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[156]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[192]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[228]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[264]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[13]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[49]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[85]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[121]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[157]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[193]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[229]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[265]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[14]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[50]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[86]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[122]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[158]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[194]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[230]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[266]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[15]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[51]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[87]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[123]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[159]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[195]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[231]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[267]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[16]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[52]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[88]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[124]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[160]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[196]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[232]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[268]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[17]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[53]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[89]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[125]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[161]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[197]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[233]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[269]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 162)] * kernel_shared[18]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 162)] * kernel_shared[54]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 162)] * kernel_shared[90]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 162)] * kernel_shared[126]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 162)] * kernel_shared[162]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 162)] * kernel_shared[198]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 162)] * kernel_shared[234]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 162)] * kernel_shared[270]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 163)] * kernel_shared[19]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 163)] * kernel_shared[55]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 163)] * kernel_shared[91]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 163)] * kernel_shared[127]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 163)] * kernel_shared[163]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 163)] * kernel_shared[199]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 163)] * kernel_shared[235]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 163)] * kernel_shared[271]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 164)] * kernel_shared[20]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 164)] * kernel_shared[56]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 164)] * kernel_shared[92]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 164)] * kernel_shared[128]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 164)] * kernel_shared[164]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 164)] * kernel_shared[200]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 164)] * kernel_shared[236]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 164)] * kernel_shared[272]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[21]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[57]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[93]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[129]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[165]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[201]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[237]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[273]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[22]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[58]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[94]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[130]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[166]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[202]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[238]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[274]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[23]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[59]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[95]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[131]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[167]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[203]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[239]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[275]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[24]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[60]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[96]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[132]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[168]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[204]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[240]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[276]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[25]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[61]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[97]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[133]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[169]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[205]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[241]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[277]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[26]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[62]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[98]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[134]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[170]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[206]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[242]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[278]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 243)] * kernel_shared[27]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 243)] * kernel_shared[63]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 243)] * kernel_shared[99]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 243)] * kernel_shared[135]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 243)] * kernel_shared[171]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 243)] * kernel_shared[207]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 243)] * kernel_shared[243]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 243)] * kernel_shared[279]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 244)] * kernel_shared[28]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 244)] * kernel_shared[64]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 244)] * kernel_shared[100]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 244)] * kernel_shared[136]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 244)] * kernel_shared[172]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 244)] * kernel_shared[208]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 244)] * kernel_shared[244]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 244)] * kernel_shared[280]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 245)] * kernel_shared[29]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 245)] * kernel_shared[65]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 245)] * kernel_shared[101]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 245)] * kernel_shared[137]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 245)] * kernel_shared[173]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 245)] * kernel_shared[209]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 245)] * kernel_shared[245]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 245)] * kernel_shared[281]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[30]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[66]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[102]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[138]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[174]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[210]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[246]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[282]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[31]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[67]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[103]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[139]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[175]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[211]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[247]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[283]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[32]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[68]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[104]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[140]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[176]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[212]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[248]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[284]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[33]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[69]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[105]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[141]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[177]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[213]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[249]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[285]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[34]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[70]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[106]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[142]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[178]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[214]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[250]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[286]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[35]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[71]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[107]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[143]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[179]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[215]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[251]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[287]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[50]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[98]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[146]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[3]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[51]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[99]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[147]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[4]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[52]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[100]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[148]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[5]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[53]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[101]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[149]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[6]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[54]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[102]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[150]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[7]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[55]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[103]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[151]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[8]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[56]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[104]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[152]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[9]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[57]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[105]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[153]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[10]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[58]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[106]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[154]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[11]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[59]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[107]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[155]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[12]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[60]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[108]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[156]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[13]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[61]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[109]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[157]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[14]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[62]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[110]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[158]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[15]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[63]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[111]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[159]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 316)] * kernel_shared[16]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 316)] * kernel_shared[64]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 316)] * kernel_shared[112]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 316)] * kernel_shared[160]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 317)] * kernel_shared[17]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 317)] * kernel_shared[65]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 317)] * kernel_shared[113]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 317)] * kernel_shared[161]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[18]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[66]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[114]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[162]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 379)] * kernel_shared[19]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 379)] * kernel_shared[67]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 379)] * kernel_shared[115]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 379)] * kernel_shared[163]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 380)] * kernel_shared[20]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 380)] * kernel_shared[68]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 380)] * kernel_shared[116]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 380)] * kernel_shared[164]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[21]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[69]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[117]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[165]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 442)] * kernel_shared[22]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 442)] * kernel_shared[70]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 442)] * kernel_shared[118]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 442)] * kernel_shared[166]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 443)] * kernel_shared[23]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 443)] * kernel_shared[71]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 443)] * kernel_shared[119]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 443)] * kernel_shared[167]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 504)] * kernel_shared[24]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 504)] * kernel_shared[72]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 504)] * kernel_shared[120]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 504)] * kernel_shared[168]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 505)] * kernel_shared[25]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 505)] * kernel_shared[73]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 505)] * kernel_shared[121]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 505)] * kernel_shared[169]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 506)] * kernel_shared[26]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 506)] * kernel_shared[74]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 506)] * kernel_shared[122]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 506)] * kernel_shared[170]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 567)] * kernel_shared[27]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 567)] * kernel_shared[75]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 567)] * kernel_shared[123]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 567)] * kernel_shared[171]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 568)] * kernel_shared[28]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 568)] * kernel_shared[76]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 568)] * kernel_shared[124]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 568)] * kernel_shared[172]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 569)] * kernel_shared[29]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 569)] * kernel_shared[77]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 569)] * kernel_shared[125]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 569)] * kernel_shared[173]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 630)] * kernel_shared[30]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 630)] * kernel_shared[78]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 630)] * kernel_shared[126]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 630)] * kernel_shared[174]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 631)] * kernel_shared[31]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 631)] * kernel_shared[79]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 631)] * kernel_shared[127]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 631)] * kernel_shared[175]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 632)] * kernel_shared[32]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 632)] * kernel_shared[80]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 632)] * kernel_shared[128]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 632)] * kernel_shared[176]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 693)] * kernel_shared[33]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 693)] * kernel_shared[81]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 693)] * kernel_shared[129]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 693)] * kernel_shared[177]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 694)] * kernel_shared[34]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 694)] * kernel_shared[82]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 694)] * kernel_shared[130]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 694)] * kernel_shared[178]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 695)] * kernel_shared[35]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 695)] * kernel_shared[83]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 695)] * kernel_shared[131]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 695)] * kernel_shared[179]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 756)] * kernel_shared[36]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 756)] * kernel_shared[84]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 756)] * kernel_shared[132]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 756)] * kernel_shared[180]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 757)] * kernel_shared[37]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 757)] * kernel_shared[85]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 757)] * kernel_shared[133]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 757)] * kernel_shared[181]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 758)] * kernel_shared[38]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 758)] * kernel_shared[86]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 758)] * kernel_shared[134]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 758)] * kernel_shared[182]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 819)] * kernel_shared[39]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 819)] * kernel_shared[87]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 819)] * kernel_shared[135]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 819)] * kernel_shared[183]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 820)] * kernel_shared[40]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 820)] * kernel_shared[88]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 820)] * kernel_shared[136]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 820)] * kernel_shared[184]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 821)] * kernel_shared[41]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 821)] * kernel_shared[89]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 821)] * kernel_shared[137]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 821)] * kernel_shared[185]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 882)] * kernel_shared[42]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 882)] * kernel_shared[90]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 882)] * kernel_shared[138]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 882)] * kernel_shared[186]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 883)] * kernel_shared[43]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 883)] * kernel_shared[91]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 883)] * kernel_shared[139]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 883)] * kernel_shared[187]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 884)] * kernel_shared[44]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 884)] * kernel_shared[92]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 884)] * kernel_shared[140]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 884)] * kernel_shared[188]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 945)] * kernel_shared[45]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 945)] * kernel_shared[93]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 945)] * kernel_shared[141]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 945)] * kernel_shared[189]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 946)] * kernel_shared[46]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 946)] * kernel_shared[94]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 946)] * kernel_shared[142]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 946)] * kernel_shared[190]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 947)] * kernel_shared[47]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 947)] * kernel_shared[95]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 947)] * kernel_shared[143]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 947)] * kernel_shared[191]));
+ }
+ for (int i1_inner = 0; i1_inner < 4; ++i1_inner) {
+ compute[(((((int)blockIdx.x) * 196) + (i1_inner * 49)) + ((int)threadIdx.x))] = max((conv2d_nchw[i1_inner] + bias[((((int)blockIdx.x) * 4) + i1_inner)]), 0.000000e+00f);
}
- compute[((((int)blockIdx.x) * 392) + ((int)threadIdx.x))] = max((conv2d_nchw[0] + bias[(((int)blockIdx.x) * 8)]), 0.000000e+00f);
- compute[(((((int)blockIdx.x) * 392) + ((int)threadIdx.x)) + 49)] = max((conv2d_nchw[1] + bias[((((int)blockIdx.x) * 8) + 1)]), 0.000000e+00f);
- compute[(((((int)blockIdx.x) * 392) + ((int)threadIdx.x)) + 98)] = max((conv2d_nchw[2] + bias[((((int)blockIdx.x) * 8) + 2)]), 0.000000e+00f);
- compute[(((((int)blockIdx.x) * 392) + ((int)threadIdx.x)) + 147)] = max((conv2d_nchw[3] + bias[((((int)blockIdx.x) * 8) + 3)]), 0.000000e+00f);
- compute[(((((int)blockIdx.x) * 392) + ((int)threadIdx.x)) + 196)] = max((conv2d_nchw[4] + bias[((((int)blockIdx.x) * 8) + 4)]), 0.000000e+00f);
- compute[(((((int)blockIdx.x) * 392) + ((int)threadIdx.x)) + 245)] = max((conv2d_nchw[5] + bias[((((int)blockIdx.x) * 8) + 5)]), 0.000000e+00f);
- compute[(((((int)blockIdx.x) * 392) + ((int)threadIdx.x)) + 294)] = max((conv2d_nchw[6] + bias[((((int)blockIdx.x) * 8) + 6)]), 0.000000e+00f);
- compute[(((((int)blockIdx.x) * 392) + ((int)threadIdx.x)) + 343)] = max((conv2d_nchw[7] + bias[((((int)blockIdx.x) * 8) + 7)]), 0.000000e+00f);
}
@@ -1156,7 +1913,7 @@ In the example below we resume the status and do more 5 trials.
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 2 minutes 29.143 seconds)
+ **Total running time of the script:** ( 2 minutes 35.377 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 4ed5227db..ba46492f4 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)
- 10.0936 10.0958 10.1276 10.0573 0.0288
+ 9.8384 9.8287 9.8629 9.8236 0.0174
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 17eff20aa..5ba188c46 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)
- 751.6743 751.1777 752.8035 751.0416 0.8004
+ 743.2332 743.7093 743.7099 742.2805 0.6737
@@ -694,7 +694,7 @@ Other Tips
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 20.297 seconds)
+ **Total running time of the script:** ( 1 minutes 19.283 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 051dcb030..ed42a4452 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
@@ -397,32 +397,1298 @@ layout transformation, parallelization, vectorization, unrolling, and operator f
placeholder_4: Buffer(placeholder_14: Pointer(float32), float32, [65536], []),
compute: Buffer(compute_2: Pointer(float32), float32, [65536], [])}
buffer_map = {placeholder_5: placeholder, placeholder_6: placeholder_1, placeholder_7: placeholder_2, placeholder_8: placeholder_3, placeholder_9: placeholder_4, compute_1: compute}
- preflattened_buffer_map = {placeholder_8: placeholder_15: Buffer(placeholder_13, int32, [33], []), placeholder_7: placeholder_16: Buffer(placeholder_12, int32, [4916], []), placeholder_9: placeholder_17: Buffer(placeholder_14, float32, [128, 512], []), placeholder_5: placeholder_18: Buffer(placeholder_10, float32, [128, 256], []), placeholder_6: placeholder_19: Buffer(placeholder_11, float32, [4916, 16, 1], []), compute_1: compute_3: Buffer(compute_2, float32, [128, 512], [])} {
- for (i0.outer.i1.outer.fused: int32, 0, 16) "parallel" {
- allocate(compute_4: Pointer(global float32), float32, [4096]), storage_scope = global {
- for (i.outer.inner: int32, 0, 32) {
- for (nb_j.inner: int32, 0, 2) {
- for (i.inner.init: int32, 0, 4) {
- for (j.init: int32, 0, 16) {
- compute_5: Buffer(compute_4, float32, [4096], [])[((((i.outer.inner*128) + (i.inner.init*32)) + (nb_j.inner*16)) + j.init)] = 0f32
+ preflattened_buffer_map = {placeholder_7: placeholder_15: Buffer(placeholder_12, int32, [4916], []), placeholder_8: placeholder_16: Buffer(placeholder_13, int32, [33], []), placeholder_9: placeholder_17: Buffer(placeholder_14, float32, [128, 512], []), placeholder_5: placeholder_18: Buffer(placeholder_10, float32, [128, 256], []), placeholder_6: placeholder_19: Buffer(placeholder_11, float32, [4916, 16, 1], []), compute_1: compute_3: Buffer(compute_2, float32, [128, 512], [])} {
+ for (i0.outer.i1.outer.fused: int32, 0, 32) "parallel" {
+ allocate(compute_4: Pointer(global float32), float32, [2048]), storage_scope = global {
+ for (i.outer.inner: int32, 0, 8) {
+ let cse_var_1: int32 = (i.outer.inner*256)
+ {
+ compute_5: Buffer(compute_4, float32, [2048], [])[cse_var_1] = 0f32
+ compute_5[(cse_var_1 + 1)] = 0f32
+ compute_5[(cse_var_1 + 2)] = 0f32
+ compute_5[(cse_var_1 + 3)] = 0f32
+ compute_5[(cse_var_1 + 4)] = 0f32
+ compute_5[(cse_var_1 + 5)] = 0f32
+ compute_5[(cse_var_1 + 6)] = 0f32
+ compute_5[(cse_var_1 + 7)] = 0f32
+ compute_5[(cse_var_1 + 8)] = 0f32
+ compute_5[(cse_var_1 + 9)] = 0f32
+ compute_5[(cse_var_1 + 10)] = 0f32
+ compute_5[(cse_var_1 + 11)] = 0f32
+ compute_5[(cse_var_1 + 12)] = 0f32
+ compute_5[(cse_var_1 + 13)] = 0f32
+ compute_5[(cse_var_1 + 14)] = 0f32
+ compute_5[(cse_var_1 + 15)] = 0f32
+ compute_5[(cse_var_1 + 16)] = 0f32
+ compute_5[(cse_var_1 + 17)] = 0f32
+ compute_5[(cse_var_1 + 18)] = 0f32
+ compute_5[(cse_var_1 + 19)] = 0f32
+ compute_5[(cse_var_1 + 20)] = 0f32
+ compute_5[(cse_var_1 + 21)] = 0f32
+ compute_5[(cse_var_1 + 22)] = 0f32
+ compute_5[(cse_var_1 + 23)] = 0f32
+ compute_5[(cse_var_1 + 24)] = 0f32
+ compute_5[(cse_var_1 + 25)] = 0f32
+ compute_5[(cse_var_1 + 26)] = 0f32
+ compute_5[(cse_var_1 + 27)] = 0f32
+ compute_5[(cse_var_1 + 28)] = 0f32
+ compute_5[(cse_var_1 + 29)] = 0f32
+ compute_5[(cse_var_1 + 30)] = 0f32
+ compute_5[(cse_var_1 + 31)] = 0f32
+ compute_5[(cse_var_1 + 32)] = 0f32
+ compute_5[(cse_var_1 + 33)] = 0f32
+ compute_5[(cse_var_1 + 34)] = 0f32
+ compute_5[(cse_var_1 + 35)] = 0f32
+ compute_5[(cse_var_1 + 36)] = 0f32
+ compute_5[(cse_var_1 + 37)] = 0f32
+ compute_5[(cse_var_1 + 38)] = 0f32
+ compute_5[(cse_var_1 + 39)] = 0f32
+ compute_5[(cse_var_1 + 40)] = 0f32
+ compute_5[(cse_var_1 + 41)] = 0f32
+ compute_5[(cse_var_1 + 42)] = 0f32
+ compute_5[(cse_var_1 + 43)] = 0f32
+ compute_5[(cse_var_1 + 44)] = 0f32
+ compute_5[(cse_var_1 + 45)] = 0f32
+ compute_5[(cse_var_1 + 46)] = 0f32
+ compute_5[(cse_var_1 + 47)] = 0f32
+ compute_5[(cse_var_1 + 48)] = 0f32
+ compute_5[(cse_var_1 + 49)] = 0f32
+ compute_5[(cse_var_1 + 50)] = 0f32
+ compute_5[(cse_var_1 + 51)] = 0f32
+ compute_5[(cse_var_1 + 52)] = 0f32
+ compute_5[(cse_var_1 + 53)] = 0f32
+ compute_5[(cse_var_1 + 54)] = 0f32
+ compute_5[(cse_var_1 + 55)] = 0f32
+ compute_5[(cse_var_1 + 56)] = 0f32
+ compute_5[(cse_var_1 + 57)] = 0f32
+ compute_5[(cse_var_1 + 58)] = 0f32
+ compute_5[(cse_var_1 + 59)] = 0f32
+ compute_5[(cse_var_1 + 60)] = 0f32
+ compute_5[(cse_var_1 + 61)] = 0f32
+ compute_5[(cse_var_1 + 62)] = 0f32
+ compute_5[(cse_var_1 + 63)] = 0f32
+ compute_5[(cse_var_1 + 64)] = 0f32
+ compute_5[(cse_var_1 + 65)] = 0f32
+ compute_5[(cse_var_1 + 66)] = 0f32
+ compute_5[(cse_var_1 + 67)] = 0f32
+ compute_5[(cse_var_1 + 68)] = 0f32
+ compute_5[(cse_var_1 + 69)] = 0f32
+ compute_5[(cse_var_1 + 70)] = 0f32
+ compute_5[(cse_var_1 + 71)] = 0f32
+ compute_5[(cse_var_1 + 72)] = 0f32
+ compute_5[(cse_var_1 + 73)] = 0f32
+ compute_5[(cse_var_1 + 74)] = 0f32
+ compute_5[(cse_var_1 + 75)] = 0f32
+ compute_5[(cse_var_1 + 76)] = 0f32
+ compute_5[(cse_var_1 + 77)] = 0f32
+ compute_5[(cse_var_1 + 78)] = 0f32
+ compute_5[(cse_var_1 + 79)] = 0f32
+ compute_5[(cse_var_1 + 80)] = 0f32
+ compute_5[(cse_var_1 + 81)] = 0f32
+ compute_5[(cse_var_1 + 82)] = 0f32
+ compute_5[(cse_var_1 + 83)] = 0f32
+ compute_5[(cse_var_1 + 84)] = 0f32
+ compute_5[(cse_var_1 + 85)] = 0f32
+ compute_5[(cse_var_1 + 86)] = 0f32
+ compute_5[(cse_var_1 + 87)] = 0f32
+ compute_5[(cse_var_1 + 88)] = 0f32
+ compute_5[(cse_var_1 + 89)] = 0f32
+ compute_5[(cse_var_1 + 90)] = 0f32
+ compute_5[(cse_var_1 + 91)] = 0f32
+ compute_5[(cse_var_1 + 92)] = 0f32
+ compute_5[(cse_var_1 + 93)] = 0f32
+ compute_5[(cse_var_1 + 94)] = 0f32
+ compute_5[(cse_var_1 + 95)] = 0f32
+ compute_5[(cse_var_1 + 96)] = 0f32
+ compute_5[(cse_var_1 + 97)] = 0f32
+ compute_5[(cse_var_1 + 98)] = 0f32
+ compute_5[(cse_var_1 + 99)] = 0f32
+ compute_5[(cse_var_1 + 100)] = 0f32
+ compute_5[(cse_var_1 + 101)] = 0f32
+ compute_5[(cse_var_1 + 102)] = 0f32
+ compute_5[(cse_var_1 + 103)] = 0f32
+ compute_5[(cse_var_1 + 104)] = 0f32
+ compute_5[(cse_var_1 + 105)] = 0f32
+ compute_5[(cse_var_1 + 106)] = 0f32
+ compute_5[(cse_var_1 + 107)] = 0f32
+ compute_5[(cse_var_1 + 108)] = 0f32
+ compute_5[(cse_var_1 + 109)] = 0f32
+ compute_5[(cse_var_1 + 110)] = 0f32
+ compute_5[(cse_var_1 + 111)] = 0f32
+ compute_5[(cse_var_1 + 112)] = 0f32
+ compute_5[(cse_var_1 + 113)] = 0f32
+ compute_5[(cse_var_1 + 114)] = 0f32
+ compute_5[(cse_var_1 + 115)] = 0f32
+ compute_5[(cse_var_1 + 116)] = 0f32
+ compute_5[(cse_var_1 + 117)] = 0f32
+ compute_5[(cse_var_1 + 118)] = 0f32
+ compute_5[(cse_var_1 + 119)] = 0f32
+ compute_5[(cse_var_1 + 120)] = 0f32
+ compute_5[(cse_var_1 + 121)] = 0f32
+ compute_5[(cse_var_1 + 122)] = 0f32
+ compute_5[(cse_var_1 + 123)] = 0f32
+ compute_5[(cse_var_1 + 124)] = 0f32
+ compute_5[(cse_var_1 + 125)] = 0f32
+ compute_5[(cse_var_1 + 126)] = 0f32
+ compute_5[(cse_var_1 + 127)] = 0f32
+ compute_5[(cse_var_1 + 128)] = 0f32
+ compute_5[(cse_var_1 + 129)] = 0f32
+ compute_5[(cse_var_1 + 130)] = 0f32
+ compute_5[(cse_var_1 + 131)] = 0f32
+ compute_5[(cse_var_1 + 132)] = 0f32
+ compute_5[(cse_var_1 + 133)] = 0f32
+ compute_5[(cse_var_1 + 134)] = 0f32
+ compute_5[(cse_var_1 + 135)] = 0f32
+ compute_5[(cse_var_1 + 136)] = 0f32
+ compute_5[(cse_var_1 + 137)] = 0f32
+ compute_5[(cse_var_1 + 138)] = 0f32
+ compute_5[(cse_var_1 + 139)] = 0f32
+ compute_5[(cse_var_1 + 140)] = 0f32
+ compute_5[(cse_var_1 + 141)] = 0f32
+ compute_5[(cse_var_1 + 142)] = 0f32
+ compute_5[(cse_var_1 + 143)] = 0f32
+ compute_5[(cse_var_1 + 144)] = 0f32
+ compute_5[(cse_var_1 + 145)] = 0f32
+ compute_5[(cse_var_1 + 146)] = 0f32
+ compute_5[(cse_var_1 + 147)] = 0f32
+ compute_5[(cse_var_1 + 148)] = 0f32
+ compute_5[(cse_var_1 + 149)] = 0f32
+ compute_5[(cse_var_1 + 150)] = 0f32
+ compute_5[(cse_var_1 + 151)] = 0f32
+ compute_5[(cse_var_1 + 152)] = 0f32
+ compute_5[(cse_var_1 + 153)] = 0f32
+ compute_5[(cse_var_1 + 154)] = 0f32
+ compute_5[(cse_var_1 + 155)] = 0f32
+ compute_5[(cse_var_1 + 156)] = 0f32
+ compute_5[(cse_var_1 + 157)] = 0f32
+ compute_5[(cse_var_1 + 158)] = 0f32
+ compute_5[(cse_var_1 + 159)] = 0f32
+ compute_5[(cse_var_1 + 160)] = 0f32
+ compute_5[(cse_var_1 + 161)] = 0f32
+ compute_5[(cse_var_1 + 162)] = 0f32
+ compute_5[(cse_var_1 + 163)] = 0f32
+ compute_5[(cse_var_1 + 164)] = 0f32
+ compute_5[(cse_var_1 + 165)] = 0f32
+ compute_5[(cse_var_1 + 166)] = 0f32
+ compute_5[(cse_var_1 + 167)] = 0f32
+ compute_5[(cse_var_1 + 168)] = 0f32
+ compute_5[(cse_var_1 + 169)] = 0f32
+ compute_5[(cse_var_1 + 170)] = 0f32
+ compute_5[(cse_var_1 + 171)] = 0f32
+ compute_5[(cse_var_1 + 172)] = 0f32
+ compute_5[(cse_var_1 + 173)] = 0f32
+ compute_5[(cse_var_1 + 174)] = 0f32
+ compute_5[(cse_var_1 + 175)] = 0f32
+ compute_5[(cse_var_1 + 176)] = 0f32
+ compute_5[(cse_var_1 + 177)] = 0f32
+ compute_5[(cse_var_1 + 178)] = 0f32
+ compute_5[(cse_var_1 + 179)] = 0f32
+ compute_5[(cse_var_1 + 180)] = 0f32
+ compute_5[(cse_var_1 + 181)] = 0f32
+ compute_5[(cse_var_1 + 182)] = 0f32
+ compute_5[(cse_var_1 + 183)] = 0f32
+ compute_5[(cse_var_1 + 184)] = 0f32
+ compute_5[(cse_var_1 + 185)] = 0f32
+ compute_5[(cse_var_1 + 186)] = 0f32
+ compute_5[(cse_var_1 + 187)] = 0f32
+ compute_5[(cse_var_1 + 188)] = 0f32
+ compute_5[(cse_var_1 + 189)] = 0f32
+ compute_5[(cse_var_1 + 190)] = 0f32
+ compute_5[(cse_var_1 + 191)] = 0f32
+ compute_5[(cse_var_1 + 192)] = 0f32
+ compute_5[(cse_var_1 + 193)] = 0f32
+ compute_5[(cse_var_1 + 194)] = 0f32
+ compute_5[(cse_var_1 + 195)] = 0f32
+ compute_5[(cse_var_1 + 196)] = 0f32
+ compute_5[(cse_var_1 + 197)] = 0f32
+ compute_5[(cse_var_1 + 198)] = 0f32
+ compute_5[(cse_var_1 + 199)] = 0f32
+ compute_5[(cse_var_1 + 200)] = 0f32
+ compute_5[(cse_var_1 + 201)] = 0f32
+ compute_5[(cse_var_1 + 202)] = 0f32
+ compute_5[(cse_var_1 + 203)] = 0f32
+ compute_5[(cse_var_1 + 204)] = 0f32
+ compute_5[(cse_var_1 + 205)] = 0f32
+ compute_5[(cse_var_1 + 206)] = 0f32
+ compute_5[(cse_var_1 + 207)] = 0f32
+ compute_5[(cse_var_1 + 208)] = 0f32
+ compute_5[(cse_var_1 + 209)] = 0f32
+ compute_5[(cse_var_1 + 210)] = 0f32
+ compute_5[(cse_var_1 + 211)] = 0f32
+ compute_5[(cse_var_1 + 212)] = 0f32
+ compute_5[(cse_var_1 + 213)] = 0f32
+ compute_5[(cse_var_1 + 214)] = 0f32
+ compute_5[(cse_var_1 + 215)] = 0f32
+ compute_5[(cse_var_1 + 216)] = 0f32
+ compute_5[(cse_var_1 + 217)] = 0f32
+ compute_5[(cse_var_1 + 218)] = 0f32
+ compute_5[(cse_var_1 + 219)] = 0f32
+ compute_5[(cse_var_1 + 220)] = 0f32
+ compute_5[(cse_var_1 + 221)] = 0f32
+ compute_5[(cse_var_1 + 222)] = 0f32
+ compute_5[(cse_var_1 + 223)] = 0f32
+ compute_5[(cse_var_1 + 224)] = 0f32
+ compute_5[(cse_var_1 + 225)] = 0f32
+ compute_5[(cse_var_1 + 226)] = 0f32
+ compute_5[(cse_var_1 + 227)] = 0f32
+ compute_5[(cse_var_1 + 228)] = 0f32
+ compute_5[(cse_var_1 + 229)] = 0f32
+ compute_5[(cse_var_1 + 230)] = 0f32
+ compute_5[(cse_var_1 + 231)] = 0f32
+ compute_5[(cse_var_1 + 232)] = 0f32
+ compute_5[(cse_var_1 + 233)] = 0f32
+ compute_5[(cse_var_1 + 234)] = 0f32
+ compute_5[(cse_var_1 + 235)] = 0f32
+ compute_5[(cse_var_1 + 236)] = 0f32
+ compute_5[(cse_var_1 + 237)] = 0f32
+ compute_5[(cse_var_1 + 238)] = 0f32
+ compute_5[(cse_var_1 + 239)] = 0f32
+ compute_5[(cse_var_1 + 240)] = 0f32
+ compute_5[(cse_var_1 + 241)] = 0f32
+ compute_5[(cse_var_1 + 242)] = 0f32
+ compute_5[(cse_var_1 + 243)] = 0f32
+ compute_5[(cse_var_1 + 244)] = 0f32
+ compute_5[(cse_var_1 + 245)] = 0f32
+ compute_5[(cse_var_1 + 246)] = 0f32
+ compute_5[(cse_var_1 + 247)] = 0f32
+ compute_5[(cse_var_1 + 248)] = 0f32
+ compute_5[(cse_var_1 + 249)] = 0f32
+ compute_5[(cse_var_1 + 250)] = 0f32
+ compute_5[(cse_var_1 + 251)] = 0f32
+ compute_5[(cse_var_1 + 252)] = 0f32
+ compute_5[(cse_var_1 + 253)] = 0f32
+ compute_5[(cse_var_1 + 254)] = 0f32
+ compute_5[(cse_var_1 + 255)] = 0f32
+ for (elem_idx: int32, 0, (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])) {
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ compute_5[cse_var_1] = (compute_5[cse_var_1] + (placeholder_1[((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16))]*max(placeholder[((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)])], 0f32)))
}
- }
- for (elem_idx: int32, 0, let cse_var_1: int32 = ((i0.outer.i1.outer.fused*2) + nb_j.inner) in (placeholder_3[(cse_var_1 + 1)] - placeholder_3[cse_var_1])) {
- for (i.inner: int32, 0, 4) {
- for (j: int32, 0, 16) {
- let cse_var_3: int32 = ((i0.outer.i1.outer.fused*2) + nb_j.inner)
- let cse_var_2: int32 = ((((i.outer.inner*128) + (i.inner*32)) + (nb_j.inner*16)) + j)
- compute_5[cse_var_2] = (compute_5[cse_var_2] + (placeholder_1[(((placeholder_3[cse_var_3]*16) + (elem_idx*16)) + j)]*max(placeholder[(((i.outer.inner*1024) + (i.inner*256)) + placeholder_2[(placeholder_3[cse_var_3] + elem_idx)])], 0f32)))
- }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_2: int32 = (cse_var_1 + 1)
+ compute_5[cse_var_2] = (compute_5[cse_var_2] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 1)]*max(placeholder[((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)])], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_3: int32 = (cse_var_1 + 2)
+ compute_5[cse_var_3] = (compute_5[cse_var_3] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 2)]*max(placeholder[((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)])], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_4: int32 = (cse_var_1 + 3)
+ compute_5[cse_var_4] = (compute_5[cse_var_4] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 3)]*max(placeholder[((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)])], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_5: int32 = (cse_var_1 + 4)
+ compute_5[cse_var_5] = (compute_5[cse_var_5] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 4)]*max(placeholder[((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)])], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_6: int32 = (cse_var_1 + 5)
+ compute_5[cse_var_6] = (compute_5[cse_var_6] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 5)]*max(placeholder[((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)])], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_7: int32 = (cse_var_1 + 6)
+ compute_5[cse_var_7] = (compute_5[cse_var_7] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 6)]*max(placeholder[((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)])], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_8: int32 = (cse_var_1 + 7)
+ compute_5[cse_var_8] = (compute_5[cse_var_8] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 7)]*max(placeholder[((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)])], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_9: int32 = (cse_var_1 + 8)
+ compute_5[cse_var_9] = (compute_5[cse_var_9] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 8)]*max(placeholder[((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)])], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_10: int32 = (cse_var_1 + 9)
+ compute_5[cse_var_10] = (compute_5[cse_var_10] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 9)]*max(placeholder[((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)])], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_11: int32 = (cse_var_1 + 10)
+ compute_5[cse_var_11] = (compute_5[cse_var_11] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 10)]*max(placeholder[((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)])], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_12: int32 = (cse_var_1 + 11)
+ compute_5[cse_var_12] = (compute_5[cse_var_12] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 11)]*max(placeholder[((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)])], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_13: int32 = (cse_var_1 + 12)
+ compute_5[cse_var_13] = (compute_5[cse_var_13] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 12)]*max(placeholder[((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)])], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_14: int32 = (cse_var_1 + 13)
+ compute_5[cse_var_14] = (compute_5[cse_var_14] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 13)]*max(placeholder[((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)])], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_15: int32 = (cse_var_1 + 14)
+ compute_5[cse_var_15] = (compute_5[cse_var_15] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 14)]*max(placeholder[((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)])], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_16: int32 = (cse_var_1 + 15)
+ compute_5[cse_var_16] = (compute_5[cse_var_16] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 15)]*max(placeholder[((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)])], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_17: int32 = (cse_var_1 + 16)
+ compute_5[cse_var_17] = (compute_5[cse_var_17] + (placeholder_1[((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16))]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 256)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_18: int32 = (cse_var_1 + 17)
+ compute_5[cse_var_18] = (compute_5[cse_var_18] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 1)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 256)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_19: int32 = (cse_var_1 + 18)
+ compute_5[cse_var_19] = (compute_5[cse_var_19] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 2)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 256)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_20: int32 = (cse_var_1 + 19)
+ compute_5[cse_var_20] = (compute_5[cse_var_20] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 3)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 256)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_21: int32 = (cse_var_1 + 20)
+ compute_5[cse_var_21] = (compute_5[cse_var_21] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 4)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 256)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_22: int32 = (cse_var_1 + 21)
+ compute_5[cse_var_22] = (compute_5[cse_var_22] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 5)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 256)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_23: int32 = (cse_var_1 + 22)
+ compute_5[cse_var_23] = (compute_5[cse_var_23] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 6)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 256)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_24: int32 = (cse_var_1 + 23)
+ compute_5[cse_var_24] = (compute_5[cse_var_24] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 7)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 256)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_25: int32 = (cse_var_1 + 24)
+ compute_5[cse_var_25] = (compute_5[cse_var_25] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 8)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 256)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_26: int32 = (cse_var_1 + 25)
+ compute_5[cse_var_26] = (compute_5[cse_var_26] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 9)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 256)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_27: int32 = (cse_var_1 + 26)
+ compute_5[cse_var_27] = (compute_5[cse_var_27] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 10)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 256)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_28: int32 = (cse_var_1 + 27)
+ compute_5[cse_var_28] = (compute_5[cse_var_28] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 11)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 256)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_29: int32 = (cse_var_1 + 28)
+ compute_5[cse_var_29] = (compute_5[cse_var_29] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 12)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 256)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_30: int32 = (cse_var_1 + 29)
+ compute_5[cse_var_30] = (compute_5[cse_var_30] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 13)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 256)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_31: int32 = (cse_var_1 + 30)
+ compute_5[cse_var_31] = (compute_5[cse_var_31] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 14)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 256)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_32: int32 = (cse_var_1 + 31)
+ compute_5[cse_var_32] = (compute_5[cse_var_32] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 15)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 256)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_33: int32 = (cse_var_1 + 32)
+ compute_5[cse_var_33] = (compute_5[cse_var_33] + (placeholder_1[((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16))]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 512)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_34: int32 = (cse_var_1 + 33)
+ compute_5[cse_var_34] = (compute_5[cse_var_34] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 1)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 512)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_35: int32 = (cse_var_1 + 34)
+ compute_5[cse_var_35] = (compute_5[cse_var_35] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 2)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 512)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_36: int32 = (cse_var_1 + 35)
+ compute_5[cse_var_36] = (compute_5[cse_var_36] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 3)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 512)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_37: int32 = (cse_var_1 + 36)
+ compute_5[cse_var_37] = (compute_5[cse_var_37] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 4)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 512)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_38: int32 = (cse_var_1 + 37)
+ compute_5[cse_var_38] = (compute_5[cse_var_38] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 5)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 512)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_39: int32 = (cse_var_1 + 38)
+ compute_5[cse_var_39] = (compute_5[cse_var_39] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 6)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 512)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_40: int32 = (cse_var_1 + 39)
+ compute_5[cse_var_40] = (compute_5[cse_var_40] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 7)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 512)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_41: int32 = (cse_var_1 + 40)
+ compute_5[cse_var_41] = (compute_5[cse_var_41] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 8)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 512)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_42: int32 = (cse_var_1 + 41)
+ compute_5[cse_var_42] = (compute_5[cse_var_42] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 9)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 512)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_43: int32 = (cse_var_1 + 42)
+ compute_5[cse_var_43] = (compute_5[cse_var_43] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 10)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 512)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_44: int32 = (cse_var_1 + 43)
+ compute_5[cse_var_44] = (compute_5[cse_var_44] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 11)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 512)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_45: int32 = (cse_var_1 + 44)
+ compute_5[cse_var_45] = (compute_5[cse_var_45] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 12)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 512)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_46: int32 = (cse_var_1 + 45)
+ compute_5[cse_var_46] = (compute_5[cse_var_46] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 13)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 512)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_47: int32 = (cse_var_1 + 46)
+ compute_5[cse_var_47] = (compute_5[cse_var_47] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 14)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 512)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_48: int32 = (cse_var_1 + 47)
+ compute_5[cse_var_48] = (compute_5[cse_var_48] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 15)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 512)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_49: int32 = (cse_var_1 + 48)
+ compute_5[cse_var_49] = (compute_5[cse_var_49] + (placeholder_1[((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16))]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 768)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_50: int32 = (cse_var_1 + 49)
+ compute_5[cse_var_50] = (compute_5[cse_var_50] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 1)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 768)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_51: int32 = (cse_var_1 + 50)
+ compute_5[cse_var_51] = (compute_5[cse_var_51] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 2)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 768)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_52: int32 = (cse_var_1 + 51)
+ compute_5[cse_var_52] = (compute_5[cse_var_52] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 3)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 768)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_53: int32 = (cse_var_1 + 52)
+ compute_5[cse_var_53] = (compute_5[cse_var_53] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 4)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 768)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_54: int32 = (cse_var_1 + 53)
+ compute_5[cse_var_54] = (compute_5[cse_var_54] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 5)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 768)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_55: int32 = (cse_var_1 + 54)
+ compute_5[cse_var_55] = (compute_5[cse_var_55] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 6)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 768)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_56: int32 = (cse_var_1 + 55)
+ compute_5[cse_var_56] = (compute_5[cse_var_56] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 7)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 768)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_57: int32 = (cse_var_1 + 56)
+ compute_5[cse_var_57] = (compute_5[cse_var_57] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 8)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 768)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_58: int32 = (cse_var_1 + 57)
+ compute_5[cse_var_58] = (compute_5[cse_var_58] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 9)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 768)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_59: int32 = (cse_var_1 + 58)
+ compute_5[cse_var_59] = (compute_5[cse_var_59] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 10)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 768)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_60: int32 = (cse_var_1 + 59)
+ compute_5[cse_var_60] = (compute_5[cse_var_60] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 11)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 768)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_61: int32 = (cse_var_1 + 60)
+ compute_5[cse_var_61] = (compute_5[cse_var_61] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 12)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 768)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_62: int32 = (cse_var_1 + 61)
+ compute_5[cse_var_62] = (compute_5[cse_var_62] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 13)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 768)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_63: int32 = (cse_var_1 + 62)
+ compute_5[cse_var_63] = (compute_5[cse_var_63] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 14)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 768)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_64: int32 = (cse_var_1 + 63)
+ compute_5[cse_var_64] = (compute_5[cse_var_64] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 15)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 768)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_65: int32 = (cse_var_1 + 64)
+ compute_5[cse_var_65] = (compute_5[cse_var_65] + (placeholder_1[((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16))]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1024)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_66: int32 = (cse_var_1 + 65)
+ compute_5[cse_var_66] = (compute_5[cse_var_66] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 1)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1024)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_67: int32 = (cse_var_1 + 66)
+ compute_5[cse_var_67] = (compute_5[cse_var_67] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 2)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1024)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_68: int32 = (cse_var_1 + 67)
+ compute_5[cse_var_68] = (compute_5[cse_var_68] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 3)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1024)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_69: int32 = (cse_var_1 + 68)
+ compute_5[cse_var_69] = (compute_5[cse_var_69] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 4)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1024)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_70: int32 = (cse_var_1 + 69)
+ compute_5[cse_var_70] = (compute_5[cse_var_70] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 5)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1024)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_71: int32 = (cse_var_1 + 70)
+ compute_5[cse_var_71] = (compute_5[cse_var_71] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 6)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1024)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_72: int32 = (cse_var_1 + 71)
+ compute_5[cse_var_72] = (compute_5[cse_var_72] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 7)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1024)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_73: int32 = (cse_var_1 + 72)
+ compute_5[cse_var_73] = (compute_5[cse_var_73] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 8)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1024)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_74: int32 = (cse_var_1 + 73)
+ compute_5[cse_var_74] = (compute_5[cse_var_74] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 9)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1024)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_75: int32 = (cse_var_1 + 74)
+ compute_5[cse_var_75] = (compute_5[cse_var_75] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 10)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1024)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_76: int32 = (cse_var_1 + 75)
+ compute_5[cse_var_76] = (compute_5[cse_var_76] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 11)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1024)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_77: int32 = (cse_var_1 + 76)
+ compute_5[cse_var_77] = (compute_5[cse_var_77] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 12)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1024)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_78: int32 = (cse_var_1 + 77)
+ compute_5[cse_var_78] = (compute_5[cse_var_78] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 13)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1024)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_79: int32 = (cse_var_1 + 78)
+ compute_5[cse_var_79] = (compute_5[cse_var_79] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 14)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1024)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_80: int32 = (cse_var_1 + 79)
+ compute_5[cse_var_80] = (compute_5[cse_var_80] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 15)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1024)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_81: int32 = (cse_var_1 + 80)
+ compute_5[cse_var_81] = (compute_5[cse_var_81] + (placeholder_1[((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16))]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1280)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_82: int32 = (cse_var_1 + 81)
+ compute_5[cse_var_82] = (compute_5[cse_var_82] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 1)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1280)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_83: int32 = (cse_var_1 + 82)
+ compute_5[cse_var_83] = (compute_5[cse_var_83] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 2)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1280)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_84: int32 = (cse_var_1 + 83)
+ compute_5[cse_var_84] = (compute_5[cse_var_84] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 3)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1280)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_85: int32 = (cse_var_1 + 84)
+ compute_5[cse_var_85] = (compute_5[cse_var_85] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 4)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1280)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_86: int32 = (cse_var_1 + 85)
+ compute_5[cse_var_86] = (compute_5[cse_var_86] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 5)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1280)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_87: int32 = (cse_var_1 + 86)
+ compute_5[cse_var_87] = (compute_5[cse_var_87] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 6)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1280)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_88: int32 = (cse_var_1 + 87)
+ compute_5[cse_var_88] = (compute_5[cse_var_88] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 7)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1280)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_89: int32 = (cse_var_1 + 88)
+ compute_5[cse_var_89] = (compute_5[cse_var_89] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 8)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1280)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_90: int32 = (cse_var_1 + 89)
+ compute_5[cse_var_90] = (compute_5[cse_var_90] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 9)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1280)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_91: int32 = (cse_var_1 + 90)
+ compute_5[cse_var_91] = (compute_5[cse_var_91] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 10)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1280)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_92: int32 = (cse_var_1 + 91)
+ compute_5[cse_var_92] = (compute_5[cse_var_92] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 11)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1280)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_93: int32 = (cse_var_1 + 92)
+ compute_5[cse_var_93] = (compute_5[cse_var_93] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 12)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1280)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_94: int32 = (cse_var_1 + 93)
+ compute_5[cse_var_94] = (compute_5[cse_var_94] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 13)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1280)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_95: int32 = (cse_var_1 + 94)
+ compute_5[cse_var_95] = (compute_5[cse_var_95] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 14)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1280)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_96: int32 = (cse_var_1 + 95)
+ compute_5[cse_var_96] = (compute_5[cse_var_96] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 15)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1280)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_97: int32 = (cse_var_1 + 96)
+ compute_5[cse_var_97] = (compute_5[cse_var_97] + (placeholder_1[((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16))]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1536)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_98: int32 = (cse_var_1 + 97)
+ compute_5[cse_var_98] = (compute_5[cse_var_98] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 1)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1536)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_99: int32 = (cse_var_1 + 98)
+ compute_5[cse_var_99] = (compute_5[cse_var_99] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 2)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1536)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_100: int32 = (cse_var_1 + 99)
+ compute_5[cse_var_100] = (compute_5[cse_var_100] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 3)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1536)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_101: int32 = (cse_var_1 + 100)
+ compute_5[cse_var_101] = (compute_5[cse_var_101] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 4)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1536)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_102: int32 = (cse_var_1 + 101)
+ compute_5[cse_var_102] = (compute_5[cse_var_102] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 5)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1536)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_103: int32 = (cse_var_1 + 102)
+ compute_5[cse_var_103] = (compute_5[cse_var_103] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 6)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1536)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_104: int32 = (cse_var_1 + 103)
+ compute_5[cse_var_104] = (compute_5[cse_var_104] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 7)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1536)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_105: int32 = (cse_var_1 + 104)
+ compute_5[cse_var_105] = (compute_5[cse_var_105] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 8)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1536)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_106: int32 = (cse_var_1 + 105)
+ compute_5[cse_var_106] = (compute_5[cse_var_106] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 9)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1536)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_107: int32 = (cse_var_1 + 106)
+ compute_5[cse_var_107] = (compute_5[cse_var_107] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 10)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1536)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_108: int32 = (cse_var_1 + 107)
+ compute_5[cse_var_108] = (compute_5[cse_var_108] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 11)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1536)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_109: int32 = (cse_var_1 + 108)
+ compute_5[cse_var_109] = (compute_5[cse_var_109] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 12)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1536)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_110: int32 = (cse_var_1 + 109)
+ compute_5[cse_var_110] = (compute_5[cse_var_110] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 13)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1536)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_111: int32 = (cse_var_1 + 110)
+ compute_5[cse_var_111] = (compute_5[cse_var_111] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 14)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1536)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_112: int32 = (cse_var_1 + 111)
+ compute_5[cse_var_112] = (compute_5[cse_var_112] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 15)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1536)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_113: int32 = (cse_var_1 + 112)
+ compute_5[cse_var_113] = (compute_5[cse_var_113] + (placeholder_1[((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16))]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1792)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_114: int32 = (cse_var_1 + 113)
+ compute_5[cse_var_114] = (compute_5[cse_var_114] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 1)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1792)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_115: int32 = (cse_var_1 + 114)
+ compute_5[cse_var_115] = (compute_5[cse_var_115] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 2)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1792)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_116: int32 = (cse_var_1 + 115)
+ compute_5[cse_var_116] = (compute_5[cse_var_116] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 3)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1792)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_117: int32 = (cse_var_1 + 116)
+ compute_5[cse_var_117] = (compute_5[cse_var_117] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 4)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1792)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_118: int32 = (cse_var_1 + 117)
+ compute_5[cse_var_118] = (compute_5[cse_var_118] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 5)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1792)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_119: int32 = (cse_var_1 + 118)
+ compute_5[cse_var_119] = (compute_5[cse_var_119] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 6)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1792)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_120: int32 = (cse_var_1 + 119)
+ compute_5[cse_var_120] = (compute_5[cse_var_120] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 7)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1792)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_121: int32 = (cse_var_1 + 120)
+ compute_5[cse_var_121] = (compute_5[cse_var_121] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 8)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1792)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_122: int32 = (cse_var_1 + 121)
+ compute_5[cse_var_122] = (compute_5[cse_var_122] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 9)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1792)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_123: int32 = (cse_var_1 + 122)
+ compute_5[cse_var_123] = (compute_5[cse_var_123] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 10)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1792)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_124: int32 = (cse_var_1 + 123)
+ compute_5[cse_var_124] = (compute_5[cse_var_124] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 11)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1792)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_125: int32 = (cse_var_1 + 124)
+ compute_5[cse_var_125] = (compute_5[cse_var_125] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 12)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1792)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_126: int32 = (cse_var_1 + 125)
+ compute_5[cse_var_126] = (compute_5[cse_var_126] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 13)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1792)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_127: int32 = (cse_var_1 + 126)
+ compute_5[cse_var_127] = (compute_5[cse_var_127] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 14)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1792)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_128: int32 = (cse_var_1 + 127)
+ compute_5[cse_var_128] = (compute_5[cse_var_128] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 15)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1792)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_129: int32 = (cse_var_1 + 128)
+ compute_5[cse_var_129] = (compute_5[cse_var_129] + (placeholder_1[((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16))]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2048)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_130: int32 = (cse_var_1 + 129)
+ compute_5[cse_var_130] = (compute_5[cse_var_130] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 1)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2048)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_131: int32 = (cse_var_1 + 130)
+ compute_5[cse_var_131] = (compute_5[cse_var_131] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 2)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2048)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_132: int32 = (cse_var_1 + 131)
+ compute_5[cse_var_132] = (compute_5[cse_var_132] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 3)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2048)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_133: int32 = (cse_var_1 + 132)
+ compute_5[cse_var_133] = (compute_5[cse_var_133] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 4)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2048)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_134: int32 = (cse_var_1 + 133)
+ compute_5[cse_var_134] = (compute_5[cse_var_134] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 5)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2048)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_135: int32 = (cse_var_1 + 134)
+ compute_5[cse_var_135] = (compute_5[cse_var_135] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 6)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2048)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_136: int32 = (cse_var_1 + 135)
+ compute_5[cse_var_136] = (compute_5[cse_var_136] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 7)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2048)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_137: int32 = (cse_var_1 + 136)
+ compute_5[cse_var_137] = (compute_5[cse_var_137] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 8)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2048)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_138: int32 = (cse_var_1 + 137)
+ compute_5[cse_var_138] = (compute_5[cse_var_138] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 9)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2048)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_139: int32 = (cse_var_1 + 138)
+ compute_5[cse_var_139] = (compute_5[cse_var_139] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 10)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2048)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_140: int32 = (cse_var_1 + 139)
+ compute_5[cse_var_140] = (compute_5[cse_var_140] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 11)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2048)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_141: int32 = (cse_var_1 + 140)
+ compute_5[cse_var_141] = (compute_5[cse_var_141] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 12)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2048)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_142: int32 = (cse_var_1 + 141)
+ compute_5[cse_var_142] = (compute_5[cse_var_142] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 13)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2048)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_143: int32 = (cse_var_1 + 142)
+ compute_5[cse_var_143] = (compute_5[cse_var_143] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 14)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2048)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_144: int32 = (cse_var_1 + 143)
+ compute_5[cse_var_144] = (compute_5[cse_var_144] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 15)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2048)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_145: int32 = (cse_var_1 + 144)
+ compute_5[cse_var_145] = (compute_5[cse_var_145] + (placeholder_1[((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16))]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2304)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_146: int32 = (cse_var_1 + 145)
+ compute_5[cse_var_146] = (compute_5[cse_var_146] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 1)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2304)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_147: int32 = (cse_var_1 + 146)
+ compute_5[cse_var_147] = (compute_5[cse_var_147] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 2)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2304)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_148: int32 = (cse_var_1 + 147)
+ compute_5[cse_var_148] = (compute_5[cse_var_148] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 3)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2304)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_149: int32 = (cse_var_1 + 148)
+ compute_5[cse_var_149] = (compute_5[cse_var_149] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 4)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2304)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_150: int32 = (cse_var_1 + 149)
+ compute_5[cse_var_150] = (compute_5[cse_var_150] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 5)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2304)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_151: int32 = (cse_var_1 + 150)
+ compute_5[cse_var_151] = (compute_5[cse_var_151] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 6)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2304)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_152: int32 = (cse_var_1 + 151)
+ compute_5[cse_var_152] = (compute_5[cse_var_152] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 7)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2304)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_153: int32 = (cse_var_1 + 152)
+ compute_5[cse_var_153] = (compute_5[cse_var_153] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 8)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2304)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_154: int32 = (cse_var_1 + 153)
+ compute_5[cse_var_154] = (compute_5[cse_var_154] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 9)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2304)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_155: int32 = (cse_var_1 + 154)
+ compute_5[cse_var_155] = (compute_5[cse_var_155] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 10)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2304)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_156: int32 = (cse_var_1 + 155)
+ compute_5[cse_var_156] = (compute_5[cse_var_156] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 11)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2304)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_157: int32 = (cse_var_1 + 156)
+ compute_5[cse_var_157] = (compute_5[cse_var_157] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 12)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2304)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_158: int32 = (cse_var_1 + 157)
+ compute_5[cse_var_158] = (compute_5[cse_var_158] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 13)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2304)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_159: int32 = (cse_var_1 + 158)
+ compute_5[cse_var_159] = (compute_5[cse_var_159] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 14)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2304)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_160: int32 = (cse_var_1 + 159)
+ compute_5[cse_var_160] = (compute_5[cse_var_160] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 15)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2304)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_161: int32 = (cse_var_1 + 160)
+ compute_5[cse_var_161] = (compute_5[cse_var_161] + (placeholder_1[((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16))]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2560)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_162: int32 = (cse_var_1 + 161)
+ compute_5[cse_var_162] = (compute_5[cse_var_162] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 1)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2560)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_163: int32 = (cse_var_1 + 162)
+ compute_5[cse_var_163] = (compute_5[cse_var_163] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 2)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2560)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_164: int32 = (cse_var_1 + 163)
+ compute_5[cse_var_164] = (compute_5[cse_var_164] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 3)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2560)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_165: int32 = (cse_var_1 + 164)
+ compute_5[cse_var_165] = (compute_5[cse_var_165] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 4)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2560)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_166: int32 = (cse_var_1 + 165)
+ compute_5[cse_var_166] = (compute_5[cse_var_166] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 5)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2560)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_167: int32 = (cse_var_1 + 166)
+ compute_5[cse_var_167] = (compute_5[cse_var_167] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 6)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2560)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_168: int32 = (cse_var_1 + 167)
+ compute_5[cse_var_168] = (compute_5[cse_var_168] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 7)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2560)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_169: int32 = (cse_var_1 + 168)
+ compute_5[cse_var_169] = (compute_5[cse_var_169] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 8)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2560)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_170: int32 = (cse_var_1 + 169)
+ compute_5[cse_var_170] = (compute_5[cse_var_170] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 9)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2560)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_171: int32 = (cse_var_1 + 170)
+ compute_5[cse_var_171] = (compute_5[cse_var_171] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 10)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2560)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_172: int32 = (cse_var_1 + 171)
+ compute_5[cse_var_172] = (compute_5[cse_var_172] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 11)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2560)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_173: int32 = (cse_var_1 + 172)
+ compute_5[cse_var_173] = (compute_5[cse_var_173] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 12)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2560)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_174: int32 = (cse_var_1 + 173)
+ compute_5[cse_var_174] = (compute_5[cse_var_174] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 13)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2560)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_175: int32 = (cse_var_1 + 174)
+ compute_5[cse_var_175] = (compute_5[cse_var_175] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 14)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2560)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_176: int32 = (cse_var_1 + 175)
+ compute_5[cse_var_176] = (compute_5[cse_var_176] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 15)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2560)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_177: int32 = (cse_var_1 + 176)
+ compute_5[cse_var_177] = (compute_5[cse_var_177] + (placeholder_1[((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16))]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2816)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_178: int32 = (cse_var_1 + 177)
+ compute_5[cse_var_178] = (compute_5[cse_var_178] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 1)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2816)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_179: int32 = (cse_var_1 + 178)
+ compute_5[cse_var_179] = (compute_5[cse_var_179] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 2)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2816)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_180: int32 = (cse_var_1 + 179)
+ compute_5[cse_var_180] = (compute_5[cse_var_180] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 3)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2816)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_181: int32 = (cse_var_1 + 180)
+ compute_5[cse_var_181] = (compute_5[cse_var_181] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 4)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2816)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_182: int32 = (cse_var_1 + 181)
+ compute_5[cse_var_182] = (compute_5[cse_var_182] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 5)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2816)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_183: int32 = (cse_var_1 + 182)
+ compute_5[cse_var_183] = (compute_5[cse_var_183] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 6)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2816)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_184: int32 = (cse_var_1 + 183)
+ compute_5[cse_var_184] = (compute_5[cse_var_184] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 7)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2816)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_185: int32 = (cse_var_1 + 184)
+ compute_5[cse_var_185] = (compute_5[cse_var_185] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 8)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2816)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_186: int32 = (cse_var_1 + 185)
+ compute_5[cse_var_186] = (compute_5[cse_var_186] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 9)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2816)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_187: int32 = (cse_var_1 + 186)
+ compute_5[cse_var_187] = (compute_5[cse_var_187] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 10)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2816)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_188: int32 = (cse_var_1 + 187)
+ compute_5[cse_var_188] = (compute_5[cse_var_188] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 11)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2816)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_189: int32 = (cse_var_1 + 188)
+ compute_5[cse_var_189] = (compute_5[cse_var_189] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 12)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2816)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_190: int32 = (cse_var_1 + 189)
+ compute_5[cse_var_190] = (compute_5[cse_var_190] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 13)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2816)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_191: int32 = (cse_var_1 + 190)
+ compute_5[cse_var_191] = (compute_5[cse_var_191] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 14)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2816)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_192: int32 = (cse_var_1 + 191)
+ compute_5[cse_var_192] = (compute_5[cse_var_192] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 15)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2816)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_193: int32 = (cse_var_1 + 192)
+ compute_5[cse_var_193] = (compute_5[cse_var_193] + (placeholder_1[((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16))]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3072)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_194: int32 = (cse_var_1 + 193)
+ compute_5[cse_var_194] = (compute_5[cse_var_194] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 1)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3072)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_195: int32 = (cse_var_1 + 194)
+ compute_5[cse_var_195] = (compute_5[cse_var_195] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 2)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3072)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_196: int32 = (cse_var_1 + 195)
+ compute_5[cse_var_196] = (compute_5[cse_var_196] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 3)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3072)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_197: int32 = (cse_var_1 + 196)
+ compute_5[cse_var_197] = (compute_5[cse_var_197] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 4)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3072)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_198: int32 = (cse_var_1 + 197)
+ compute_5[cse_var_198] = (compute_5[cse_var_198] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 5)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3072)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_199: int32 = (cse_var_1 + 198)
+ compute_5[cse_var_199] = (compute_5[cse_var_199] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 6)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3072)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_200: int32 = (cse_var_1 + 199)
+ compute_5[cse_var_200] = (compute_5[cse_var_200] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 7)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3072)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_201: int32 = (cse_var_1 + 200)
+ compute_5[cse_var_201] = (compute_5[cse_var_201] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 8)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3072)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_202: int32 = (cse_var_1 + 201)
+ compute_5[cse_var_202] = (compute_5[cse_var_202] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 9)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3072)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_203: int32 = (cse_var_1 + 202)
+ compute_5[cse_var_203] = (compute_5[cse_var_203] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 10)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3072)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_204: int32 = (cse_var_1 + 203)
+ compute_5[cse_var_204] = (compute_5[cse_var_204] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 11)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3072)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_205: int32 = (cse_var_1 + 204)
+ compute_5[cse_var_205] = (compute_5[cse_var_205] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 12)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3072)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_206: int32 = (cse_var_1 + 205)
+ compute_5[cse_var_206] = (compute_5[cse_var_206] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 13)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3072)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_207: int32 = (cse_var_1 + 206)
+ compute_5[cse_var_207] = (compute_5[cse_var_207] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 14)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3072)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_208: int32 = (cse_var_1 + 207)
+ compute_5[cse_var_208] = (compute_5[cse_var_208] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 15)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3072)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_209: int32 = (cse_var_1 + 208)
+ compute_5[cse_var_209] = (compute_5[cse_var_209] + (placeholder_1[((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16))]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3328)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_210: int32 = (cse_var_1 + 209)
+ compute_5[cse_var_210] = (compute_5[cse_var_210] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 1)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3328)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_211: int32 = (cse_var_1 + 210)
+ compute_5[cse_var_211] = (compute_5[cse_var_211] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 2)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3328)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_212: int32 = (cse_var_1 + 211)
+ compute_5[cse_var_212] = (compute_5[cse_var_212] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 3)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3328)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_213: int32 = (cse_var_1 + 212)
+ compute_5[cse_var_213] = (compute_5[cse_var_213] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 4)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3328)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_214: int32 = (cse_var_1 + 213)
+ compute_5[cse_var_214] = (compute_5[cse_var_214] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 5)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3328)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_215: int32 = (cse_var_1 + 214)
+ compute_5[cse_var_215] = (compute_5[cse_var_215] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 6)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3328)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_216: int32 = (cse_var_1 + 215)
+ compute_5[cse_var_216] = (compute_5[cse_var_216] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 7)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3328)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_217: int32 = (cse_var_1 + 216)
+ compute_5[cse_var_217] = (compute_5[cse_var_217] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 8)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3328)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_218: int32 = (cse_var_1 + 217)
+ compute_5[cse_var_218] = (compute_5[cse_var_218] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 9)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3328)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_219: int32 = (cse_var_1 + 218)
+ compute_5[cse_var_219] = (compute_5[cse_var_219] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 10)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3328)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_220: int32 = (cse_var_1 + 219)
+ compute_5[cse_var_220] = (compute_5[cse_var_220] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 11)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3328)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_221: int32 = (cse_var_1 + 220)
+ compute_5[cse_var_221] = (compute_5[cse_var_221] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 12)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3328)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_222: int32 = (cse_var_1 + 221)
+ compute_5[cse_var_222] = (compute_5[cse_var_222] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 13)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3328)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_223: int32 = (cse_var_1 + 222)
+ compute_5[cse_var_223] = (compute_5[cse_var_223] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 14)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3328)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_224: int32 = (cse_var_1 + 223)
+ compute_5[cse_var_224] = (compute_5[cse_var_224] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 15)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3328)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_225: int32 = (cse_var_1 + 224)
+ compute_5[cse_var_225] = (compute_5[cse_var_225] + (placeholder_1[((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16))]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3584)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_226: int32 = (cse_var_1 + 225)
+ compute_5[cse_var_226] = (compute_5[cse_var_226] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 1)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3584)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_227: int32 = (cse_var_1 + 226)
+ compute_5[cse_var_227] = (compute_5[cse_var_227] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 2)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3584)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_228: int32 = (cse_var_1 + 227)
+ compute_5[cse_var_228] = (compute_5[cse_var_228] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 3)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3584)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_229: int32 = (cse_var_1 + 228)
+ compute_5[cse_var_229] = (compute_5[cse_var_229] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 4)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3584)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_230: int32 = (cse_var_1 + 229)
+ compute_5[cse_var_230] = (compute_5[cse_var_230] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 5)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3584)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_231: int32 = (cse_var_1 + 230)
+ compute_5[cse_var_231] = (compute_5[cse_var_231] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 6)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3584)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_232: int32 = (cse_var_1 + 231)
+ compute_5[cse_var_232] = (compute_5[cse_var_232] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 7)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3584)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_233: int32 = (cse_var_1 + 232)
+ compute_5[cse_var_233] = (compute_5[cse_var_233] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 8)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3584)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_234: int32 = (cse_var_1 + 233)
+ compute_5[cse_var_234] = (compute_5[cse_var_234] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 9)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3584)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_235: int32 = (cse_var_1 + 234)
+ compute_5[cse_var_235] = (compute_5[cse_var_235] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 10)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3584)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_236: int32 = (cse_var_1 + 235)
+ compute_5[cse_var_236] = (compute_5[cse_var_236] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 11)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3584)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_237: int32 = (cse_var_1 + 236)
+ compute_5[cse_var_237] = (compute_5[cse_var_237] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 12)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3584)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_238: int32 = (cse_var_1 + 237)
+ compute_5[cse_var_238] = (compute_5[cse_var_238] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 13)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3584)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_239: int32 = (cse_var_1 + 238)
+ compute_5[cse_var_239] = (compute_5[cse_var_239] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 14)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3584)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_240: int32 = (cse_var_1 + 239)
+ compute_5[cse_var_240] = (compute_5[cse_var_240] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 15)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3584)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_241: int32 = (cse_var_1 + 240)
+ compute_5[cse_var_241] = (compute_5[cse_var_241] + (placeholder_1[((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16))]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3840)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_242: int32 = (cse_var_1 + 241)
+ compute_5[cse_var_242] = (compute_5[cse_var_242] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 1)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3840)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_243: int32 = (cse_var_1 + 242)
+ compute_5[cse_var_243] = (compute_5[cse_var_243] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 2)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3840)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_244: int32 = (cse_var_1 + 243)
+ compute_5[cse_var_244] = (compute_5[cse_var_244] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 3)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3840)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_245: int32 = (cse_var_1 + 244)
+ compute_5[cse_var_245] = (compute_5[cse_var_245] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 4)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3840)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_246: int32 = (cse_var_1 + 245)
+ compute_5[cse_var_246] = (compute_5[cse_var_246] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 5)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3840)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_247: int32 = (cse_var_1 + 246)
+ compute_5[cse_var_247] = (compute_5[cse_var_247] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 6)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3840)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_248: int32 = (cse_var_1 + 247)
+ compute_5[cse_var_248] = (compute_5[cse_var_248] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 7)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3840)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_249: int32 = (cse_var_1 + 248)
+ compute_5[cse_var_249] = (compute_5[cse_var_249] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 8)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3840)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_250: int32 = (cse_var_1 + 249)
+ compute_5[cse_var_250] = (compute_5[cse_var_250] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 9)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3840)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_251: int32 = (cse_var_1 + 250)
+ compute_5[cse_var_251] = (compute_5[cse_var_251] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 10)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3840)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_252: int32 = (cse_var_1 + 251)
+ compute_5[cse_var_252] = (compute_5[cse_var_252] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 11)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3840)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_253: int32 = (cse_var_1 + 252)
+ compute_5[cse_var_253] = (compute_5[cse_var_253] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 12)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3840)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_254: int32 = (cse_var_1 + 253)
+ compute_5[cse_var_254] = (compute_5[cse_var_254] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 13)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3840)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_255: int32 = (cse_var_1 + 254)
+ compute_5[cse_var_255] = (compute_5[cse_var_255] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 14)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3840)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_256: int32 = (cse_var_1 + 255)
+ compute_5[cse_var_256] = (compute_5[cse_var_256] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 15)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3840)], 0f32)))
}
}
}
}
for (i0.inner: int32, 0, 128) {
- for (i1.inner: int32, 0, 32) {
- let cse_var_4: int32 = (((i0.inner*512) + (i0.outer.i1.outer.fused*32)) + i1.inner)
- compute[cse_var_4] = max((compute_5[((i0.inner*32) + i1.inner)] + placeholder_4[cse_var_4]), 0f32)
- }
+ let cse_var_257: int32 = ((i0.inner*512) + (i0.outer.i1.outer.fused*16))
+ compute[ramp(cse_var_257, 1, 16)] = max((compute_5[ramp((i0.inner*16), 1, 16)] + placeholder_4[ramp(cse_var_257, 1, 16)]), broadcast(0f32, 16))
}
}
}
@@ -478,7 +1744,7 @@ We build the binary and check its correctness and performance.
.. code-block:: none
- Execution time of this operator: 1.443 ms
+ Execution time of this operator: 3.119 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 fba661fe2..e0a1c674e 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
@@ -5,10 +5,10 @@
Computation times
=================
-**00:43.512** total execution time for **how_to_tune_with_autotvm** files:
+**00:43.808** 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``) | 00:43.476 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autotvm_tune_conv2d_cuda.py` (``tune_conv2d_cuda.py``) | 00:43.772 | 0.0 MB |
+--------------------------------------------------------------------------------------------------+-----------+--------+
| :ref:`sphx_glr_how_to_tune_with_autotvm_tune_relay_x86.py` (``tune_relay_x86.py``) | 00:00.020 | 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 395946dfb..2974dae03 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
@@ -892,8 +892,8 @@ for this template
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 871, 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, 32]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 1, 128]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,2885496
- No: 6 GFLOPS: 42.45/42.45 result: MeasureResult(costs=(0.005453405210526316,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.6411683559417725, timestamp=1657505886.9299095) [('tile_f', [-1, 1, 1, 1]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 4, 4]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,3754080
- No: 7 GFLOPS: 0.00/42.45 result: Traceback (most recent call last):
+ No: 6 GFLOPS: 104.00/104.00 result: MeasureResult(costs=(0.002225995604166667,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.636479139328003, timestamp=1657533659.40923) [('tile_f', [-1, 1, 1, 1]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 4, 4]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,3754080
+ No: 7 GFLOPS: 0.00/104.00 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 588, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 540, in _build_func_common
@@ -1016,7 +1016,7 @@ for this template
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 871, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 1, 16, 32]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 256, 1]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,6225319
- No: 8 GFLOPS: 0.00/42.45 result: Traceback (most recent call last):
+ No: 8 GFLOPS: 0.00/104.00 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 588, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 540, in _build_func_common
@@ -1139,7 +1139,7 @@ for this template
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 871, 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, 32]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 8, 64]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,943546
- No: 9 GFLOPS: 0.00/42.45 result: Traceback (most recent call last):
+ No: 9 GFLOPS: 0.00/104.00 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 588, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 540, in _build_func_common
@@ -1262,7 +1262,7 @@ for this template
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 871, 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, 4]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 16, 32]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,2868708
- No: 10 GFLOPS: 0.00/42.45 result: Traceback (most recent call last):
+ No: 10 GFLOPS: 0.00/104.00 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 142, in build
res = future.result()
File "/usr/lib/python3.7/concurrent/futures/_base.py", line 435, in result
@@ -1280,7 +1280,7 @@ for this template
TimeoutError
[('tile_f', [-1, 32, 2, 4]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 4, 2]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,4691833
- No: 11 GFLOPS: 0.00/42.45 result: Traceback (most recent call last):
+ No: 11 GFLOPS: 0.00/104.00 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 588, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 540, in _build_func_common
@@ -1403,7 +1403,7 @@ for this template
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 871, 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, 64]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 4, 4]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,1042124
- No: 12 GFLOPS: 0.00/42.45 result: Traceback (most recent call last):
+ No: 12 GFLOPS: 0.00/104.00 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 588, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 540, in _build_func_common
@@ -1526,7 +1526,7 @@ for this template
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 871, 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, 4]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 32, 16]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,10013405
- No: 13 GFLOPS: 0.00/42.45 result: Traceback (most recent call last):
+ No: 13 GFLOPS: 0.00/104.00 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 588, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 540, in _build_func_common
@@ -1649,7 +1649,7 @@ for this template
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 871, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 8, 8, 2]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 4, 32]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,6732082
- No: 14 GFLOPS: 0.00/42.45 result: Traceback (most recent call last):
+ No: 14 GFLOPS: 0.00/104.00 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 588, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 540, in _build_func_common
@@ -1772,7 +1772,7 @@ for this template
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 871, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 4, 32]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 4, 128]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,7536735
- No: 15 GFLOPS: 0.00/42.45 result: Traceback (most recent call last):
+ No: 15 GFLOPS: 0.00/104.00 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 588, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 540, in _build_func_common
@@ -1895,7 +1895,7 @@ for this template
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 871, 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, 4]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 128, 4]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,482121
- No: 16 GFLOPS: 0.00/42.45 result: Traceback (most recent call last):
+ No: 16 GFLOPS: 0.00/104.00 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 588, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 540, in _build_func_common
@@ -2018,7 +2018,7 @@ for this template
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 871, 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, 16]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 32, 8]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,2824525
- No: 17 GFLOPS: 0.00/42.45 result: Traceback (most recent call last):
+ No: 17 GFLOPS: 0.00/104.00 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 588, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 540, in _build_func_common
@@ -2141,7 +2141,7 @@ for this template
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 871, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 64, 1, 1]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 8, 8]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,4559286
- No: 18 GFLOPS: 0.00/42.45 result: Traceback (most recent call last):
+ No: 18 GFLOPS: 0.00/104.00 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 588, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 540, in _build_func_common
@@ -2264,7 +2264,7 @@ for this template
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 871, 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, 32, 16]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 1, 512]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,9677544
- No: 19 GFLOPS: 0.00/42.45 result: Traceback (most recent call last):
+ No: 19 GFLOPS: 0.00/104.00 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 738, in __call__
yield remote, remote.load_module(os.path.split(build_result.filename)[1])
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 702, in run_through_rpc
@@ -2352,7 +2352,7 @@ for this template
15: _PyEval_EvalFrameDefault
14: 0x0000000000537c30
13: _PyObject_FastCallKeywords
- 12: 0x00007ff304e22fa2
+ 12: 0x00007f0ddc2e4fa2
11: _ctypes_callproc
10: ffi_call
9: ffi_call_unix64
@@ -2417,7 +2417,7 @@ for this template
21: _PyFunction_FastCallKeywords
20: _PyEval_EvalFrameDefault
19: _PyFunction_FastCall [('tile_f', [-1, 8, 2, 16]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 1, 1]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,6390073
- No: 20 GFLOPS: 142.88/142.88 result: MeasureResult(costs=(0.00162021567,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.442824363708496, timestamp=1657505912.7780535) [('tile_f', [-1, 1, 4, 1]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 4, 1]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,9881539
+ No: 20 GFLOPS: 143.59/143.59 result: MeasureResult(costs=(0.00161221691,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.4332330226898193, timestamp=1657533685.933291) [('tile_f', [-1, 1, 4, 1]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 4, 1]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,9881539
@@ -2474,7 +2474,7 @@ and measure running time.
Best config:
[('tile_f', [-1, 1, 4, 1]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 4, 1]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,9881539
Finish loading 20 records
- Time cost of this operator: 0.002013
+ Time cost of this operator: 0.001964
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 4a1811cde..e51a988cf 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
@@ -329,10 +329,10 @@ Timing the untuned program
########## Build without Autotuning ##########
Node Name Ops Time(us) Time(%) Shape Inputs Outputs Measurements(us)
--------- --- -------- ------- ----- ------ ------- ----------------
- tvmgen_default_fused_nn_contrib_conv2d_NCHWc tvmgen_default_fused_nn_contrib_conv2d_NCHWc 315.1 98.729 (1, 2, 10, 10, 3) 2 1 [315.1]
- tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 3.064 0.96 (1, 6, 10, 10) 1 1 [3.064]
- tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.994 0.311 (1, 1, 10, 10, 3) 1 1 [0.994]
- Total_time - 319.158 - - - - -
+ tvmgen_default_fused_nn_contrib_conv2d_NCHWc tvmgen_default_fused_nn_contrib_conv2d_NCHWc 308.2 98.734 (1, 2, 10, 10, 3) 2 1 [308.2]
+ tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 2.983 0.956 (1, 6, 10, 10) 1 1 [2.983]
+ tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.969 0.31 (1, 1, 10, 10, 3) 1 1 [0.969]
+ Total_time - 312.152 - - - - -
@@ -398,10 +398,10 @@ 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 209.7 98.55 (1, 1, 10, 10, 6) 2 1 [209.7]
- tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 2.21 1.039 (1, 6, 10, 10) 1 1 [2.21]
- tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.875 0.411 (1, 3, 10, 10, 1) 1 1 [0.875]
- Total_time - 212.785 - - - - -
+ tvmgen_default_fused_nn_contrib_conv2d_NCHWc tvmgen_default_fused_nn_contrib_conv2d_NCHWc 79.562 96.684 (1, 6, 10, 10, 1) 2 1 [79.562]
+ tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 1.713 2.082 (1, 6, 10, 10) 1 1 [1.713]
+ tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 1.015 1.234 (1, 1, 10, 10, 3) 1 1 [1.015]
+ Total_time - 82.291 - - - - -
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 c85f5fb94..f7f0af986 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
@@ -225,7 +225,7 @@ take about **2 minutes** to download the Stanford Cars, while COCO 2017 validati
.. code-block:: none
- '/tmp/tmpcd8odi49/images/random'
+ '/tmp/tmpj7dioiiz/images/random'
@@ -325,8 +325,8 @@ objects to other stuff? We can display some examples from our datasets using ``m
.. code-block:: none
- /tmp/tmpcd8odi49/images/target contains 8144 images
- /tmp/tmpcd8odi49/images/random contains 5000 images
+ /tmp/tmpj7dioiiz/images/target contains 8144 images
+ /tmp/tmpj7dioiiz/images/random contains 5000 images
@@ -501,13 +501,13 @@ the time on our validation set).
.. code-block:: none
Epoch 1/3
- 328/328 - 55s - loss: 0.2250 - accuracy: 0.9231 - val_loss: 0.1352 - val_accuracy: 0.9588
+ 328/328 - 53s - loss: 0.2160 - accuracy: 0.9251 - val_loss: 0.1530 - val_accuracy: 0.9494
Epoch 2/3
- 328/328 - 53s - loss: 0.0951 - accuracy: 0.9639 - val_loss: 0.1336 - val_accuracy: 0.9622
+ 328/328 - 50s - loss: 0.0966 - accuracy: 0.9638 - val_loss: 0.1054 - val_accuracy: 0.9641
Epoch 3/3
- 328/328 - 53s - loss: 0.0654 - accuracy: 0.9757 - val_loss: 0.1372 - val_accuracy: 0.9558
+ 328/328 - 50s - loss: 0.0623 - accuracy: 0.9769 - val_loss: 0.1160 - val_accuracy: 0.9637
- <keras.callbacks.History object at 0x7f486eff1790>
+ <keras.callbacks.History object at 0x7fd72d0ebe10>
@@ -864,7 +864,7 @@ 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 51.961 seconds)
+ **Total running time of the script:** ( 4 minutes 50.207 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 7c62ca045..1bf8215b0 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
@@ -5,14 +5,14 @@
Computation times
=================
-**05:39.131** total execution time for **how_to_work_with_microtvm** files:
+**05:36.322** total execution time for **how_to_work_with_microtvm** files:
+---------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_microtvm_micro_train.py` (``micro_train.py``) | 04:51.961 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_microtvm_micro_train.py` (``micro_train.py``) | 04:50.207 | 0.0 MB |
+---------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_microtvm_micro_autotune.py` (``micro_autotune.py``) | 00:43.801 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_microtvm_micro_autotune.py` (``micro_autotune.py``) | 00:42.750 | 0.0 MB |
+---------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_microtvm_micro_tflite.py` (``micro_tflite.py``) | 00:03.368 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_microtvm_micro_tflite.py` (``micro_tflite.py``) | 00:03.362 | 0.0 MB |
+---------------------------------------------------------------------------------------------+-----------+--------+
| :ref:`sphx_glr_how_to_work_with_microtvm_micro_ethosu.py` (``micro_ethosu.py``) | 00:00.001 | 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 55a586d50..adb35f0a5 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
@@ -5,12 +5,12 @@
Computation times
=================
-**00:11.307** total execution time for **how_to_work_with_relay** files:
+**00:11.582** total execution time for **how_to_work_with_relay** files:
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_relay_using_external_lib.py` (``using_external_lib.py``) | 00:09.789 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_relay_using_external_lib.py` (``using_external_lib.py``) | 00:10.125 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_relay_build_gcn.py` (``build_gcn.py``) | 00:01.512 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_relay_build_gcn.py` (``build_gcn.py``) | 00:01.451 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
| :ref:`sphx_glr_how_to_work_with_relay_using_relay_viz.py` (``using_relay_viz.py``) | 00:00.006 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/how_to/work_with_schedules/intrin_math.rst.txt b/docs/_sources/how_to/work_with_schedules/intrin_math.rst.txt
index 8dfe20ed4..f203abe6b 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
@@ -261,7 +261,7 @@ The following example customizes CUDA lowering rule for :code:`exp`.
.. code-block:: none
- <function my_cuda_math_rule at 0x7f4969c349e0>
+ <function my_cuda_math_rule at 0x7fd7381487a0>
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 af30b0966..2ae423249 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
=================
-**00:04.024** total execution time for **how_to_work_with_schedules** files:
+**00:04.105** total execution time for **how_to_work_with_schedules** files:
+------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_intrin_math.py` (``intrin_math.py``) | 00:01.875 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_intrin_math.py` (``intrin_math.py``) | 00:01.904 | 0.0 MB |
+------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_tensorize.py` (``tensorize.py``) | 00:00.944 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_tensorize.py` (``tensorize.py``) | 00:00.973 | 0.0 MB |
+------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_reduction.py` (``reduction.py``) | 00:00.522 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_reduction.py` (``reduction.py``) | 00:00.531 | 0.0 MB |
+------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_scan.py` (``scan.py``) | 00:00.506 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_scan.py` (``scan.py``) | 00:00.517 | 0.0 MB |
+------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_extern_op.py` (``extern_op.py``) | 00:00.100 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_extern_op.py` (``extern_op.py``) | 00:00.102 | 0.0 MB |
+------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_schedule_primitives.py` (``schedule_primitives.py``) | 00:00.035 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_schedule_primitives.py` (``schedule_primitives.py``) | 00:00.036 | 0.0 MB |
+------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_tedd.py` (``tedd.py``) | 00:00.027 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_tedd.py` (``tedd.py``) | 00:00.028 | 0.0 MB |
+------------------------------------------------------------------------------------------------+-----------+--------+
| :ref:`sphx_glr_how_to_work_with_schedules_tuple_inputs.py` (``tuple_inputs.py``) | 00:00.014 | 0.0 MB |
+------------------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/how_to/work_with_schedules/tensorize.rst.txt b/docs/_sources/how_to/work_with_schedules/tensorize.rst.txt
index ef9c0ebeb..e1a936692 100644
--- a/docs/_sources/how_to/work_with_schedules/tensorize.rst.txt
+++ b/docs/_sources/how_to/work_with_schedules/tensorize.rst.txt
@@ -347,7 +347,7 @@ The importing needs to happen before the tensorized GEMV being executed.
C: Buffer(C_2: Pointer(float32), float32, [524288], [])}
buffer_map = {A_1: A, B_1: B, C_1: C}
preflattened_buffer_map = {A_1: A_3: Buffer(A_2, float32, [1024, 64], []), B_1: B_3: Buffer(B_2, float32, [512, 64], []), C_1: C_3: Buffer(C_2, float32, [1024, 512], [])} {
- attr [IterVar(i: int32, (nullptr), "DataPar", "")] "pragma_import_llvm" = "; ModuleID = '/tmp/tmpi6dlqhp0/input0.cc'\nsource_filename = \"/tmp/tmpi6dlqhp0/input0.cc\"\ntarget datalayout = \"e-m:e-i64:64-f80:128-n8:16:32:64-S128\"\ntarget triple = \"x86_64-pc-linux-gnu\"\n\n; Function Attrs: noinline nounwind optnone uwtable\ndefine dso_local i32 @gemv_update(float*, float*, float*, i32, i32, i32) #0 {\n %7 = alloca float*, align 8\n %8 = alloca float*, align 8\n %9 = alloca floa [...]
+ attr [IterVar(i: int32, (nullptr), "DataPar", "")] "pragma_import_llvm" = "; ModuleID = '/tmp/tmp91ffoxoj/input0.cc'\nsource_filename = \"/tmp/tmp91ffoxoj/input0.cc\"\ntarget datalayout = \"e-m:e-i64:64-f80:128-n8:16:32:64-S128\"\ntarget triple = \"x86_64-pc-linux-gnu\"\n\n; Function Attrs: noinline nounwind optnone uwtable\ndefine dso_local i32 @gemv_update(float*, float*, float*, i32, i32, i32) #0 {\n %7 = alloca float*, align 8\n %8 = alloca float*, align 8\n %9 = alloca floa [...]
for (i, 0, 1024) {
for (j.outer: int32, 0, 32) {
@tir.call_extern("gemv_update", @tir.tvm_access_ptr(@tir.type_annotation(, dtype=float32), C_2, ((i*512) + (j.outer*16)), 16, 2, dtype=handle), @tir.tvm_access_ptr(@tir.type_annotation(, dtype=float32), A_2, (i*64), 64, 1, dtype=handle), @tir.tvm_access_ptr(@tir.type_annotation(, dtype=float32), B_2, (j.outer*1024), 1024, 1, dtype=handle), 16, 64, 64, dtype=int32)
diff --git a/docs/_sources/install/from_source.rst.txt b/docs/_sources/install/from_source.rst.txt
index 92795da9a..4b5810cb2 100644
--- a/docs/_sources/install/from_source.rst.txt
+++ b/docs/_sources/install/from_source.rst.txt
@@ -323,7 +323,7 @@ like ``virtualenv``.
.. code:: bash
- pip3 install --user tornado psutil xgboost cloudpickle
+ pip3 install --user tornado psutil 'xgboost<1.6.0' cloudpickle
Note on M1 macs, you may have trouble installing xgboost / scipy. scipy and xgboost requires some additional dependencies to be installed,
including openblas and its dependencies. Use the following commands to install scipy and xgboost with the required dependencies and
@@ -339,7 +339,7 @@ configuration. A workaround for this is to do the following commands:
pip install scipy --no-use-pep517
- pip install xgboost
+ pip install 'xgboost<1.6.0'
Install Contrib Libraries
-------------------------
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 361c03c1a..e96ed8d5c 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
=================
-**00:21.636** total execution time for **topic_vta_tutorials_autotvm** files:
+**00:20.282** total execution time for **topic_vta_tutorials_autotvm** files:
+---------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_autotvm_tune_relay_vta.py` (``tune_relay_vta.py``) | 00:21.630 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_autotvm_tune_relay_vta.py` (``tune_relay_vta.py``) | 00:20.276 | 0.0 MB |
+---------------------------------------------------------------------------------------+-----------+--------+
| :ref:`sphx_glr_topic_vta_tutorials_autotvm_tune_alu_vta.py` (``tune_alu_vta.py``) | 00:00.006 | 0.0 MB |
+---------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/topic/vta/tutorials/frontend/deploy_classification.rst.txt b/docs/_sources/topic/vta/tutorials/frontend/deploy_classification.rst.txt
index 025c44213..d72e9043c 100644
--- a/docs/_sources/topic/vta/tutorials/frontend/deploy_classification.rst.txt
+++ b/docs/_sources/topic/vta/tutorials/frontend/deploy_classification.rst.txt
@@ -291,7 +291,7 @@ 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 22.61s!
+ resnet18_v1 inference graph built in 21.65s!
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 8a5757fc8..5c33df92a 100644
--- a/docs/_sources/topic/vta/tutorials/frontend/deploy_detection.rst.txt
+++ b/docs/_sources/topic/vta/tutorials/frontend/deploy_detection.rst.txt
@@ -335,7 +335,7 @@ The compilation steps are:
"target_host parameter is going to be deprecated. "
/workspace/python/tvm/relay/build_module.py:411: DeprecationWarning: Please use input parameter mod (tvm.IRModule) instead of deprecated parameter mod (tvm.relay.function.Function)
DeprecationWarning,
- yolov3-tiny inference graph built in 16.30s!
+ yolov3-tiny inference graph built in 15.23s!
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 1ca7fd63e..53ba7dc85 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
=================
-**01:31.802** total execution time for **topic_vta_tutorials_frontend** files:
+**01:29.450** total execution time for **topic_vta_tutorials_frontend** files:
+------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_frontend_deploy_detection.py` (``deploy_detection.py``) | 00:48.976 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_frontend_deploy_detection.py` (``deploy_detection.py``) | 00:47.792 | 0.0 MB |
+------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_frontend_deploy_classification.py` (``deploy_classification.py``) | 00:42.826 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_frontend_deploy_classification.py` (``deploy_classification.py``) | 00:41.658 | 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 b2651b284..ee1d37f06 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
=================
-**00:03.214** total execution time for **topic_vta_tutorials_optimize** files:
+**00:03.181** total execution time for **topic_vta_tutorials_optimize** files:
+--------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_optimize_convolution_opt.py` (``convolution_opt.py``) | 00:02.815 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_optimize_convolution_opt.py` (``convolution_opt.py``) | 00:02.786 | 0.0 MB |
+--------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_optimize_matrix_multiply_opt.py` (``matrix_multiply_opt.py``) | 00:00.399 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_optimize_matrix_multiply_opt.py` (``matrix_multiply_opt.py``) | 00:00.395 | 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 bd23f948c..821e533b9 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
=================
-**00:00.703** total execution time for **topic_vta_tutorials** files:
+**00:00.707** total execution time for **topic_vta_tutorials** files:
+---------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_matrix_multiply.py` (``matrix_multiply.py``) | 00:00.373 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_matrix_multiply.py` (``matrix_multiply.py``) | 00:00.377 | 0.0 MB |
+---------------------------------------------------------------------------------+-----------+--------+
| :ref:`sphx_glr_topic_vta_tutorials_vta_get_started.py` (``vta_get_started.py``) | 00:00.330 | 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 7be4c0eb8..20465e9b3 100644
--- a/docs/_sources/tutorial/auto_scheduler_matmul_x86.rst.txt
+++ b/docs/_sources/tutorial/auto_scheduler_matmul_x86.rst.txt
@@ -328,7 +328,7 @@ We build the binary and check its correctness and performance.
.. code-block:: none
- Execution time of this operator: 93.466 ms
+ Execution time of this operator: 93.361 ms
diff --git a/docs/_sources/tutorial/autotvm_matmul_x86.rst.txt b/docs/_sources/tutorial/autotvm_matmul_x86.rst.txt
index b6772246a..29db51004 100644
--- a/docs/_sources/tutorial/autotvm_matmul_x86.rst.txt
+++ b/docs/_sources/tutorial/autotvm_matmul_x86.rst.txt
@@ -462,16 +462,16 @@ reduce variance, we take 5 measurements and average them.
waiting for device...
device available
Get devices for measurement successfully!
- No: 1 GFLOPS: 9.21/9.21 result: MeasureResult(costs=(0.029154213800000002,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.6001706123352051, timestamp=1657504753.2182171) [('tile_y', [-1, 1]), ('tile_x', [-1, 256])],None,80
- No: 2 GFLOPS: 2.47/9.21 result: MeasureResult(costs=(0.10880303499999999,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.8876526355743408, timestamp=1657504755.12157) [('tile_y', [-1, 4]), ('tile_x', [-1, 8])],None,32
- No: 3 GFLOPS: 11.75/11.75 result: MeasureResult(costs=(0.022841615200000002,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.575455904006958, timestamp=1657504756.1836452) [('tile_y', [-1, 64]), ('tile_x', [-1, 32])],None,56
- No: 4 GFLOPS: 1.89/11.75 result: MeasureResult(costs=(0.142067211,), error_no=MeasureErrorNo.NO_ERROR, all_cost=2.393825054168701, timestamp=1657504759.1453855) [('tile_y', [-1, 1]), ('tile_x', [-1, 4])],None,20
- No: 5 GFLOPS: 3.66/11.75 result: MeasureResult(costs=(0.07333695659999999,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.3122856616973877, timestamp=1657504760.5847883) [('tile_y', [-1, 256]), ('tile_x', [-1, 16])],None,48
- No: 6 GFLOPS: 1.71/11.75 result: MeasureResult(costs=(0.157348668,), error_no=MeasureErrorNo.NO_ERROR, all_cost=2.6468584537506104, timestamp=1657504763.8008442) [('tile_y', [-1, 512]), ('tile_x', [-1, 4])],None,29
- No: 7 GFLOPS: 0.87/11.75 result: MeasureResult(costs=(0.30696147259999995,), error_no=MeasureErrorNo.NO_ERROR, all_cost=5.027516603469849, timestamp=1657504768.8753772) [('tile_y', [-1, 512]), ('tile_x', [-1, 2])],None,19
- No: 8 GFLOPS: 10.48/11.75 result: MeasureResult(costs=(0.025606279599999997,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.5558183193206787, timestamp=1657504769.4442372) [('tile_y', [-1, 4]), ('tile_x', [-1, 64])],None,62
- No: 9 GFLOPS: 1.90/11.75 result: MeasureResult(costs=(0.14157979299999998,), error_no=MeasureErrorNo.NO_ERROR, all_cost=2.366572380065918, timestamp=1657504771.9293792) [('tile_y', [-1, 2]), ('tile_x', [-1, 2])],None,11
- No: 10 GFLOPS: 2.80/11.75 result: MeasureResult(costs=(0.0959876314,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.6463425159454346, timestamp=1657504773.6292558) [('tile_y', [-1, 4]), ('tile_x', [-1, 4])],None,22
+ No: 1 GFLOPS: 9.98/9.98 result: MeasureResult(costs=(0.0268887274,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.5620629787445068, timestamp=1657532521.2429855) [('tile_y', [-1, 1]), ('tile_x', [-1, 256])],None,80
+ No: 2 GFLOPS: 2.92/9.98 result: MeasureResult(costs=(0.0917741092,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.616034746170044, timestamp=1657532523.4018276) [('tile_y', [-1, 4]), ('tile_x', [-1, 8])],None,32
+ No: 3 GFLOPS: 11.82/11.82 result: MeasureResult(costs=(0.022714058000000002,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.5528194904327393, timestamp=1657532524.4561114) [('tile_y', [-1, 64]), ('tile_x', [-1, 32])],None,56
+ No: 4 GFLOPS: 1.85/11.82 result: MeasureResult(costs=(0.1448481544,), error_no=MeasureErrorNo.NO_ERROR, all_cost=2.4391255378723145, timestamp=1657532526.9392235) [('tile_y', [-1, 1]), ('tile_x', [-1, 4])],None,20
+ No: 5 GFLOPS: 3.63/11.82 result: MeasureResult(costs=(0.0739733334,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.323929786682129, timestamp=1657532528.3962824) [('tile_y', [-1, 256]), ('tile_x', [-1, 16])],None,48
+ No: 6 GFLOPS: 1.68/11.82 result: MeasureResult(costs=(0.1594500364,), error_no=MeasureErrorNo.NO_ERROR, all_cost=2.713550090789795, timestamp=1657532531.1562076) [('tile_y', [-1, 512]), ('tile_x', [-1, 4])],None,29
+ No: 7 GFLOPS: 0.87/11.82 result: MeasureResult(costs=(0.3098437456,), error_no=MeasureErrorNo.NO_ERROR, all_cost=5.076806306838989, timestamp=1657532536.8197832) [('tile_y', [-1, 512]), ('tile_x', [-1, 2])],None,19
+ No: 8 GFLOPS: 10.46/11.82 result: MeasureResult(costs=(0.0256587246,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.5558922290802002, timestamp=1657532537.393337) [('tile_y', [-1, 4]), ('tile_x', [-1, 64])],None,62
+ No: 9 GFLOPS: 1.87/11.82 result: MeasureResult(costs=(0.1434236372,), error_no=MeasureErrorNo.NO_ERROR, all_cost=2.3893957138061523, timestamp=1657532539.9031358) [('tile_y', [-1, 2]), ('tile_x', [-1, 2])],None,11
+ No: 10 GFLOPS: 2.35/11.82 result: MeasureResult(costs=(0.1143380446,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.935478687286377, timestamp=1657532541.8986776) [('tile_y', [-1, 4]), ('tile_x', [-1, 4])],None,22
diff --git a/docs/_sources/tutorial/autotvm_relay_x86.rst.txt b/docs/_sources/tutorial/autotvm_relay_x86.rst.txt
index ebc420065..ed87bf209 100644
--- a/docs/_sources/tutorial/autotvm_relay_x86.rst.txt
+++ b/docs/_sources/tutorial/autotvm_relay_x86.rst.txt
@@ -327,7 +327,7 @@ standard deviation.
.. code-block:: none
- {'mean': 497.285720330002, 'median': 497.179591150001, 'std': 0.4704793355308102}
+ {'mean': 500.172581780007, 'median': 500.1141345000178, 'std': 0.43971723624319015}
@@ -563,31 +563,31 @@ the tuning data to.
/workspace/python/tvm/driver/build_module.py:268: UserWarning: target_host parameter is going to be deprecated. Please pass in tvm.target.Target(target, host=target_host) instead.
"target_host parameter is going to be deprecated. "
-
[Task 1/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 1/25] Current/Best: 17.51/ 17.51 GFLOPS | Progress: (4/20) | 6.29 s
[Task 1/25] Current/Best: 6.16/ 17.51 GFLOPS | Progress: (8/20) | 9.30 s
[Task 1/25] Current/Best: 11.50/ 22.70 GFLOPS | Progress: (12/20) | 11.81 s
[Task 1/25] Current/Best: 16.86/ 22.73 GFLOPS | Progress: (16/20) | 13.49 s
[Task 1/25] Current/Best: 11.55/ 23.86 GFLOPS | Progress: (20/20) | 15.24 s Done.
-
[Task 2/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 2/25] Current/Best: 12.27/ 13.22 GFLOPS | Progress: (4/20) | 3.98 s
[Task 2/25] Current/Best: 14.17/ 18.53 GFLOPS | Progress: (8/20) | 5.28 s
[Task 2/25] Current/Best: 20.79/ 20.79 GFLOPS | Progress: (12/20) | 6.60 s
[Task 2/25] Current/Best: 12.55/ 20.79 GFLOPS | Progress: (16/20) | 7.86 s
[Task 2/25] Current/Best: 19.44/ 20.79 GFLOPS | Progress: (20/20) | 9.48 s Done.
-
[Task 3/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 3/25] Current/Best: 1.63/ 10.54 GFLOPS | Progress: (4/20) | 5.87 s
[Task 3/25] Current/Best: 15.56/ 16.87 GFLOPS | Progress: (8/20) | 7.78 s
[Task 3/25] Current/Best: 14.89/ 16.87 GFLOPS | Progress: (12/20) | 9.52 s
[Task 3/25] Current/Best: 7.20/ 23.40 GFLOPS | Progress: (16/20) | 11.43 s
[Task 3/25] Current/Best: 12.49/ 23.40 GFLOPS | Progress: (20/20) | 16.06 s Done.
-
[Task 4/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 4/25] Current/Best: 9.56/ 20.53 GFLOPS | Progress: (4/20) | 2.41 s
[Task 4/25] Current/Best: 6.67/ 20.53 GFLOPS | Progress: (8/20) | 7.25 s
[Task 4/25] Current/Best: 22.07/ 22.07 GFLOPS | Progress: (12/20) | 12.33 s
[Task 4/25] Current/Best: 17.38/ 22.07 GFLOPS | Progress: (16/20) | 14.76 s
[Task 4/25] Current/Best: 13.21/ 22.07 GFLOPS | Progress: (20/20) | 16.74 s Done.
-
[Task 5/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 5/25] Current/Best: 9.60/ 10.22 GFLOPS | Progress: (4/20) | 2.58 s
[Task 5/25] Current/Best: 11.73/ 12.90 GFLOPS | Progress: (8/20) | 4.64 s
[Task 5/25] Current/Best: 11.08/ 17.98 GFLOPS | Progress: (12/20) | 7.87 s
[Task 5/25] Current/Best: 11.56/ 22.50 GFLOPS | Progress: (16/20) | 9.29 s
[Task 5/25] Current/Best: 11.99/ 22.50 GFLOPS | Progress: (20/20) | 11.17 s Done.
-
[Task 6/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 6/25] Current/Best: 12.22/ 20.67 GFLOPS | Progress: (4/20) | 4.12 s
[Task 6/25] Current/Best: 19.02/ 20.67 GFLOPS | Progress: (8/20) | 5.88 s
[Task 6/25] Current/Best: 13.30/ 20.67 GFLOPS | Progress: (12/20) | 7.84 s
[Task 6/25] Current/Best: 19.94/ 20.67 GFLOPS | Progress: (16/20) | 10.10 s
[Task 6/25] Current/Best: 3.69/ 20.67 GFLOPS | Progress: (20/20) | 12.60 s Done.
-
[Task 7/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 7/25] Current/Best: 11.24/ 12.97 GFLOPS | Progress: (4/20) | 3.65 s
[Task 7/25] Current/Best: 20.22/ 20.85 GFLOPS | Progress: (8/20) | 5.18 s
[Task 7/25] Current/Best: 16.13/ 20.92 GFLOPS | Progress: (12/20) | 7.10 s
[Task 7/25] Current/Best: 12.24/ 20.92 GFLOPS | Progress: (16/20) | 9.13 s
[Task 7/25] Current/Best: 6.40/ 21.67 GFLOPS | Progress: (20/20) | 11.58 s Done.
-
[Task 8/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 8/25] Current/Best: 10.00/ 13.80 GFLOPS | Progress: (4/20) | 2.95 s
[Task 8/25] Current/Best: 9.51/ 13.80 GFLOPS | Progress: (8/20) | 8.18 s
[Task 8/25] Current/Best: 12.48/ 13.80 GFLOPS | Progress: (12/20) | 14.80 s
[Task 8/25] Current/Best: 19.06/ 19.06 GFLOPS | Progress: (16/20) | 16.91 s
[Task 8/25] Current/Best: 19.84/ 19.84 GFLOPS | Progress: (20/20) | 24.04 s Done.
-
[Task 9/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 9/25] Current/Best: 14.38/ 15.47 GFLOPS | Progress: (4/20) | 11.98 s
[Task 9/25] Current/Best: 23.37/ 23.37 GFLOPS | Progress: (8/20) | 13.78 s
[Task 9/25] Current/Best: 8.23/ 23.37 GFLOPS | Progress: (12/20) | 16.35 s
[Task 9/25] Current/Best: 17.95/ 23.37 GFLOPS | Progress: (16/20) | 19.22 s
[Task 9/25] Current/Best: 8.96/ 23.37 GFLOPS | Progress: (20/20) | 28.08 s
[Task 10/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 10/25] Current/Best: 18.27/ 18.27 GFLOPS | Progress: (4/20) | 2.60 s
[Task 10/25] Current/Best: 15.45/ 18.27 GFLOPS | Progress: (8/20) | 4.24 s
[Task 10/25] Current/Best: 12.85/ 18.80 GFLOPS | Progress: (12/20) | 5.78 s
[Task 10/25] Current/Best: 19.16/ 20.16 GFLOPS | Progress: (16/20) | 6.90 s
[Task 10/25] Current/Best: 8.96/ 20.16 GFLOPS | Progress: (20/20
) | 8.45 s Done.
-
[Task 11/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 11/25] Current/Best: 12.05/ 18.11 GFLOPS | Progress: (4/20) | 3.39 s
[Task 11/25] Current/Best: 16.84/ 18.11 GFLOPS | Progress: (8/20) | 6.23 s
[Task 11/25] Current/Best: 17.09/ 18.11 GFLOPS | Progress: (12/20) | 8.33 s
[Task 11/25] Current/Best: 12.22/ 21.15 GFLOPS | Progress: (16/20) | 11.35 s
[Task 11/25] Current/Best: 19.40/ 21.56 GFLOPS | Progress: (20/20) | 13.48 s Done.
-
[Task 12/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 12/25] Current/Best: 7.78/ 17.99 GFLOPS | Progress: (4/20) | 5.88 s
[Task 12/25] Current/Best: 5.24/ 17.99 GFLOPS | Progress: (8/20) | 9.85 s
[Task 12/25] Current/Best: 19.26/ 19.26 GFLOPS | Progress: (12/20) | 11.83 s
[Task 12/25] Current/Best: 13.35/ 19.26 GFLOPS | Progress: (16/20) | 14.79 s
[Task 12/25] Current/Best: 15.11/ 19.26 GFLOPS | Progress: (20/20) | 16.69 s Done.
-
[Task 13/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 13/25] Current/Best: 8.49/ 17.27 GFLOPS | Progress: (4/20) | 3.79 s
[Task 13/25] Current/Best: 16.06/ 20.76 GFLOPS | Progress: (8/20) | 6.42 s
[Task 13/25] Current/Best: 19.39/ 21.82 GFLOPS | Progress: (12/20) | 9.51 s
[Task 13/25] Current/Best: 12.21/ 21.82 GFLOPS | Progress: (16/20) | 13.09 s
[Task 13/25] Current/Best: 17.53/ 21.82 GFLOPS | Progress: (20/20) | 15.40 s Done.
-
[Task 14/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 14/25] Current/Best: 13.58/ 13.58 GFLOPS | Progress: (4/20) | 3.39 s
[Task 14/25] Current/Best: 6.05/ 13.58 GFLOPS | Progress: (8/20) | 5.60 s
[Task 14/25] Current/Best: 20.03/ 20.03 GFLOPS | Progress: (12/20) | 8.28 s
[Task 14/25] Current/Best: 16.59/ 20.03 GFLOPS | Progress: (16/20) | 9.93 s Done.
-
[Task 14/25] Current/Best: 16.98/ 20.03 GFLOPS | Progress: (20/20) | 11.70 s
[Task 15/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 15/25] Current/Best: 16.10/ 17.58 GFLOPS | Progress: (4/20) | 2.75 s
[Task 15/25] Current/Best: 14.34/ 17.94 GFLOPS | Progress: (8/20) | 4.10 s
[Task 15/25] Current/Best: 10.25/ 22.00 GFLOPS | Progress: (12/20) | 6.38 s
[Task 15/25] Current/Best: 20.29/ 22.00 GFLOPS | Progress: (16/20) | 9.66 s
[Task 15/25] Current/Best: 9.69/ 22.00 GFLOPS | Progress: (20/20) | 10.69 s
[Task 16/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 16/25] Current/Best: 20.56/ 20.56 GFLOPS | Progress: (4/20) | 3.00 s
[Task 16/25] Current/Best: 3.04/ 20.56 GFLOPS | Progress: (8/20) | 4.61 s
[Task 16/25] Current/Best: 19.56/ 20.56 GFLOPS | Progress: (12/20) | 5.83 s
[Task 16/25] Current/Best: 18.07/ 20.56 GFLOPS | Progress: (16/20) |
7.21 s
[Task 16/25] Current/Best: 10.01/ 22.19 GFLOPS | Progress: (20/20) | 9.40 s Done.
-
[Task 17/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 17/25] Current/Best: 13.18/ 18.83 GFLOPS | Progress: (4/20) | 4.88 s
[Task 17/25] Current/Best: 14.16/ 23.00 GFLOPS | Progress: (8/20) | 7.77 s
[Task 17/25] Current/Best: 16.80/ 23.00 GFLOPS | Progress: (12/20) | 9.82 s
[Task 17/25] Current/Best: 16.64/ 23.00 GFLOPS | Progress: (16/20) | 12.07 s
[Task 17/25] Current/Best: 10.02/ 23.00 GFLOPS | Progress: (20/20) | 14.25 s Done.
-
[Task 18/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 18/25] Current/Best: 11.53/ 17.06 GFLOPS | Progress: (4/20) | 3.85 s
[Task 18/25] Current/Best: 10.62/ 18.40 GFLOPS | Progress: (8/20) | 7.58 s
[Task 18/25] Current/Best: 19.57/ 19.57 GFLOPS | Progress: (12/20) | 9.52 s
[Task 18/25] Current/Best: 9.93/ 19.57 GFLOPS | Progress: (16/20) | 13.47 s
[Task 18/25] Current/Best: 20.57/ 20.57 GFLOPS | Progress: (20/20) | 15.01 s Done.
-
[Task 19/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 19/25] Current/Best: 7.05/ 20.29 GFLOPS | Progress: (4/20) | 6.27 s
[Task 19/25] Current/Best: 2.60/ 20.29 GFLOPS | Progress: (8/20) | 9.61 s
[Task 19/25] Current/Best: 19.14/ 20.76 GFLOPS | Progress: (12/20) | 12.61 s
[Task 19/25] Current/Best: 13.92/ 20.76 GFLOPS | Progress: (16/20) | 15.75 s
[Task 19/25] Current/Best: 2.69/ 23.22 GFLOPS | Progress: (20/20) | 18.55 s Done.
-
[Task 20/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 20/25] Current/Best: 8.62/ 14.94 GFLOPS | Progress: (4/20) | 3.40 s Done.
+
[Task 1/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 1/25] Current/Best: 17.43/ 17.43 GFLOPS | Progress: (4/20) | 6.40 s
[Task 1/25] Current/Best: 6.16/ 17.43 GFLOPS | Progress: (8/20) | 9.42 s
[Task 1/25] Current/Best: 11.49/ 22.72 GFLOPS | Progress: (12/20) | 11.90 s
[Task 1/25] Current/Best: 16.77/ 22.72 GFLOPS | Progress: (16/20) | 13.59 s
[Task 1/25] Current/Best: 11.60/ 23.82 GFLOPS | Progress: (20/20) | 15.33 s Done.
+
[Task 2/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 2/25] Current/Best: 12.13/ 12.34 GFLOPS | Progress: (4/20) | 3.84 s
[Task 2/25] Current/Best: 13.81/ 18.30 GFLOPS | Progress: (8/20) | 5.15 s
[Task 2/25] Current/Best: 21.01/ 21.01 GFLOPS | Progress: (12/20) | 6.48 s
[Task 2/25] Current/Best: 12.17/ 21.01 GFLOPS | Progress: (16/20) | 7.78 s
[Task 2/25] Current/Best: 18.15/ 21.01 GFLOPS | Progress: (20/20) | 9.40 s Done.
+
[Task 3/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 3/25] Current/Best: 1.63/ 10.54 GFLOPS | Progress: (4/20) | 5.90 s
[Task 3/25] Current/Best: 15.51/ 16.84 GFLOPS | Progress: (8/20) | 7.82 s
[Task 3/25] Current/Best: 14.82/ 16.84 GFLOPS | Progress: (12/20) | 9.59 s
[Task 3/25] Current/Best: 7.19/ 23.66 GFLOPS | Progress: (16/20) | 11.51 s
[Task 3/25] Current/Best: 12.56/ 23.66 GFLOPS | Progress: (20/20) | 16.11 s Done.
+
[Task 4/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 4/25] Current/Best: 9.54/ 20.50 GFLOPS | Progress: (4/20) | 2.41 s
[Task 4/25] Current/Best: 6.56/ 20.50 GFLOPS | Progress: (8/20) | 7.24 s
[Task 4/25] Current/Best: 20.69/ 20.69 GFLOPS | Progress: (12/20) | 12.22 s
[Task 4/25] Current/Best: 17.34/ 21.14 GFLOPS | Progress: (16/20) | 14.66 s
[Task 4/25] Current/Best: 13.02/ 21.14 GFLOPS | Progress: (20/20) | 16.80 s Done.
+
[Task 5/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 5/25] Current/Best: 9.35/ 10.04 GFLOPS | Progress: (4/20) | 2.63 s
[Task 5/25] Current/Best: 11.69/ 13.16 GFLOPS | Progress: (8/20) | 4.70 s
[Task 5/25] Current/Best: 11.08/ 17.96 GFLOPS | Progress: (12/20) | 7.94 s
[Task 5/25] Current/Best: 11.71/ 22.44 GFLOPS | Progress: (16/20) | 9.35 s
[Task 5/25] Current/Best: 11.99/ 22.44 GFLOPS | Progress: (20/20) | 11.28 s Done.
+
[Task 6/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 6/25] Current/Best: 12.32/ 20.64 GFLOPS | Progress: (4/20) | 4.16 s
[Task 6/25] Current/Best: 18.95/ 20.64 GFLOPS | Progress: (8/20) | 5.92 s
[Task 6/25] Current/Best: 13.32/ 20.64 GFLOPS | Progress: (12/20) | 7.87 s
[Task 6/25] Current/Best: 19.92/ 20.64 GFLOPS | Progress: (16/20) | 10.13 s
[Task 6/25] Current/Best: 3.73/ 20.64 GFLOPS | Progress: (20/20) | 12.68 s Done.
+
[Task 7/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 7/25] Current/Best: 10.95/ 12.73 GFLOPS | Progress: (4/20) | 3.60 s
[Task 7/25] Current/Best: 20.15/ 20.81 GFLOPS | Progress: (8/20) | 5.11 s
[Task 7/25] Current/Best: 16.11/ 20.81 GFLOPS | Progress: (12/20) | 7.04 s
[Task 7/25] Current/Best: 12.26/ 20.86 GFLOPS | Progress: (16/20) | 9.10 s
[Task 7/25] Current/Best: 6.32/ 21.37 GFLOPS | Progress: (20/20) | 11.58 s Done.
+
[Task 8/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 8/25] Current/Best: 9.68/ 14.09 GFLOPS | Progress: (4/20) | 2.96 s
[Task 8/25] Current/Best: 9.94/ 14.09 GFLOPS | Progress: (8/20) | 8.13 s
[Task 8/25] Current/Best: 12.80/ 14.09 GFLOPS | Progress: (12/20) | 14.77 s
[Task 8/25] Current/Best: 18.81/ 18.81 GFLOPS | Progress: (16/20) | 16.87 s
[Task 8/25] Current/Best: 20.13/ 20.13 GFLOPS | Progress: (20/20) | 24.13 s Done.
+
[Task 9/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 9/25] Current/Best: 14.25/ 15.44 GFLOPS | Progress: (4/20) | 11.96 s
[Task 9/25] Current/Best: 23.32/ 23.32 GFLOPS | Progress: (8/20) | 13.76 s
[Task 9/25] Current/Best: 8.24/ 23.32 GFLOPS | Progress: (12/20) | 16.32 s
[Task 9/25] Current/Best: 17.96/ 23.32 GFLOPS | Progress: (16/20) | 19.12 s
[Task 9/25] Current/Best: 8.96/ 23.32 GFLOPS | Progress: (20/20) | 27.98 s
[Task 10/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 10/25] Current/Best: 18.01/ 18.01 GFLOPS | Progress: (4/20) | 2.62 s
[Task 10/25] Current/Best: 15.41/ 18.01 GFLOPS | Progress: (8/20) | 4.28 s
[Task 10/25] Current/Best: 12.55/ 18.83 GFLOPS | Progress: (12/20) | 5.83 s
[Task 10/25] Current/Best: 19.09/ 20.22 GFLOPS | Progress: (16/20) | 6.94 s
[Task 10/25] Current/Best: 8.83/ 20.22 GFLOPS | Progress: (20/20
) | 8.47 s Done.
+
[Task 11/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 11/25] Current/Best: 12.30/ 17.93 GFLOPS | Progress: (4/20) | 3.35 s
[Task 11/25] Current/Best: 16.96/ 17.93 GFLOPS | Progress: (8/20) | 6.19 s
[Task 11/25] Current/Best: 18.09/ 18.09 GFLOPS | Progress: (12/20) | 8.29 s
[Task 11/25] Current/Best: 13.07/ 21.09 GFLOPS | Progress: (16/20) | 11.29 s
[Task 11/25] Current/Best: 19.38/ 21.25 GFLOPS | Progress: (20/20) | 13.41 s Done.
+
[Task 12/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 12/25] Current/Best: 7.80/ 17.97 GFLOPS | Progress: (4/20) | 5.86 s
[Task 12/25] Current/Best: 5.13/ 17.97 GFLOPS | Progress: (8/20) | 9.82 s
[Task 12/25] Current/Best: 19.11/ 19.11 GFLOPS | Progress: (12/20) | 11.81 s
[Task 12/25] Current/Best: 15.19/ 19.11 GFLOPS | Progress: (16/20) | 14.80 s
[Task 12/25] Current/Best: 15.09/ 19.11 GFLOPS | Progress: (20/20) | 16.74 s Done.
+
[Task 13/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 13/25] Current/Best: 8.60/ 16.49 GFLOPS | Progress: (4/20) | 3.83 s
[Task 13/25] Current/Best: 15.61/ 20.65 GFLOPS | Progress: (8/20) | 6.48 s
[Task 13/25] Current/Best: 19.52/ 21.63 GFLOPS | Progress: (12/20) | 9.54 s
[Task 13/25] Current/Best: 12.19/ 21.63 GFLOPS | Progress: (16/20) | 13.03 s
[Task 13/25] Current/Best: 18.58/ 21.63 GFLOPS | Progress: (20/20) | 15.40 s Done.
+
[Task 14/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 14/25] Current/Best: 13.68/ 13.68 GFLOPS | Progress: (4/20) | 3.46 s
[Task 14/25] Current/Best: 6.12/ 13.68 GFLOPS | Progress: (8/20) | 5.65 s
[Task 14/25] Current/Best: 20.36/ 20.36 GFLOPS | Progress: (12/20) | 8.39 s
[Task 14/25] Current/Best: 16.77/ 20.36 GFLOPS | Progress: (16/20) | 10.11 s Done.
+
[Task 14/25] Current/Best: 16.97/ 20.36 GFLOPS | Progress: (20/20) | 11.91 s
[Task 15/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 15/25] Current/Best: 16.11/ 17.58 GFLOPS | Progress: (4/20) | 2.79 s
[Task 15/25] Current/Best: 14.43/ 17.97 GFLOPS | Progress: (8/20) | 4.10 s
[Task 15/25] Current/Best: 10.37/ 22.24 GFLOPS | Progress: (12/20) | 6.38 s
[Task 15/25] Current/Best: 20.24/ 22.24 GFLOPS | Progress: (16/20) | 9.46 s
[Task 15/25] Current/Best: 9.69/ 22.24 GFLOPS | Progress: (20/20) | 10.49 s
[Task 16/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 16/25] Current/Best: 20.50/ 20.50 GFLOPS | Progress: (4/20) | 3.06 s
[Task 16/25] Current/Best: 3.04/ 20.50 GFLOPS | Progress: (8/20) | 4.69 s
[Task 16/25] Current/Best: 19.80/ 20.50 GFLOPS | Progress: (12/20) | 5.92 s
[Task 16/25] Current/Best: 17.30/ 20.50 GFLOPS | Progress: (16/20) |
7.29 s
[Task 16/25] Current/Best: 10.03/ 21.98 GFLOPS | Progress: (20/20) | 9.49 s Done.
+
[Task 17/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 17/25] Current/Best: 13.41/ 18.80 GFLOPS | Progress: (4/20) | 4.88 s
[Task 17/25] Current/Best: 14.35/ 23.06 GFLOPS | Progress: (8/20) | 7.72 s
[Task 17/25] Current/Best: 17.22/ 23.06 GFLOPS | Progress: (12/20) | 9.77 s
[Task 17/25] Current/Best: 16.39/ 23.06 GFLOPS | Progress: (16/20) | 12.03 s
[Task 17/25] Current/Best: 10.00/ 23.06 GFLOPS | Progress: (20/20) | 14.22 s Done.
+
[Task 18/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 18/25] Current/Best: 11.03/ 17.34 GFLOPS | Progress: (4/20) | 3.88 s
[Task 18/25] Current/Best: 10.57/ 19.55 GFLOPS | Progress: (8/20) | 7.66 s
[Task 18/25] Current/Best: 19.29/ 19.55 GFLOPS | Progress: (12/20) | 9.62 s
[Task 18/25] Current/Best: 9.96/ 19.55 GFLOPS | Progress: (16/20) | 13.60 s
[Task 18/25] Current/Best: 20.66/ 20.66 GFLOPS | Progress: (20/20) | 15.14 s Done.
+
[Task 19/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 19/25] Current/Best: 6.89/ 20.09 GFLOPS | Progress: (4/20) | 6.36 s
[Task 19/25] Current/Best: 2.60/ 20.09 GFLOPS | Progress: (8/20) | 9.70 s
[Task 19/25] Current/Best: 19.03/ 20.91 GFLOPS | Progress: (12/20) | 12.70 s
[Task 19/25] Current/Best: 15.11/ 20.91 GFLOPS | Progress: (16/20) | 15.71 s
[Task 19/25] Current/Best: 2.69/ 23.24 GFLOPS | Progress: (20/20) | 18.51 s Done.
+
[Task 20/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 20/25] Current/Best: 9.25/ 14.83 GFLOPS | Progress: (4/20) | 3.42 s Done.
Done.
-
[Task 20/25] Current/Best: 10.21/ 14.94 GFLOPS | Progress: (8/20) | 6.96 s
[Task 20/25] Current/Best: 2.32/ 16.65 GFLOPS | Progress: (12/20) | 11.07 s
[Task 20/25] Current/Best: 12.38/ 16.65 GFLOPS | Progress: (16/20) | 14.85 s
[Task 20/25] Current/Best: 12.79/ 21.81 GFLOPS | Progress: (20/20) | 16.98 s
[Task 21/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 21/25] Current/Best: 6.36/ 17.71 GFLOPS | Progress: (4/20) | 3.34 s
[Task 21/25] Current/Best: 14.31/ 17.71 GFLOPS | Progress: (8/20) | 4.95 s
[Task 21/25] Current/Best: 1.61/ 17.71 GFLOPS | Progress: (12/20) | 7.11 s
[Task 21/25] Current/Best: 17.89/ 17.89 GFLOPS | Progress: (16/20) | 10.68 s
[Task 21/25] Current/Best: 4.46/ 17.89 GFLOPS | Progress: (20/20) | 18.14 s
[Task 22/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 22/25] Current/Best: 2.70/ 16.97 GFLOPS | Progress: (4/20
) | 2.73 s
[Task 22/25] Current/Best: 9.19/ 21.68 GFLOPS | Progress: (8/20) | 4.70 s
[Task 22/25] Current/Best: 19.64/ 21.68 GFLOPS | Progress: (12/20) | 7.10 s
[Task 22/25] Current/Best: 15.26/ 21.68 GFLOPS | Progress: (16/20) | 9.23 s
[Task 22/25] Current/Best: 14.10/ 21.68 GFLOPS | Progress: (20/20) | 10.98 s Done.
-
[Task 23/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 23/25] Current/Best: 17.34/ 20.12 GFLOPS | Progress: (4/20) | 3.30 s
[Task 23/25] Current/Best: 15.70/ 20.12 GFLOPS | Progress: (8/20) | 6.60 s
[Task 23/25] Current/Best: 20.74/ 21.47 GFLOPS | Progress: (12/20) | 8.46 s
[Task 23/25] Current/Best: 6.28/ 21.47 GFLOPS | Progress: (16/20) | 15.56 s
[Task 23/25] Current/Best: 7.69/ 21.47 GFLOPS | Progress: (20/20) | 19.84 s Done.
-
[Task 24/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 24/25] Current/Best: 8.76/ 8.76 GFLOPS | Progress: (4/20) | 11.83 s
[Task 24/25] Current/Best: 3.37/ 8.76 GFLOPS | Progress: (8/20) | 23.10 s
[Task 24/25] Current/Best: 4.59/ 8.76 GFLOPS | Progress: (12/20) | 33.84 s Done.
+
[Task 20/25] Current/Best: 10.23/ 14.83 GFLOPS | Progress: (8/20) | 7.01 s
[Task 20/25] Current/Best: 2.32/ 15.15 GFLOPS | Progress: (12/20) | 11.03 s
[Task 20/25] Current/Best: 12.41/ 15.15 GFLOPS | Progress: (16/20) | 15.06 s
[Task 20/25] Current/Best: 13.24/ 21.47 GFLOPS | Progress: (20/20) | 17.18 s
[Task 21/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 21/25] Current/Best: 6.37/ 17.56 GFLOPS | Progress: (4/20) | 3.37 s
[Task 21/25] Current/Best: 14.42/ 17.56 GFLOPS | Progress: (8/20) | 5.01 s
[Task 21/25] Current/Best: 1.61/ 17.56 GFLOPS | Progress: (12/20) | 7.19 s
[Task 21/25] Current/Best: 18.01/ 18.01 GFLOPS | Progress: (16/20) | 10.79 s
[Task 21/25] Current/Best: 4.47/ 18.01 GFLOPS | Progress: (20/20) | 18.41 s
[Task 22/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 22/25] Current/Best: 2.70/ 17.05 GFLOPS | Progress: (4/20
) | 2.75 s
[Task 22/25] Current/Best: 9.15/ 21.41 GFLOPS | Progress: (8/20) | 4.74 s
[Task 22/25] Current/Best: 19.58/ 21.41 GFLOPS | Progress: (12/20) | 7.18 s
[Task 22/25] Current/Best: 15.05/ 21.41 GFLOPS | Progress: (16/20) | 9.32 s
[Task 22/25] Current/Best: 14.84/ 21.41 GFLOPS | Progress: (20/20) | 11.09 s Done.
+
[Task 23/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 23/25] Current/Best: 17.30/ 20.03 GFLOPS | Progress: (4/20) | 3.32 s
[Task 23/25] Current/Best: 15.71/ 20.03 GFLOPS | Progress: (8/20) | 6.74 s
[Task 23/25] Current/Best: 20.47/ 21.12 GFLOPS | Progress: (12/20) | 8.62 s
[Task 23/25] Current/Best: 6.05/ 21.12 GFLOPS | Progress: (16/20) | 15.99 s
[Task 23/25] Current/Best: 7.54/ 21.12 GFLOPS | Progress: (20/20) | 20.28 s Done.
+
[Task 24/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 24/25] Current/Best: 8.43/ 8.43 GFLOPS | Progress: (4/20) | 11.87 s
[Task 24/25] Current/Best: 1.98/ 8.43 GFLOPS | Progress: (8/20) | 22.97 s
[Task 24/25] Current/Best: 4.12/ 8.43 GFLOPS | Progress: (12/20) | 34.53 s Done.
Done.
-
[Task 24/25] Current/Best: 6.83/ 8.76 GFLOPS | Progress: (16/20) | 39.77 s
[Task 24/25] Current/Best: 3.33/ 8.93 GFLOPS | Progress: (20/20) | 45.84 s Done.
-
[Task 25/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 25/25] Current/Best: 1.54/ 2.91 GFLOPS | Progress: (4/20) | 11.65 s
[Task 25/25] Current/Best: 5.66/ 7.70 GFLOPS | Progress: (8/20) | 22.93 s
[Task 25/25] Current/Best: 5.86/ 7.70 GFLOPS | Progress: (12/20) | 34.24 s
[Task 25/25] Current/Best: 5.81/ 9.15 GFLOPS | Progress: (16/20) | 36.08 s
[Task 25/25] Current/Best: 2.94/ 9.15 GFLOPS | Progress: (20/20) | 46.76 s
+
[Task 24/25] Current/Best: 6.99/ 8.52 GFLOPS | Progress: (16/20) | 40.42 s
[Task 24/25] Current/Best: 3.32/ 8.68 GFLOPS | Progress: (20/20) | 46.54 s Done.
+
[Task 25/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 25/25] Current/Best: 1.54/ 2.89 GFLOPS | Progress: (4/20) | 11.64 s
[Task 25/25] Current/Best: 5.50/ 7.42 GFLOPS | Progress: (8/20) | 22.98 s
[Task 25/25] Current/Best: 5.76/ 7.42 GFLOPS | Progress: (12/20) | 34.48 s
[Task 25/25] Current/Best: 5.86/ 8.61 GFLOPS | Progress: (16/20) | 36.22 s
[Task 25/25] Current/Best: 2.93/ 8.61 GFLOPS | Progress: (20/20) | 46.89 s
@@ -748,8 +748,8 @@ improvement in comparing the optimized model to the unoptimized model.
.. code-block:: none
- optimized: {'mean': 410.96856623999884, 'median': 411.07895865000046, 'std': 0.6519231574679785}
- unoptimized: {'mean': 497.285720330002, 'median': 497.179591150001, 'std': 0.4704793355308102}
+ optimized: {'mean': 415.12749731998156, 'median': 415.12617939997654, 'std': 1.2218218604940612}
+ unoptimized: {'mean': 500.172581780007, 'median': 500.1141345000178, 'std': 0.43971723624319015}
@@ -772,7 +772,7 @@ profiling/benchmarking.
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 10 minutes 29.418 seconds)
+ **Total running time of the script:** ( 10 minutes 32.989 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 01066b352..172e1f209 100644
--- a/docs/_sources/tutorial/cross_compilation_and_rpc.rst.txt
+++ b/docs/_sources/tutorial/cross_compilation_and_rpc.rst.txt
@@ -282,7 +282,7 @@ device and returns the measured cost. Network overhead is excluded.
.. code-block:: none
- 1.275e-07 secs/op
+ 1.221e-07 secs/op
diff --git a/docs/_sources/tutorial/intro_topi.rst.txt b/docs/_sources/tutorial/intro_topi.rst.txt
index 07dce4ddf..76d322f4b 100644
--- a/docs/_sources/tutorial/intro_topi.rst.txt
+++ b/docs/_sources/tutorial/intro_topi.rst.txt
@@ -263,7 +263,7 @@ As you can see, scheduled stages of computation have been accumulated and we can
.. code-block:: none
- [stage(a, placeholder(a, 0xe034b50)), stage(b, placeholder(b, 0x61084d0)), stage(T_add, compute(T_add, body=[(a[ax0, ax1, ax2] + b[ax1, ax2])], axis=[iter_var(ax0, range(min=0, ext=100)), iter_var(ax1, range(min=0, ext=10)), iter_var(ax2, range(min=0, ext=10))], reduce_axis=[], tag=broadcast, attrs={})), stage(T_multiply, compute(T_multiply, body=[(a[ax0, ax1, ax2]*b[ax1, ax2])], axis=[iter_var(ax0, range(min=0, ext=100)), iter_var(ax1, range(min=0, ext=10)), iter_var(ax2, range(min= [...]
+ [stage(a, placeholder(a, 0xcce3610)), stage(b, placeholder(b, 0x4819510)), stage(T_add, compute(T_add, body=[(a[ax0, ax1, ax2] + b[ax1, ax2])], axis=[iter_var(ax0, range(min=0, ext=100)), iter_var(ax1, range(min=0, ext=10)), iter_var(ax2, range(min=0, ext=10))], reduce_axis=[], tag=broadcast, attrs={})), stage(T_multiply, compute(T_multiply, body=[(a[ax0, ax1, ax2]*b[ax1, ax2])], axis=[iter_var(ax0, range(min=0, ext=100)), iter_var(ax1, range(min=0, ext=10)), iter_var(ax2, range(min= [...]
diff --git a/docs/_sources/tutorial/sg_execution_times.rst.txt b/docs/_sources/tutorial/sg_execution_times.rst.txt
index 471ce32c6..3f445ca12 100644
--- a/docs/_sources/tutorial/sg_execution_times.rst.txt
+++ b/docs/_sources/tutorial/sg_execution_times.rst.txt
@@ -5,30 +5,30 @@
Computation times
=================
-**13:13.033** total execution time for **tutorial** files:
+**13:21.540** total execution time for **tutorial** files:
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_autotvm_relay_x86.py` (``autotvm_relay_x86.py``) | 10:29.418 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_autotvm_relay_x86.py` (``autotvm_relay_x86.py``) | 10:32.989 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_tensor_expr_get_started.py` (``tensor_expr_get_started.py``) | 01:01.053 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_tensor_expr_get_started.py` (``tensor_expr_get_started.py``) | 01:01.394 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_auto_scheduler_matmul_x86.py` (``auto_scheduler_matmul_x86.py``) | 00:48.185 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_auto_scheduler_matmul_x86.py` (``auto_scheduler_matmul_x86.py``) | 00:52.796 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_relay_quick_start.py` (``relay_quick_start.py``) | 00:28.345 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_relay_quick_start.py` (``relay_quick_start.py``) | 00:28.792 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_autotvm_matmul_x86.py` (``autotvm_matmul_x86.py``) | 00:23.945 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_autotvm_matmul_x86.py` (``autotvm_matmul_x86.py``) | 00:24.143 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_tensor_ir_blitz_course.py` (``tensor_ir_blitz_course.py``) | 00:01.231 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_intro_topi.py` (``intro_topi.py``) | 00:00.706 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_intro_topi.py` (``intro_topi.py``) | 00:00.699 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_tensor_ir_blitz_course.py` (``tensor_ir_blitz_course.py``) | 00:00.549 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_cross_compilation_and_rpc.py` (``cross_compilation_and_rpc.py``) | 00:00.149 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_cross_compilation_and_rpc.py` (``cross_compilation_and_rpc.py``) | 00:00.164 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
| :ref:`sphx_glr_tutorial_introduction.py` (``introduction.py``) | 00:00.005 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_tvmc_python.py` (``tvmc_python.py``) | 00:00.001 | 0.0 MB |
-+------------------------------------------------------------------------------------------+-----------+--------+
| :ref:`sphx_glr_tutorial_install.py` (``install.py``) | 00:00.001 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
+| :ref:`sphx_glr_tutorial_tvmc_python.py` (``tvmc_python.py``) | 00:00.001 | 0.0 MB |
++------------------------------------------------------------------------------------------+-----------+--------+
| :ref:`sphx_glr_tutorial_tvmc_command_line_driver.py` (``tvmc_command_line_driver.py``) | 00:00.001 | 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 233fae2c5..c1b0c1bc3 100644
--- a/docs/_sources/tutorial/tensor_expr_get_started.rst.txt
+++ b/docs/_sources/tutorial/tensor_expr_get_started.rst.txt
@@ -301,8 +301,8 @@ helper function to run a profile of the TVM generated code.
.. code-block:: none
- Numpy running time: 0.000008
- naive: 0.000007
+ Numpy running time: 0.000007
+ naive: 0.000008
@@ -403,7 +403,7 @@ compile and run this new schedule with the parallel operation applied:
/workspace/python/tvm/driver/build_module.py:268: UserWarning: target_host parameter is going to be deprecated. Please pass in tvm.target.Target(target, host=target_host) instead.
"target_host parameter is going to be deprecated. "
- parallel: 0.000006
+ parallel: 0.000007
@@ -512,10 +512,10 @@ We can now compare the different schedules
.. code-block:: none
Operator Timing Performance
- numpy 7.814130001406738e-06 1.0
- naive 6.715599999999999e-06 0.859417490979933
- parallel 6.0443e-06 0.773509014939843
- vector 2.45751e-05 3.1449566356812433
+ numpy 6.864689994472428e-06 1.0
+ naive 7.603899999999999e-06 1.1076829406896447
+ parallel 6.990699999999999e-06 1.0183562558001944
+ vector 2.45305e-05 3.5734315780832637
@@ -936,7 +936,7 @@ matrix multiplication.
.. code-block:: none
- Numpy running time: 0.018873
+ Numpy running time: 0.018571
@@ -996,7 +996,7 @@ optimizations.
/workspace/python/tvm/driver/build_module.py:268: UserWarning: target_host parameter is going to be deprecated. Please pass in tvm.target.Target(target, host=target_host) instead.
"target_host parameter is going to be deprecated. "
- none: 3.429490
+ none: 3.434227
@@ -1101,7 +1101,7 @@ schedule.
/workspace/python/tvm/driver/build_module.py:268: UserWarning: target_host parameter is going to be deprecated. Please pass in tvm.target.Target(target, host=target_host) instead.
"target_host parameter is going to be deprecated. "
- blocking: 0.303072
+ blocking: 0.314475
@@ -1199,7 +1199,7 @@ already cache friendly from our previous optimizations.
/workspace/python/tvm/driver/build_module.py:268: UserWarning: target_host parameter is going to be deprecated. Please pass in tvm.target.Target(target, host=target_host) instead.
"target_host parameter is going to be deprecated. "
- vectorization: 0.342671
+ vectorization: 0.344531
@main = primfn(A_1: handle, B_1: handle, C_1: handle) -> ()
attr = {"from_legacy_te_schedule": True, "global_symbol": "main", "tir.noalias": True}
buffers = {A: Buffer(A_2: Pointer(float32), float32, [1048576], []),
@@ -1275,7 +1275,7 @@ more cache friendly.
/workspace/python/tvm/driver/build_module.py:268: UserWarning: target_host parameter is going to be deprecated. Please pass in tvm.target.Target(target, host=target_host) instead.
"target_host parameter is going to be deprecated. "
- loop permutation: 0.120775
+ loop permutation: 0.117761
@main = primfn(A_1: handle, B_1: handle, C_1: handle) -> ()
attr = {"from_legacy_te_schedule": True, "global_symbol": "main", "tir.noalias": True}
buffers = {A: Buffer(A_2: Pointer(float32), float32, [1048576], []),
@@ -1376,7 +1376,7 @@ optimized schedule.
/workspace/python/tvm/driver/build_module.py:268: UserWarning: target_host parameter is going to be deprecated. Please pass in tvm.target.Target(target, host=target_host) instead.
"target_host parameter is going to be deprecated. "
- array packing: 0.109393
+ array packing: 0.108552
@main = primfn(A_1: handle, B_1: handle, C_1: handle) -> ()
attr = {"from_legacy_te_schedule": True, "global_symbol": "main", "tir.noalias": True}
buffers = {A: Buffer(A_2: Pointer(float32), float32, [1048576], []),
@@ -1471,7 +1471,7 @@ to `C` when all the block results are ready.
/workspace/python/tvm/driver/build_module.py:268: UserWarning: target_host parameter is going to be deprecated. Please pass in tvm.target.Target(target, host=target_host) instead.
"target_host parameter is going to be deprecated. "
- block caching: 0.111355
+ block caching: 0.109521
@main = primfn(A_1: handle, B_1: handle, C_1: handle) -> ()
attr = {"from_legacy_te_schedule": True, "global_symbol": "main", "tir.noalias": True}
buffers = {A: Buffer(A_2: Pointer(float32), float32, [1048576], []),
@@ -1559,7 +1559,7 @@ of thread-level parallelization.
/workspace/python/tvm/driver/build_module.py:268: UserWarning: target_host parameter is going to be deprecated. Please pass in tvm.target.Target(target, host=target_host) instead.
"target_host parameter is going to be deprecated. "
- parallelization: 0.144799
+ parallelization: 0.144144
@main = primfn(A_1: handle, B_1: handle, C_1: handle) -> ()
attr = {"from_legacy_te_schedule": True, "global_symbol": "main", "tir.noalias": True}
buffers = {A: Buffer(A_2: Pointer(float32), float32, [1048576], []),
@@ -1640,13 +1640,13 @@ working, we can compare the results.
.. code-block:: none
Operator Timing Performance
- none 3.4294904799999997 1.0
- blocking 0.3030718857 0.0883722778842646
- vectorization 0.3426707194 0.09991884257978755
- loop permutation 0.12077503819999999 0.03521661276050546
- array packing 0.1093933389 0.031897840083813266
- block caching 0.1113546419 0.032469733492305836
- parallelization 0.1447990736 0.042221745313023876
+ none 3.4342266564999995 1.0
+ blocking 0.3144748072 0.09157077812694453
+ vectorization 0.3445306166 0.10032262021724833
+ loop permutation 0.1177608365 0.03429035071902221
+ array packing 0.10855221 0.031608924179346756
+ block caching 0.1095208572 0.03189098104305627
+ parallelization 0.1441438333 0.0419727198340789
@@ -1688,7 +1688,7 @@ the computation for specific platforms.
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 1.053 seconds)
+ **Total running time of the script:** ( 1 minutes 1.394 seconds)
.. _sphx_glr_download_tutorial_tensor_expr_get_started.py:
diff --git a/docs/commit_hash b/docs/commit_hash
index 54b31b459..8369c65bb 100644
--- a/docs/commit_hash
+++ b/docs/commit_hash
@@ -1 +1 @@
-cf15375e20cd1fb2e201e47afda169732ef88eea
+fbb7b5d1a0d82acb1f581dd2ec362b4dcad2638e
diff --git a/docs/how_to/compile_models/from_darknet.html b/docs/how_to/compile_models/from_darknet.html
index 307da67fa..bde28b61b 100644
--- a/docs/how_to/compile_models/from_darknet.html
+++ b/docs/how_to/compile_models/from_darknet.html
@@ -569,7 +569,7 @@ 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 1.137 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 2.100 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 d9e0ceaf2..cd4dc88ce 100644
--- a/docs/how_to/compile_models/from_mxnet.html
+++ b/docs/how_to/compile_models/from_mxnet.html
@@ -422,7 +422,7 @@ to download the full example code</p>
<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.zip60b9a966-12eb-4861-83d9-5634ac1241ee 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.zip1cf8aeaa-377a-4b9f-8caf-ece2bf0475c4 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 f79f25099..c51ad3158 100644
--- a/docs/how_to/compile_models/from_oneflow.html
+++ b/docs/how_to/compile_models/from_oneflow.html
@@ -427,13 +427,13 @@ python3 -m pip install -f https://release.oneflow.info <span class="nv">oneflow<
<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, 60.1MB/s]
- 35%|###4 | 14.3M/41.5M [00:00<00:00, 55.8MB/s]
- 47%|####7 | 19.6M/41.5M [00:00<00:00, 55.0MB/s]
- 60%|#####9 | 24.9M/41.5M [00:00<00:00, 35.0MB/s]
- 77%|#######7 | 32.1M/41.5M [00:00<00:00, 44.5MB/s]
- 90%|########9 | 37.2M/41.5M [00:00<00:00, 44.3MB/s]
-100%|##########| 41.5M/41.5M [00:00<00:00, 47.3MB/s]
+ 15%|#5 | 6.33M/41.5M [00:00<00:01, 32.3MB/s]
+ 24%|##4 | 10.1M/41.5M [00:00<00:00, 35.2MB/s]
+ 39%|###8 | 16.0M/41.5M [00:00<00:00, 42.5MB/s]
+ 58%|#####7 | 24.0M/41.5M [00:00<00:00, 41.9MB/s]
+ 77%|#######7 | 32.0M/41.5M [00:00<00:00, 49.0MB/s]
+ 96%|#########6| 40.0M/41.5M [00:00<00:00, 52.6MB/s]
+100%|##########| 41.5M/41.5M [00:00<00:00, 48.1MB/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 c94f35258..babbb280d 100644
--- a/docs/how_to/compile_models/from_pytorch.html
+++ b/docs/how_to/compile_models/from_pytorch.html
@@ -409,14 +409,8 @@ be unstable.</p>
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>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]
- 2%|1 | 760k/44.7M [00:00<00:05, 7.74MB/s]
- 4%|4 | 1.98M/44.7M [00:00<00:04, 10.8MB/s]
- 9%|9 | 4.03M/44.7M [00:00<00:02, 15.7MB/s]
- 17%|#6 | 7.42M/44.7M [00:00<00:01, 23.5MB/s]
- 29%|##9 | 13.1M/44.7M [00:00<00:00, 36.3MB/s]
- 51%|#####1 | 22.8M/44.7M [00:00<00:00, 58.6MB/s]
- 86%|########5 | 38.3M/44.7M [00:00<00:00, 92.5MB/s]
-100%|##########| 44.7M/44.7M [00:00<00:00, 63.2MB/s]
+ 42%|####2 | 19.0M/44.7M [00:00<00:00, 199MB/s]
+100%|##########| 44.7M/44.7M [00:00<00:00, 235MB/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 457683e90..a76251226 100644
--- a/docs/how_to/compile_models/from_tensorflow.html
+++ b/docs/how_to/compile_models/from_tensorflow.html
@@ -631,7 +631,7 @@ 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 1.537 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 2.920 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 4c17b1ea4..f6e8caf82 100644
--- a/docs/how_to/compile_models/sg_execution_times.html
+++ b/docs/how_to/compile_models/sg_execution_times.html
@@ -322,7 +322,7 @@
<div class="section" id="computation-times">
<span id="sphx-glr-how-to-compile-models-sg-execution-times"></span><h1>Computation times<a class="headerlink" href="#computation-times" title="Permalink to this headline">¶</a></h1>
-<p><strong>05:13.044</strong> total execution time for <strong>how_to_compile_models</strong> files:</p>
+<p><strong>05:17.547</strong> total execution time for <strong>how_to_compile_models</strong> files:</p>
<table class="docutils align-default">
<colgroup>
<col style="width: 81%" />
@@ -331,43 +331,43 @@
</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>01:01.537</p></td>
+<td><p>01:02.920</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>01:01.137</p></td>
+<td><p>01:02.100</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>00:39.852</p></td>
+<td><p>00:40.181</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><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>00:31.538</p></td>
+<td><p>00:32.825</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><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>00:26.001</p></td>
+<td><p>00:26.237</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><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>00:25.569</p></td>
+<td><p>00:25.090</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><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>00:23.003</p></td>
+<td><p>00:23.089</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><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>00:22.003</p></td>
+<td><p>00:22.144</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><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>00:20.045</p></td>
+<td><p>00:20.073</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>00:02.358</p></td>
+<td><p>00:02.890</p></td>
<td><p>0.0 MB</p></td>
</tr>
</tbody>
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 c565f66be..db7c03608 100644
--- a/docs/how_to/deploy_models/deploy_model_on_android.html
+++ b/docs/how_to/deploy_models/deploy_model_on_android.html
@@ -648,7 +648,7 @@ to the remote android device.</p>
Evaluate inference time cost...
Execution time summary:
mean (ms) median (ms) max (ms) min (ms) std (ms)
- 16.0296 16.1118 16.3556 15.5997 0.2841
+ 16.2902 16.2610 16.6797 16.0264 0.2077
</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 e2e434a9b..cc55fe649 100644
--- a/docs/how_to/deploy_models/deploy_object_detection_pytorch.html
+++ b/docs/how_to/deploy_models/deploy_object_detection_pytorch.html
@@ -431,16 +431,15 @@ be unstable.</p>
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>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]
- 6%|6 | 10.5M/170M [00:00<00:01, 109MB/s]
- 18%|#7 | 29.8M/170M [00:00<00:00, 164MB/s]
- 28%|##8 | 48.3M/170M [00:00<00:00, 173MB/s]
- 40%|###9 | 67.8M/170M [00:00<00:00, 186MB/s]
- 51%|#####1 | 87.4M/170M [00:00<00:00, 193MB/s]
- 63%|######3 | 107M/170M [00:00<00:00, 197MB/s]
- 75%|#######4 | 127M/170M [00:00<00:00, 200MB/s]
- 86%|########6 | 146M/170M [00:00<00:00, 201MB/s]
- 97%|#########7| 165M/170M [00:00<00:00, 200MB/s]
-100%|##########| 170M/170M [00:00<00:00, 191MB/s]
+ 3%|2 | 4.45M/170M [00:00<00:03, 46.6MB/s]
+ 5%|5 | 9.01M/170M [00:00<00:03, 47.3MB/s]
+ 21%|## | 35.2M/170M [00:00<00:00, 151MB/s]
+ 37%|###6 | 62.2M/170M [00:00<00:00, 203MB/s]
+ 51%|#####1 | 87.5M/170M [00:00<00:00, 224MB/s]
+ 67%|######7 | 114M/170M [00:00<00:00, 244MB/s]
+ 83%|########2 | 141M/170M [00:00<00:00, 255MB/s]
+ 98%|#########7| 166M/170M [00:00<00:00, 259MB/s]
+100%|##########| 170M/170M [00:00<00:00, 218MB/s]
/usr/local/lib/python3.7/dist-packages/torch/nn/functional.py:3878: 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)
/usr/local/lib/python3.7/dist-packages/torchvision/models/detection/anchor_utils.py:127: 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').
@@ -535,7 +534,7 @@ 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> ( 2 minutes 55.692 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 2 minutes 57.631 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 131292a3c..f9c5df4d7 100644
--- a/docs/how_to/deploy_models/deploy_prequantized.html
+++ b/docs/how_to/deploy_models/deploy_prequantized.html
@@ -475,10 +475,7 @@ training. Other models require a full post training calibration.</p>
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>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]
- 27%|##7 | 3.69M/13.6M [00:00<00:00, 38.5MB/s]
- 60%|###### | 8.19M/13.6M [00:00<00:00, 43.4MB/s]
- 91%|######### | 12.3M/13.6M [00:00<00:00, 34.7MB/s]
-100%|##########| 13.6M/13.6M [00:00<00:00, 34.2MB/s]
+100%|##########| 13.6M/13.6M [00:00<00:00, 147MB/s]
</pre></div>
</div>
</div>
@@ -567,7 +564,7 @@ output values are identical out of 1000 outputs from mobilenet v2.</p>
</div>
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Execution time summary:
mean (ms) median (ms) max (ms) min (ms) std (ms)
- 90.5425 90.3991 97.1210 90.0839 0.8348
+ 90.4966 90.4039 94.7650 90.2492 0.5064
</pre></div>
</div>
<div class="admonition note">
@@ -606,7 +603,7 @@ This includes support for the VNNI 8 bit dot product instruction (CascadeLake or
<div class="section" id="deploy-a-quantized-tflite-model">
<h2>Deploy a quantized TFLite Model<a class="headerlink" href="#deploy-a-quantized-tflite-model" title="Permalink to this headline">¶</a></h2>
<p>TODO</p>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 8.038 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 7.701 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 fc343b11e..0624e3b8b 100644
--- a/docs/how_to/deploy_models/deploy_prequantized_tflite.html
+++ b/docs/how_to/deploy_models/deploy_prequantized_tflite.html
@@ -568,7 +568,7 @@ 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)
- 120.0127 119.9937 122.4806 119.2379 0.4334
+ 119.7802 119.7182 121.4246 119.0629 0.4287
</pre></div>
</div>
<div class="admonition note">
@@ -596,7 +596,7 @@ 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 1.639 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 59.854 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 ccdb51a05..8713982d6 100644
--- a/docs/how_to/deploy_models/deploy_quantized.html
+++ b/docs/how_to/deploy_models/deploy_quantized.html
@@ -504,7 +504,7 @@ 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 25.372 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 19.144 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 ec14739eb..5fc53dda2 100644
--- a/docs/how_to/deploy_models/deploy_ssd_gluoncv.html
+++ b/docs/how_to/deploy_models/deploy_ssd_gluoncv.html
@@ -436,24 +436,24 @@ to your device.</p>
Downloading /workspace/.mxnet/models/ssd_512_resnet50_v1_voc-9c8b225a.zip from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/ssd_512_resnet50_v1_voc-9c8b225a.zip...
0%| | 0/132723 [00:00<?, ?KB/s]
- 1%|1 | 1952/132723 [00:00<00:06, 19399.67KB/s]
- 4%|4 | 5802/132723 [00:00<00:04, 30604.28KB/s]
- 10%|9 | 12820/132723 [00:00<00:02, 48644.83KB/s]
- 16%|#5 | 21041/132723 [00:00<00:01, 61882.38KB/s]
- 22%|##2 | 29257/132723 [00:00<00:01, 69187.03KB/s]
- 28%|##7 | 36556/132723 [00:00<00:01, 70478.37KB/s]
- 33%|###2 | 43767/132723 [00:00<00:01, 71008.77KB/s]
- 39%|###9 | 51860/132723 [00:00<00:01, 74163.86KB/s]
- 45%|####5 | 60034/132723 [00:00<00:00, 76530.44KB/s]
- 51%|#####1 | 68161/132723 [00:01<00:00, 77991.85KB/s]
- 58%|#####7 | 76409/132723 [00:01<00:00, 79363.71KB/s]
- 64%|######3 | 84697/132723 [00:01<00:00, 80428.89KB/s]
- 70%|####### | 93021/132723 [00:01<00:00, 81276.60KB/s]
- 76%|#######6 | 101315/132723 [00:01<00:00, 81776.29KB/s]
- 83%|########2 | 109562/132723 [00:01<00:00, 81983.22KB/s]
- 89%|########8 | 117845/132723 [00:01<00:00, 82235.45KB/s]
- 95%|#########5| 126231/132723 [00:01<00:00, 82713.67KB/s]
-100%|##########| 132723/132723 [00:01<00:00, 74562.70KB/s]
+ 4%|3 | 4752/132723 [00:00<00:02, 47515.17KB/s]
+ 10%|9 | 12617/132723 [00:00<00:01, 65826.37KB/s]
+ 15%|#5 | 20499/132723 [00:00<00:01, 71755.79KB/s]
+ 21%|##1 | 28397/132723 [00:00<00:01, 74605.24KB/s]
+ 27%|##7 | 36289/132723 [00:00<00:01, 76160.00KB/s]
+ 33%|###3 | 44319/132723 [00:00<00:01, 77565.61KB/s]
+ 39%|###9 | 52308/132723 [00:00<00:01, 78322.63KB/s]
+ 45%|####5 | 60195/132723 [00:00<00:00, 78491.36KB/s]
+ 51%|#####1 | 68045/132723 [00:00<00:00, 78400.42KB/s]
+ 57%|#####7 | 76004/132723 [00:01<00:00, 78766.15KB/s]
+ 63%|######3 | 84004/132723 [00:01<00:00, 79139.23KB/s]
+ 69%|######9 | 91918/132723 [00:01<00:00, 79138.11KB/s]
+ 75%|#######5 | 99832/132723 [00:01<00:00, 79045.28KB/s]
+ 81%|########1 | 107816/132723 [00:01<00:00, 79281.95KB/s]
+ 87%|########7 | 115806/132723 [00:01<00:00, 79465.39KB/s]
+ 93%|#########3| 123832/132723 [00:01<00:00, 79698.80KB/s]
+ 99%|#########9| 131837/132723 [00:01<00:00, 79803.01KB/s]
+100%|##########| 132723/132723 [00:01<00:00, 77466.29KB/s]
</pre></div>
</div>
<p>Create TVM runtime and do inference
@@ -496,7 +496,7 @@ 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> ( 2 minutes 24.100 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> ( 2 minutes 24.274 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 7b7405ff5..8fe020c08 100644
--- a/docs/how_to/deploy_models/sg_execution_times.html
+++ b/docs/how_to/deploy_models/sg_execution_times.html
@@ -322,7 +322,7 @@
<div class="section" id="computation-times">
<span id="sphx-glr-how-to-deploy-models-sg-execution-times"></span><h1>Computation times<a class="headerlink" href="#computation-times" title="Permalink to this headline">¶</a></h1>
-<p><strong>10:47.156</strong> total execution time for <strong>how_to_deploy_models</strong> files:</p>
+<p><strong>10:41.384</strong> total execution time for <strong>how_to_deploy_models</strong> files:</p>
<table class="docutils align-default">
<colgroup>
<col style="width: 86%" />
@@ -331,35 +331,35 @@
</colgroup>
<tbody>
<tr class="row-odd"><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>02:55.692</p></td>
+<td><p>02:57.631</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><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>02:24.100</p></td>
+<td><p>02:24.274</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>02:01.639</p></td>
+<td><p>01:59.854</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>01:25.372</p></td>
+<td><p>01:19.144</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>01:08.038</p></td>
+<td><p>01:07.701</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>00:29.707</p></td>
+<td><p>00:29.762</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="deploy_model_on_rasp.html#sphx-glr-how-to-deploy-models-deploy-model-on-rasp-py"><span class="std std-ref">Deploy the Pretrained Model on Raspberry Pi</span></a> (<code class="docutils literal notranslate"><span class="pre">deploy_model_on_rasp.py</span></code>)</p></td>
-<td><p>00:22.603</p></td>
+<td><p>00:23.011</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><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>
-<td><p>00:00.007</p></td>
+<td><p>00:00.006</p></td>
<td><p>0.0 MB</p></td>
</tr>
</tbody>
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 250d8d59a..b8e3f468b 100644
--- a/docs/how_to/extend_tvm/bring_your_own_datatypes.html
+++ b/docs/how_to/extend_tvm/bring_your_own_datatypes.html
@@ -607,7 +607,7 @@ 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.zip38f87827-2bbd-487d-90d7-9352b529a175 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.zip1a99afe2-ef49-4e65-b085-29d2f003d48b from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/mobilenet0.25-9f83e440.zip...
</pre></div>
</div>
<p>It’s easy to execute MobileNet with native TVM:</p>
@@ -671,7 +671,7 @@ In this alpha state of the Bring Your Own Datatypes framework, we have not imple
</div>
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>/workspace/python/tvm/driver/build_module.py:268: UserWarning: target_host parameter is going to be deprecated. Please pass in tvm.target.Target(target, host=target_host) instead.
"target_host parameter is going to be deprecated. "
- Check failed: (lower) is false: Intrinsic lowering function for target llvm, intrinsic name tir.sqrt, type 150 not found
+ Check failed: (lower) is false: FloatImm lowering function for target llvm type 150 not found
</pre></div>
</div>
<p>When we attempt to run the model, we get a familiar error telling us that more functions need to be registered for myfloat.</p>
diff --git a/docs/how_to/extend_tvm/sg_execution_times.html b/docs/how_to/extend_tvm/sg_execution_times.html
index 457bd11da..455b41fdf 100644
--- a/docs/how_to/extend_tvm/sg_execution_times.html
+++ b/docs/how_to/extend_tvm/sg_execution_times.html
@@ -322,7 +322,7 @@
<div class="section" id="computation-times">
<span id="sphx-glr-how-to-extend-tvm-sg-execution-times"></span><h1>Computation times<a class="headerlink" href="#computation-times" title="Permalink to this headline">¶</a></h1>
-<p><strong>00:40.080</strong> total execution time for <strong>how_to_extend_tvm</strong> files:</p>
+<p><strong>00:38.584</strong> total execution time for <strong>how_to_extend_tvm</strong> files:</p>
<table class="docutils align-default">
<colgroup>
<col style="width: 84%" />
@@ -331,19 +331,19 @@
</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>00:36.847</p></td>
+<td><p>00:35.580</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>00:02.276</p></td>
+<td><p>00:02.109</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>00:00.949</p></td>
+<td><p>00:00.887</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="low_level_custom_pass.html#sphx-glr-how-to-extend-tvm-low-level-custom-pass-py"><span class="std std-ref">Writing a Customized Pass</span></a> (<code class="docutils literal notranslate"><span class="pre">low_level_custom_pass.py</span></code>)</p></td>
-<td><p>00:00.008</p></td>
+<td><p>00:00.007</p></td>
<td><p>0.0 MB</p></td>
</tr>
</tbody>
diff --git a/docs/how_to/extend_tvm/use_pass_instrument.html b/docs/how_to/extend_tvm/use_pass_instrument.html
index 0adbf371a..f9b48446f 100644
--- a/docs/how_to/extend_tvm/use_pass_instrument.html
+++ b/docs/how_to/extend_tvm/use_pass_instrument.html
@@ -507,10 +507,10 @@ 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: 6840us [6840us] (45.60%; 45.60%)
-FoldScaleAxis: 8161us [6us] (54.40%; 54.40%)
- FoldConstant: 8156us [1652us] (54.36%; 99.93%)
- InferType: 6503us [6503us] (43.35%; 79.74%)
+InferType: 6354us [6354us] (45.75%; 45.75%)
+FoldScaleAxis: 7534us [5us] (54.25%; 54.25%)
+ FoldConstant: 7529us [1590us] (54.21%; 99.93%)
+ InferType: 5939us [5939us] (42.76%; 78.88%)
</pre></div>
</div>
</div>
@@ -532,10 +532,10 @@ 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: 6597us [6597us] (44.00%; 44.00%)
-FoldScaleAxis: 8397us [6us] (56.00%; 56.00%)
- FoldConstant: 8391us [1711us] (55.96%; 99.93%)
- InferType: 6680us [6680us] (44.55%; 79.61%)
+InferType: 6062us [6062us] (44.88%; 44.88%)
+FoldScaleAxis: 7445us [5us] (55.12%; 55.12%)
+ FoldConstant: 7440us [1526us] (55.08%; 99.94%)
+ InferType: 5914us [5914us] (43.79%; 79.49%)
</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 c493cd697..aceee381c 100644
--- a/docs/how_to/optimize_operators/opt_conv_cuda.html
+++ b/docs/how_to/optimize_operators/opt_conv_cuda.html
@@ -559,7 +559,7 @@ 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: 54.203512 ms
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Convolution: 54.167578 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 f76161a01..e82a0fbdf 100644
--- a/docs/how_to/optimize_operators/opt_conv_tensorcore.html
+++ b/docs/how_to/optimize_operators/opt_conv_tensorcore.html
@@ -901,7 +901,7 @@ 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: 6.885543 ms
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>conv2d with tensor core: 6.995651 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 abaa25ec5..33f96dacb 100644
--- a/docs/how_to/optimize_operators/opt_gemm.html
+++ b/docs/how_to/optimize_operators/opt_gemm.html
@@ -456,8 +456,8 @@ 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: 0.018863
-Baseline: 3.335876
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Numpy running time: 0.019033
+Baseline: 3.534831
</pre></div>
</div>
<p>In TVM, we can always inspect lower level IR to debug or optimize our schedule.
@@ -517,7 +517,7 @@ 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: 0.306134
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt1: 0.305472
</pre></div>
</div>
<p>Here is the generated IR after blocking.</p>
@@ -584,7 +584,7 @@ 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: 0.339910
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt2: 0.345996
</pre></div>
</div>
<p>Here is the generated IR after vectorization.</p>
@@ -645,7 +645,7 @@ 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: 0.115891
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt3: 0.121514
</pre></div>
</div>
<p>Here is the generated IR after loop permutation.</p>
@@ -728,7 +728,7 @@ 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: 0.110579
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt4: 0.110981
</pre></div>
</div>
<p>Here is the generated IR after array packing.</p>
@@ -814,7 +814,7 @@ 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: 0.111304
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt5: 0.111529
</pre></div>
</div>
<p>Here is the generated IR after blocking.</p>
@@ -904,7 +904,7 @@ write to C when all the block results are ready.</p>
<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: 0.145312
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt6: 0.145572
</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 39ef2d9b3..56b1f1300 100644
--- a/docs/how_to/optimize_operators/sg_execution_times.html
+++ b/docs/how_to/optimize_operators/sg_execution_times.html
@@ -322,7 +322,7 @@
<div class="section" id="computation-times">
<span id="sphx-glr-how-to-optimize-operators-sg-execution-times"></span><h1>Computation times<a class="headerlink" href="#computation-times" title="Permalink to this headline">¶</a></h1>
-<p><strong>00:34.430</strong> total execution time for <strong>how_to_optimize_operators</strong> files:</p>
+<p><strong>00:35.226</strong> total execution time for <strong>how_to_optimize_operators</strong> files:</p>
<table class="docutils align-default">
<colgroup>
<col style="width: 83%" />
@@ -331,15 +331,15 @@
</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>00:32.091</p></td>
+<td><p>00:32.887</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>00:01.297</p></td>
+<td><p>00:01.293</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>00:01.042</p></td>
+<td><p>00:01.046</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 8862c47e2..9334e2bb3 100644
--- a/docs/how_to/tune_with_autoscheduler/sg_execution_times.html
+++ b/docs/how_to/tune_with_autoscheduler/sg_execution_times.html
@@ -322,7 +322,7 @@
<div class="section" id="computation-times">
<span id="sphx-glr-how-to-tune-with-autoscheduler-sg-execution-times"></span><h1>Computation times<a class="headerlink" href="#computation-times" title="Permalink to this headline">¶</a></h1>
-<p><strong>05:08.137</strong> total execution time for <strong>how_to_tune_with_autoscheduler</strong> files:</p>
+<p><strong>05:16.608</strong> total execution time for <strong>how_to_tune_with_autoscheduler</strong> files:</p>
<table class="docutils align-default">
<colgroup>
<col style="width: 85%" />
@@ -331,27 +331,27 @@
</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>02:29.143</p></td>
+<td><p>02:35.377</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>01:20.297</p></td>
+<td><p>01:19.283</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>00:43.367</p></td>
+<td><p>00:42.637</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>00:17.854</p></td>
+<td><p>00:22.459</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><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>00:08.906</p></td>
+<td><p>00:08.478</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><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>00:08.570</p></td>
+<td><p>00:08.374</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 1346b86c8..f3e6082e7 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
@@ -486,355 +486,762 @@ cooperative fetching, unrolling and operator fusion.</p>
compute: Buffer(compute_2: Pointer(float32), float32, [25088], [])}
buffer_map = {data_1: data, kernel_1: kernel, bias_1: bias, compute_1: compute}
preflattened_buffer_map = {data_1: data_3: Buffer(data_2, float32, [1, 512, 7, 7], []), kernel_1: kernel_3: Buffer(kernel_2, float32, [512, 512, 3, 3], []), bias_1: bias_3: Buffer(bias_2, float32, [1, 512, 1, 1], []), compute_1: compute_3: Buffer(compute_2, float32, [1, 512, 7, 7], [])} {
- attr [IterVar(blockIdx.x: int32, (nullptr), "ThreadIndex", "blockIdx.x")] "thread_extent" = 64;
- allocate(conv2d_nchw: Pointer(local float32), float32, [8]), storage_scope = local;
- allocate(pad_temp.shared: Pointer(shared float32), float32, [324]), storage_scope = shared;
- allocate(kernel.shared: Pointer(shared float32), float32, [288]), storage_scope = shared;
+ attr [IterVar(blockIdx.x: int32, (nullptr), "ThreadIndex", "blockIdx.x")] "thread_extent" = 128;
+ allocate(conv2d_nchw: Pointer(local float32), float32, [4]), storage_scope = local;
+ allocate(pad_temp.shared: Pointer(shared float32), float32, [1008]), storage_scope = shared;
+ allocate(kernel.shared: Pointer(shared float32), float32, [192]), storage_scope = shared;
attr [IterVar(threadIdx.x: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- conv2d_nchw_1: Buffer(conv2d_nchw, float32, [1], [], scope="local", align=4)[0] = 0f32
+ conv2d_nchw_1: Buffer(conv2d_nchw, float32, [4], [], scope="local", align=16)[0] = 0f32
conv2d_nchw_1[1] = 0f32
conv2d_nchw_1[2] = 0f32
conv2d_nchw_1[3] = 0f32
- conv2d_nchw_1[4] = 0f32
- conv2d_nchw_1[5] = 0f32
- conv2d_nchw_1[6] = 0f32
- conv2d_nchw_1[7] = 0f32
- for (rc.outer.outer: int32, 0, 128) {
- let cse_var_2: int32 = (rc.outer.outer*196)
- let cse_var_1: int32 = (rc.outer.outer*36)
+ for (rc.outer.outer: int32, 0, 32) {
+ let cse_var_2: int32 = (rc.outer.outer*784)
+ let cse_var_1: int32 = (rc.outer.outer*144)
{
- attr [IterVar(threadIdx.x_1: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1: Buffer(pad_temp.shared, float32, [324], [], scope="shared")[(threadIdx.x_1*2)] = @tir.if_then_else(((((9 <= floormod((threadIdx.x_1*2), 81)) && (floormod((threadIdx.x_1*2), 81) < 72)) && (1 <= floormod((threadIdx.x_1*2), 9))) && (floormod((threadIdx.x_1*2), 9) < 8)), data[((((cse_var_2 + (floordiv((threadIdx.x_1*2), 81)*49)) + (floordiv(floormod((threadIdx.x_1*2), 81), 9)*7)) + floormod((threadIdx.x_1*2), 9)) - 8) [...]
- pad_temp.shared_1[((threadIdx.x_1*2) + 1)] = @tir.if_then_else(((((9 <= floormod(((threadIdx.x_1*2) + 1), 81)) && (floormod(((threadIdx.x_1*2) + 1), 81) < 72)) && (1 <= floormod(((threadIdx.x_1*2) + 1), 9))) && (floormod(((threadIdx.x_1*2) + 1), 9) < 8)), data[((((cse_var_2 + (floordiv(((threadIdx.x_1*2) + 1), 81)*49)) + (floordiv(floormod(((threadIdx.x_1*2) + 1), 81), 9)*7)) + floormod(((threadIdx.x_1*2) + 1), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1: Buffer(pad_temp.shared, float32, [1008], [], scope="shared")[threadIdx.x_1] = @tir.if_then_else((((9 <= threadIdx.x_1) && (1 <= floormod(threadIdx.x_1, 9))) && (floormod(threadIdx.x_1, 9) < 8)), data[(((cse_var_2 + (floordiv(threadIdx.x_1, 9)*7)) + floormod(threadIdx.x_1, 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 49)] = @tir.if_then_else((((9 <= floormod((threadIdx.x_1 + 49), 63)) && (1 <= floormod((threadIdx.x_1 + 4), 9))) && (floormod((threadIdx.x_1 + 4), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 49), 9)*7)) + floormod((threadIdx.x_1 + 4), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 98)] = @tir.if_then_else((((9 <= floormod((threadIdx.x_1 + 35), 63)) && (1 <= floormod((threadIdx.x_1 + 8), 9))) && (floormod((threadIdx.x_1 + 8), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 98), 9)*7)) + floormod((threadIdx.x_1 + 8), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 147)] = @tir.if_then_else((((9 <= floormod((threadIdx.x_1 + 21), 63)) && (1 <= floormod((threadIdx.x_1 + 3), 9))) && (floormod((threadIdx.x_1 + 3), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 147), 9)*7)) + floormod((threadIdx.x_1 + 3), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 196)] = @tir.if_then_else((((9 <= floormod((threadIdx.x_1 + 7), 63)) && (1 <= floormod((threadIdx.x_1 + 7), 9))) && (floormod((threadIdx.x_1 + 7), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 196), 9)*7)) + floormod((threadIdx.x_1 + 7), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 245)] = @tir.if_then_else((((9 <= floormod((threadIdx.x_1 + 56), 63)) && (1 <= floormod((threadIdx.x_1 + 2), 9))) && (floormod((threadIdx.x_1 + 2), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 245), 9)*7)) + floormod((threadIdx.x_1 + 2), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 294)] = @tir.if_then_else((((9 <= floormod((threadIdx.x_1 + 42), 63)) && (1 <= floormod((threadIdx.x_1 + 6), 9))) && (floormod((threadIdx.x_1 + 6), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 294), 9)*7)) + floormod((threadIdx.x_1 + 6), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 343)] = @tir.if_then_else((((9 <= floormod((threadIdx.x_1 + 28), 63)) && (1 <= floormod((threadIdx.x_1 + 1), 9))) && (floormod((threadIdx.x_1 + 1), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 343), 9)*7)) + floormod((threadIdx.x_1 + 1), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 392)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 5), 9)) && (floormod((threadIdx.x_1 + 5), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 392), 9)*7)) + floormod((threadIdx.x_1 + 5), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 441)] = @tir.if_then_else((((9 <= threadIdx.x_1) && (1 <= floormod(threadIdx.x_1, 9))) && (floormod(threadIdx.x_1, 9) < 8)), data[(((cse_var_2 + (floordiv(threadIdx.x_1, 9)*7)) + floormod(threadIdx.x_1, 9)) + 335)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 490)] = @tir.if_then_else((((9 <= floormod((threadIdx.x_1 + 49), 63)) && (1 <= floormod((threadIdx.x_1 + 4), 9))) && (floormod((threadIdx.x_1 + 4), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 490), 9)*7)) + floormod((threadIdx.x_1 + 4), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 539)] = @tir.if_then_else((((9 <= floormod((threadIdx.x_1 + 35), 63)) && (1 <= floormod((threadIdx.x_1 + 8), 9))) && (floormod((threadIdx.x_1 + 8), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 539), 9)*7)) + floormod((threadIdx.x_1 + 8), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 588)] = @tir.if_then_else((((9 <= floormod((threadIdx.x_1 + 21), 63)) && (1 <= floormod((threadIdx.x_1 + 3), 9))) && (floormod((threadIdx.x_1 + 3), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 588), 9)*7)) + floormod((threadIdx.x_1 + 3), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 637)] = @tir.if_then_else((((9 <= floormod((threadIdx.x_1 + 7), 63)) && (1 <= floormod((threadIdx.x_1 + 7), 9))) && (floormod((threadIdx.x_1 + 7), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 637), 9)*7)) + floormod((threadIdx.x_1 + 7), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 686)] = @tir.if_then_else((((9 <= floormod((threadIdx.x_1 + 56), 63)) && (1 <= floormod((threadIdx.x_1 + 2), 9))) && (floormod((threadIdx.x_1 + 2), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 686), 9)*7)) + floormod((threadIdx.x_1 + 2), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 735)] = @tir.if_then_else((((9 <= floormod((threadIdx.x_1 + 42), 63)) && (1 <= floormod((threadIdx.x_1 + 6), 9))) && (floormod((threadIdx.x_1 + 6), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 735), 9)*7)) + floormod((threadIdx.x_1 + 6), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 784)] = @tir.if_then_else((((9 <= floormod((threadIdx.x_1 + 28), 63)) && (1 <= floormod((threadIdx.x_1 + 1), 9))) && (floormod((threadIdx.x_1 + 1), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 784), 9)*7)) + floormod((threadIdx.x_1 + 1), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 833)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 5), 9)) && (floormod((threadIdx.x_1 + 5), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 833), 9)*7)) + floormod((threadIdx.x_1 + 5), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 882)] = @tir.if_then_else((((9 <= threadIdx.x_1) && (1 <= floormod(threadIdx.x_1, 9))) && (floormod(threadIdx.x_1, 9) < 8)), data[(((cse_var_2 + (floordiv(threadIdx.x_1, 9)*7)) + floormod(threadIdx.x_1, 9)) + 678)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 931)] = @tir.if_then_else((((9 <= floormod((threadIdx.x_1 + 49), 63)) && (1 <= floormod((threadIdx.x_1 + 4), 9))) && (floormod((threadIdx.x_1 + 4), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 931), 9)*7)) + floormod((threadIdx.x_1 + 4), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ if @tir.likely((threadIdx.x_1 < 28), dtype=bool) {
+ pad_temp.shared_1[(threadIdx.x_1 + 980)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 8), 9)) && (floormod((threadIdx.x_1 + 8), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 980), 9)*7)) + floormod((threadIdx.x_1 + 8), 9)) - 8)], 0f32, dtype=float32)
}
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[((threadIdx.x_1*2) + 98)] = @tir.if_then_else(((((9 <= floormod(((threadIdx.x_1*2) + 17), 81)) && (floormod(((threadIdx.x_1*2) + 17), 81) < 72)) && (1 <= floormod(((threadIdx.x_1*2) + 8), 9))) && (floormod(((threadIdx.x_1*2) + 8), 9) < 8)), data[((((cse_var_2 + (floordiv(((threadIdx.x_1*2) + 98), 81)*49)) + (floordiv(floormod(((threadIdx.x_1*2) + 17), 81), 9)*7)) + floormod(((threadIdx.x_1*2) + 8), 9)) - 8)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*2) + 99)] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*2), 9) + 2), 9)) && (floormod(((threadIdx.x_1*2) + 18), 81) < 72)) && (1 <= floormod((threadIdx.x_1*2), 9))) && (floormod((threadIdx.x_1*2), 9) < 8)), data[((((cse_var_2 + (floordiv(((threadIdx.x_1*2) + 99), 81)*49)) + (floormod((floordiv((threadIdx.x_1*2), 9) + 2), 9)*7)) + floormod((threadIdx.x_1*2), 9)) - 8)], 0f32, dtype=float32)
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[((threadIdx.x_1*2) + 196)] = @tir.if_then_else(((((9 <= floormod(((threadIdx.x_1*2) + 34), 81)) && (floormod(((threadIdx.x_1*2) + 34), 81) < 72)) && (1 <= floormod(((threadIdx.x_1*2) + 7), 9))) && (floormod(((threadIdx.x_1*2) + 7), 9) < 8)), data[((((cse_var_2 + (floordiv(((threadIdx.x_1*2) + 196), 81)*49)) + (floordiv(floormod(((threadIdx.x_1*2) + 34), 81), 9)*7)) + floormod(((threadIdx.x_1*2) + 7), 9)) - 8)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*2) + 197)] = @tir.if_then_else(((((9 <= floormod(((threadIdx.x_1*2) + 35), 81)) && (floormod(((threadIdx.x_1*2) + 35), 81) < 72)) && (1 <= floormod(((threadIdx.x_1*2) + 8), 9))) && (floormod(((threadIdx.x_1*2) + 8), 9) < 8)), data[((((cse_var_2 + (floordiv(((threadIdx.x_1*2) + 197), 81)*49)) + (floordiv(floormod(((threadIdx.x_1*2) + 35), 81), 9)*7)) + floormod(((threadIdx.x_1*2) + 8), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_2: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ kernel.shared_1: Buffer(kernel.shared, float32, [192], [], scope="shared")[threadIdx.x_2] = kernel[(((((blockIdx.x*18432) + (floordiv(threadIdx.x_2, 48)*4608)) + cse_var_1) + (floordiv(floormod(threadIdx.x_2, 48), 3)*9)) + floormod(threadIdx.x_2, 3))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ kernel.shared_1[(threadIdx.x_2 + 49)] = kernel[(((((blockIdx.x*18432) + (floordiv((threadIdx.x_2 + 49), 48)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 1), 48), 3)*9)) + floormod((threadIdx.x_2 + 1), 3))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ kernel.shared_1[(threadIdx.x_2 + 98)] = kernel[(((((blockIdx.x*18432) + (floordiv((threadIdx.x_2 + 98), 48)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 2), 48), 3)*9)) + floormod((threadIdx.x_2 + 2), 3))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ if @tir.likely((threadIdx.x_2 < 45), dtype=bool) {
+ kernel.shared_1[(threadIdx.x_2 + 147)] = kernel[((((((blockIdx.x*18432) + (floordiv((threadIdx.x_2 + 147), 48)*4608)) + cse_var_1) + (floordiv(threadIdx.x_2, 3)*9)) + floormod(threadIdx.x_2, 3)) + 9)]
}
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- if @tir.likely((threadIdx.x_1 < 15), dtype=bool) {
- pad_temp.shared_1[((threadIdx.x_1*2) + 294)] = @tir.if_then_else((((threadIdx.x_1 < 11) && (1 <= floormod(((threadIdx.x_1*2) + 6), 9))) && (floormod(((threadIdx.x_1*2) + 6), 9) < 8)), data[(((cse_var_2 + (floordiv(floormod(((threadIdx.x_1*2) + 51), 81), 9)*7)) + floormod(((threadIdx.x_1*2) + 6), 9)) + 139)], 0f32, dtype=float32)
- }
- if @tir.likely((threadIdx.x_1 < 15), dtype=bool) {
- pad_temp.shared_1[((threadIdx.x_1*2) + 295)] = @tir.if_then_else((((threadIdx.x_1 < 10) && (1 <= floormod(((threadIdx.x_1*2) + 7), 9))) && (floormod(((threadIdx.x_1*2) + 7), 9) < 8)), data[(((cse_var_2 + (floordiv(floormod(((threadIdx.x_1*2) + 52), 81), 9)*7)) + floormod(((threadIdx.x_1*2) + 7), 9)) + 139)], 0f32, dtype=float32)
- }
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[0]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[48]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[96]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[144]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[1]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[49]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[97]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[145]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[2]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[50]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[98]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[146]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[3]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[51]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[99]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[147]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[4]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[52]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[100]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[148]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[5]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[53]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[101]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[149]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[6]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[54]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[102]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[150]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[7]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[55]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[103]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[151]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[8]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[56]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[104]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[152]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[9]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[57]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[105]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[153]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[10]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[58]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[106]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[154]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[11]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[59]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[107]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[155]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[12]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[60]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[108]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[156]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[13]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[61]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[109]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[157]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[14]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[62]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[110]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[158]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 315)]*kernel.shared_1[15]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 315)]*kernel.shared_1[63]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 315)]*kernel.shared_1[111]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 315)]*kernel.shared_1[159]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 316)]*kernel.shared_1[16]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 316)]*kernel.shared_1[64]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 316)]*kernel.shared_1[112]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 316)]*kernel.shared_1[160]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 317)]*kernel.shared_1[17]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 317)]*kernel.shared_1[65]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 317)]*kernel.shared_1[113]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 317)]*kernel.shared_1[161]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 378)]*kernel.shared_1[18]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 378)]*kernel.shared_1[66]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 378)]*kernel.shared_1[114]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 378)]*kernel.shared_1[162]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 379)]*kernel.shared_1[19]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 379)]*kernel.shared_1[67]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 379)]*kernel.shared_1[115]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 379)]*kernel.shared_1[163]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 380)]*kernel.shared_1[20]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 380)]*kernel.shared_1[68]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 380)]*kernel.shared_1[116]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 380)]*kernel.shared_1[164]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 441)]*kernel.shared_1[21]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 441)]*kernel.shared_1[69]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 441)]*kernel.shared_1[117]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 441)]*kernel.shared_1[165]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 442)]*kernel.shared_1[22]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 442)]*kernel.shared_1[70]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 442)]*kernel.shared_1[118]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 442)]*kernel.shared_1[166]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 443)]*kernel.shared_1[23]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 443)]*kernel.shared_1[71]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 443)]*kernel.shared_1[119]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 443)]*kernel.shared_1[167]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 504)]*kernel.shared_1[24]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 504)]*kernel.shared_1[72]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 504)]*kernel.shared_1[120]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 504)]*kernel.shared_1[168]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 505)]*kernel.shared_1[25]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 505)]*kernel.shared_1[73]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 505)]*kernel.shared_1[121]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 505)]*kernel.shared_1[169]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 506)]*kernel.shared_1[26]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 506)]*kernel.shared_1[74]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 506)]*kernel.shared_1[122]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 506)]*kernel.shared_1[170]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 567)]*kernel.shared_1[27]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 567)]*kernel.shared_1[75]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 567)]*kernel.shared_1[123]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 567)]*kernel.shared_1[171]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 568)]*kernel.shared_1[28]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 568)]*kernel.shared_1[76]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 568)]*kernel.shared_1[124]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 568)]*kernel.shared_1[172]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 569)]*kernel.shared_1[29]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 569)]*kernel.shared_1[77]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 569)]*kernel.shared_1[125]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 569)]*kernel.shared_1[173]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 630)]*kernel.shared_1[30]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 630)]*kernel.shared_1[78]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 630)]*kernel.shared_1[126]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 630)]*kernel.shared_1[174]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 631)]*kernel.shared_1[31]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 631)]*kernel.shared_1[79]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 631)]*kernel.shared_1[127]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 631)]*kernel.shared_1[175]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 632)]*kernel.shared_1[32]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 632)]*kernel.shared_1[80]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 632)]*kernel.shared_1[128]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 632)]*kernel.shared_1[176]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 693)]*kernel.shared_1[33]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 693)]*kernel.shared_1[81]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 693)]*kernel.shared_1[129]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 693)]*kernel.shared_1[177]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 694)]*kernel.shared_1[34]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 694)]*kernel.shared_1[82]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 694)]*kernel.shared_1[130]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 694)]*kernel.shared_1[178]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 695)]*kernel.shared_1[35]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 695)]*kernel.shared_1[83]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 695)]*kernel.shared_1[131]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 695)]*kernel.shared_1[179]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 756)]*kernel.shared_1[36]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 756)]*kernel.shared_1[84]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 756)]*kernel.shared_1[132]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 756)]*kernel.shared_1[180]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 757)]*kernel.shared_1[37]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 757)]*kernel.shared_1[85]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 757)]*kernel.shared_1[133]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 757)]*kernel.shared_1[181]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 758)]*kernel.shared_1[38]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 758)]*kernel.shared_1[86]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 758)]*kernel.shared_1[134]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 758)]*kernel.shared_1[182]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 819)]*kernel.shared_1[39]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 819)]*kernel.shared_1[87]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 819)]*kernel.shared_1[135]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 819)]*kernel.shared_1[183]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 820)]*kernel.shared_1[40]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 820)]*kernel.shared_1[88]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 820)]*kernel.shared_1[136]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 820)]*kernel.shared_1[184]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 821)]*kernel.shared_1[41]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 821)]*kernel.shared_1[89]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 821)]*kernel.shared_1[137]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 821)]*kernel.shared_1[185]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 882)]*kernel.shared_1[42]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 882)]*kernel.shared_1[90]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 882)]*kernel.shared_1[138]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 882)]*kernel.shared_1[186]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 883)]*kernel.shared_1[43]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 883)]*kernel.shared_1[91]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 883)]*kernel.shared_1[139]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 883)]*kernel.shared_1[187]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 884)]*kernel.shared_1[44]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 884)]*kernel.shared_1[92]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 884)]*kernel.shared_1[140]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 884)]*kernel.shared_1[188]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 945)]*kernel.shared_1[45]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 945)]*kernel.shared_1[93]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 945)]*kernel.shared_1[141]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 945)]*kernel.shared_1[189]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 946)]*kernel.shared_1[46]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 946)]*kernel.shared_1[94]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 946)]*kernel.shared_1[142]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 946)]*kernel.shared_1[190]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 947)]*kernel.shared_1[47]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 947)]*kernel.shared_1[95]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 947)]*kernel.shared_1[143]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 947)]*kernel.shared_1[191]))
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[threadIdx.x_1] = @tir.if_then_else(((1 <= floormod(threadIdx.x_1, 9)) && (floormod(threadIdx.x_1, 9) < 8)), data[(((cse_var_2 + (floordiv(threadIdx.x_1, 9)*7)) + floormod(threadIdx.x_1, 9)) - 1)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 49)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 4), 9)) && (floormod((threadIdx.x_1 + 4), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 49), 9)*7)) + floormod((threadIdx.x_1 + 4), 9)) - 1)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 98)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 8), 9)) && (floormod((threadIdx.x_1 + 8), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 98), 9)*7)) + floormod((threadIdx.x_1 + 8), 9)) - 1)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 147)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 3), 9)) && (floormod((threadIdx.x_1 + 3), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 147), 9)*7)) + floormod((threadIdx.x_1 + 3), 9)) - 1)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 196)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 7), 9)) && (floormod((threadIdx.x_1 + 7), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 196), 9)*7)) + floormod((threadIdx.x_1 + 7), 9)) - 1)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 245)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 2), 9)) && (floormod((threadIdx.x_1 + 2), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 245), 9)*7)) + floormod((threadIdx.x_1 + 2), 9)) - 1)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 294)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 6), 9)) && (floormod((threadIdx.x_1 + 6), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 294), 9)*7)) + floormod((threadIdx.x_1 + 6), 9)) - 1)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 343)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 1), 9)) && (floormod((threadIdx.x_1 + 1), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 343), 9)*7)) + floormod((threadIdx.x_1 + 1), 9)) - 1)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 392)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 5), 9)) && (floormod((threadIdx.x_1 + 5), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 392), 9)*7)) + floormod((threadIdx.x_1 + 5), 9)) - 1)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 441)] = @tir.if_then_else(((1 <= floormod(threadIdx.x_1, 9)) && (floormod(threadIdx.x_1, 9) < 8)), data[(((cse_var_2 + (floordiv(threadIdx.x_1, 9)*7)) + floormod(threadIdx.x_1, 9)) + 342)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 490)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 4), 9)) && (floormod((threadIdx.x_1 + 4), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 490), 9)*7)) + floormod((threadIdx.x_1 + 4), 9)) - 1)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 539)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 8), 9)) && (floormod((threadIdx.x_1 + 8), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 539), 9)*7)) + floormod((threadIdx.x_1 + 8), 9)) - 1)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 588)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 3), 9)) && (floormod((threadIdx.x_1 + 3), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 588), 9)*7)) + floormod((threadIdx.x_1 + 3), 9)) - 1)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 637)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 7), 9)) && (floormod((threadIdx.x_1 + 7), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 637), 9)*7)) + floormod((threadIdx.x_1 + 7), 9)) - 1)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 686)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 2), 9)) && (floormod((threadIdx.x_1 + 2), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 686), 9)*7)) + floormod((threadIdx.x_1 + 2), 9)) - 1)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 735)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 6), 9)) && (floormod((threadIdx.x_1 + 6), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 735), 9)*7)) + floormod((threadIdx.x_1 + 6), 9)) - 1)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 784)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 1), 9)) && (floormod((threadIdx.x_1 + 1), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 784), 9)*7)) + floormod((threadIdx.x_1 + 1), 9)) - 1)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 833)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 5), 9)) && (floormod((threadIdx.x_1 + 5), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 833), 9)*7)) + floormod((threadIdx.x_1 + 5), 9)) - 1)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 882)] = @tir.if_then_else(((1 <= floormod(threadIdx.x_1, 9)) && (floormod(threadIdx.x_1, 9) < 8)), data[(((cse_var_2 + (floordiv(threadIdx.x_1, 9)*7)) + floormod(threadIdx.x_1, 9)) + 685)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 931)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 4), 9)) && (floormod((threadIdx.x_1 + 4), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 931), 9)*7)) + floormod((threadIdx.x_1 + 4), 9)) - 1)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ if @tir.likely((threadIdx.x_1 < 28), dtype=bool) {
+ pad_temp.shared_1[(threadIdx.x_1 + 980)] = @tir.if_then_else(((1 <= floormod((threadIdx.x_1 + 8), 9)) && (floormod((threadIdx.x_1 + 8), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 980), 9)*7)) + floormod((threadIdx.x_1 + 8), 9)) - 1)], 0f32, dtype=float32)
}
- attr [IterVar(threadIdx.x_2: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1: Buffer(kernel.shared, float32, [288], [], scope="shared")[threadIdx.x_2] = kernel[((((blockIdx.x*36864) + (floordiv(threadIdx.x_2, 36)*4608)) + cse_var_1) + floormod(threadIdx.x_2, 36))]
attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 49)] = kernel[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 49), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 13), 36), 9)*9)) + (floordiv(floormod((threadIdx.x_2 + 4), 9), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
+ kernel.shared_1[threadIdx.x_2] = kernel[((((((blockIdx.x*18432) + (floordiv(threadIdx.x_2, 48)*4608)) + cse_var_1) + (floordiv(floormod(threadIdx.x_2, 48), 3)*9)) + floormod(threadIdx.x_2, 3)) + 3)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ kernel.shared_1[(threadIdx.x_2 + 49)] = kernel[((((((blockIdx.x*18432) + (floordiv((threadIdx.x_2 + 49), 48)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 1), 48), 3)*9)) + floormod((threadIdx.x_2 + 1), 3)) + 3)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ kernel.shared_1[(threadIdx.x_2 + 98)] = kernel[((((((blockIdx.x*18432) + (floordiv((threadIdx.x_2 + 98), 48)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 2), 48), 3)*9)) + floormod((threadIdx.x_2 + 2), 3)) + 3)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ if @tir.likely((threadIdx.x_2 < 45), dtype=bool) {
+ kernel.shared_1[(threadIdx.x_2 + 147)] = kernel[((((((blockIdx.x*18432) + (floordiv((threadIdx.x_2 + 147), 48)*4608)) + cse_var_1) + (floordiv(threadIdx.x_2, 3)*9)) + floormod(threadIdx.x_2, 3)) + 12)]
+ }
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[0]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[48]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[96]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[144]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[1]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[49]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[97]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[145]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[2]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[50]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[98]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[146]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[3]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[51]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[99]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[147]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[4]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[52]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[100]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[148]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[5]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[53]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[101]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[149]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[6]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[54]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[102]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[150]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[7]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[55]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[103]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[151]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[8]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[56]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[104]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[152]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[9]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[57]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[105]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[153]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[10]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[58]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[106]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[154]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[11]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[59]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[107]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[155]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[12]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[60]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[108]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[156]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[13]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[61]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[109]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[157]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[14]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[62]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[110]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[158]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 315)]*kernel.shared_1[15]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 315)]*kernel.shared_1[63]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 315)]*kernel.shared_1[111]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 315)]*kernel.shared_1[159]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 316)]*kernel.shared_1[16]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 316)]*kernel.shared_1[64]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 316)]*kernel.shared_1[112]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 316)]*kernel.shared_1[160]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 317)]*kernel.shared_1[17]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 317)]*kernel.shared_1[65]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 317)]*kernel.shared_1[113]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 317)]*kernel.shared_1[161]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 378)]*kernel.shared_1[18]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 378)]*kernel.shared_1[66]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 378)]*kernel.shared_1[114]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 378)]*kernel.shared_1[162]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 379)]*kernel.shared_1[19]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 379)]*kernel.shared_1[67]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 379)]*kernel.shared_1[115]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 379)]*kernel.shared_1[163]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 380)]*kernel.shared_1[20]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 380)]*kernel.shared_1[68]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 380)]*kernel.shared_1[116]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 380)]*kernel.shared_1[164]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 441)]*kernel.shared_1[21]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 441)]*kernel.shared_1[69]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 441)]*kernel.shared_1[117]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 441)]*kernel.shared_1[165]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 442)]*kernel.shared_1[22]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 442)]*kernel.shared_1[70]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 442)]*kernel.shared_1[118]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 442)]*kernel.shared_1[166]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 443)]*kernel.shared_1[23]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 443)]*kernel.shared_1[71]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 443)]*kernel.shared_1[119]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 443)]*kernel.shared_1[167]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 504)]*kernel.shared_1[24]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 504)]*kernel.shared_1[72]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 504)]*kernel.shared_1[120]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 504)]*kernel.shared_1[168]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 505)]*kernel.shared_1[25]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 505)]*kernel.shared_1[73]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 505)]*kernel.shared_1[121]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 505)]*kernel.shared_1[169]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 506)]*kernel.shared_1[26]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 506)]*kernel.shared_1[74]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 506)]*kernel.shared_1[122]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 506)]*kernel.shared_1[170]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 567)]*kernel.shared_1[27]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 567)]*kernel.shared_1[75]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 567)]*kernel.shared_1[123]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 567)]*kernel.shared_1[171]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 568)]*kernel.shared_1[28]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 568)]*kernel.shared_1[76]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 568)]*kernel.shared_1[124]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 568)]*kernel.shared_1[172]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 569)]*kernel.shared_1[29]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 569)]*kernel.shared_1[77]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 569)]*kernel.shared_1[125]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 569)]*kernel.shared_1[173]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 630)]*kernel.shared_1[30]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 630)]*kernel.shared_1[78]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 630)]*kernel.shared_1[126]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 630)]*kernel.shared_1[174]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 631)]*kernel.shared_1[31]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 631)]*kernel.shared_1[79]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 631)]*kernel.shared_1[127]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 631)]*kernel.shared_1[175]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 632)]*kernel.shared_1[32]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 632)]*kernel.shared_1[80]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 632)]*kernel.shared_1[128]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 632)]*kernel.shared_1[176]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 693)]*kernel.shared_1[33]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 693)]*kernel.shared_1[81]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 693)]*kernel.shared_1[129]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 693)]*kernel.shared_1[177]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 694)]*kernel.shared_1[34]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 694)]*kernel.shared_1[82]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 694)]*kernel.shared_1[130]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 694)]*kernel.shared_1[178]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 695)]*kernel.shared_1[35]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 695)]*kernel.shared_1[83]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 695)]*kernel.shared_1[131]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 695)]*kernel.shared_1[179]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 756)]*kernel.shared_1[36]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 756)]*kernel.shared_1[84]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 756)]*kernel.shared_1[132]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 756)]*kernel.shared_1[180]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 757)]*kernel.shared_1[37]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 757)]*kernel.shared_1[85]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 757)]*kernel.shared_1[133]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 757)]*kernel.shared_1[181]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 758)]*kernel.shared_1[38]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 758)]*kernel.shared_1[86]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 758)]*kernel.shared_1[134]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 758)]*kernel.shared_1[182]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 819)]*kernel.shared_1[39]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 819)]*kernel.shared_1[87]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 819)]*kernel.shared_1[135]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 819)]*kernel.shared_1[183]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 820)]*kernel.shared_1[40]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 820)]*kernel.shared_1[88]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 820)]*kernel.shared_1[136]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 820)]*kernel.shared_1[184]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 821)]*kernel.shared_1[41]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 821)]*kernel.shared_1[89]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 821)]*kernel.shared_1[137]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 821)]*kernel.shared_1[185]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 882)]*kernel.shared_1[42]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 882)]*kernel.shared_1[90]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 882)]*kernel.shared_1[138]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 882)]*kernel.shared_1[186]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 883)]*kernel.shared_1[43]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 883)]*kernel.shared_1[91]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 883)]*kernel.shared_1[139]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 883)]*kernel.shared_1[187]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 884)]*kernel.shared_1[44]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 884)]*kernel.shared_1[92]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 884)]*kernel.shared_1[140]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 884)]*kernel.shared_1[188]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 945)]*kernel.shared_1[45]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 945)]*kernel.shared_1[93]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 945)]*kernel.shared_1[141]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 945)]*kernel.shared_1[189]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 946)]*kernel.shared_1[46]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 946)]*kernel.shared_1[94]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 946)]*kernel.shared_1[142]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 946)]*kernel.shared_1[190]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 947)]*kernel.shared_1[47]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 947)]*kernel.shared_1[95]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 947)]*kernel.shared_1[143]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 947)]*kernel.shared_1[191]))
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[threadIdx.x_1] = @tir.if_then_else(((1 <= floormod(threadIdx.x_1, 9)) && (floormod(threadIdx.x_1, 9) < 8)), data[(((cse_var_2 + (floordiv(threadIdx.x_1, 9)*7)) + floormod(threadIdx.x_1, 9)) + 6)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 49)] = @tir.if_then_else((((floormod((threadIdx.x_1 + 49), 63) < 54) && (1 <= floormod((threadIdx.x_1 + 4), 9))) && (floormod((threadIdx.x_1 + 4), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 49), 9)*7)) + floormod((threadIdx.x_1 + 4), 9)) + 6)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 98)] = @tir.if_then_else((((floormod((threadIdx.x_1 + 35), 63) < 54) && (1 <= floormod((threadIdx.x_1 + 8), 9))) && (floormod((threadIdx.x_1 + 8), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 98), 9)*7)) + floormod((threadIdx.x_1 + 8), 9)) + 6)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 147)] = @tir.if_then_else((((floormod((threadIdx.x_1 + 21), 63) < 54) && (1 <= floormod((threadIdx.x_1 + 3), 9))) && (floormod((threadIdx.x_1 + 3), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 147), 9)*7)) + floormod((threadIdx.x_1 + 3), 9)) + 6)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 196)] = @tir.if_then_else((((floormod((threadIdx.x_1 + 7), 63) < 54) && (1 <= floormod((threadIdx.x_1 + 7), 9))) && (floormod((threadIdx.x_1 + 7), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 196), 9)*7)) + floormod((threadIdx.x_1 + 7), 9)) + 6)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 245)] = @tir.if_then_else((((floormod((threadIdx.x_1 + 56), 63) < 54) && (1 <= floormod((threadIdx.x_1 + 2), 9))) && (floormod((threadIdx.x_1 + 2), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 245), 9)*7)) + floormod((threadIdx.x_1 + 2), 9)) + 6)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 294)] = @tir.if_then_else((((floormod((threadIdx.x_1 + 42), 63) < 54) && (1 <= floormod((threadIdx.x_1 + 6), 9))) && (floormod((threadIdx.x_1 + 6), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 294), 9)*7)) + floormod((threadIdx.x_1 + 6), 9)) + 6)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 343)] = @tir.if_then_else((((floormod((threadIdx.x_1 + 28), 63) < 54) && (1 <= floormod((threadIdx.x_1 + 1), 9))) && (floormod((threadIdx.x_1 + 1), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 343), 9)*7)) + floormod((threadIdx.x_1 + 1), 9)) + 6)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 392)] = @tir.if_then_else((((floormod((threadIdx.x_1 + 14), 63) < 54) && (1 <= floormod((threadIdx.x_1 + 5), 9))) && (floormod((threadIdx.x_1 + 5), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 392), 9)*7)) + floormod((threadIdx.x_1 + 5), 9)) + 6)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 441)] = @tir.if_then_else(((1 <= floormod(threadIdx.x_1, 9)) && (floormod(threadIdx.x_1, 9) < 8)), data[(((cse_var_2 + (floordiv(threadIdx.x_1, 9)*7)) + floormod(threadIdx.x_1, 9)) + 349)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 490)] = @tir.if_then_else((((floormod((threadIdx.x_1 + 49), 63) < 54) && (1 <= floormod((threadIdx.x_1 + 4), 9))) && (floormod((threadIdx.x_1 + 4), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 490), 9)*7)) + floormod((threadIdx.x_1 + 4), 9)) + 6)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 539)] = @tir.if_then_else((((floormod((threadIdx.x_1 + 35), 63) < 54) && (1 <= floormod((threadIdx.x_1 + 8), 9))) && (floormod((threadIdx.x_1 + 8), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 539), 9)*7)) + floormod((threadIdx.x_1 + 8), 9)) + 6)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 588)] = @tir.if_then_else((((floormod((threadIdx.x_1 + 21), 63) < 54) && (1 <= floormod((threadIdx.x_1 + 3), 9))) && (floormod((threadIdx.x_1 + 3), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 588), 9)*7)) + floormod((threadIdx.x_1 + 3), 9)) + 6)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 637)] = @tir.if_then_else((((floormod((threadIdx.x_1 + 7), 63) < 54) && (1 <= floormod((threadIdx.x_1 + 7), 9))) && (floormod((threadIdx.x_1 + 7), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 637), 9)*7)) + floormod((threadIdx.x_1 + 7), 9)) + 6)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 686)] = @tir.if_then_else((((floormod((threadIdx.x_1 + 56), 63) < 54) && (1 <= floormod((threadIdx.x_1 + 2), 9))) && (floormod((threadIdx.x_1 + 2), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 686), 9)*7)) + floormod((threadIdx.x_1 + 2), 9)) + 6)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 735)] = @tir.if_then_else((((floormod((threadIdx.x_1 + 42), 63) < 54) && (1 <= floormod((threadIdx.x_1 + 6), 9))) && (floormod((threadIdx.x_1 + 6), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 735), 9)*7)) + floormod((threadIdx.x_1 + 6), 9)) + 6)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 784)] = @tir.if_then_else((((floormod((threadIdx.x_1 + 28), 63) < 54) && (1 <= floormod((threadIdx.x_1 + 1), 9))) && (floormod((threadIdx.x_1 + 1), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 784), 9)*7)) + floormod((threadIdx.x_1 + 1), 9)) + 6)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 833)] = @tir.if_then_else((((floormod((threadIdx.x_1 + 14), 63) < 54) && (1 <= floormod((threadIdx.x_1 + 5), 9))) && (floormod((threadIdx.x_1 + 5), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 833), 9)*7)) + floormod((threadIdx.x_1 + 5), 9)) + 6)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 882)] = @tir.if_then_else(((1 <= floormod(threadIdx.x_1, 9)) && (floormod(threadIdx.x_1, 9) < 8)), data[(((cse_var_2 + (floordiv(threadIdx.x_1, 9)*7)) + floormod(threadIdx.x_1, 9)) + 692)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ pad_temp.shared_1[(threadIdx.x_1 + 931)] = @tir.if_then_else((((floormod((threadIdx.x_1 + 49), 63) < 54) && (1 <= floormod((threadIdx.x_1 + 4), 9))) && (floormod((threadIdx.x_1 + 4), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 931), 9)*7)) + floormod((threadIdx.x_1 + 4), 9)) + 6)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
+ if @tir.likely((threadIdx.x_1 < 28), dtype=bool) {
+ pad_temp.shared_1[(threadIdx.x_1 + 980)] = @tir.if_then_else((((floormod((threadIdx.x_1 + 35), 63) < 54) && (1 <= floormod((threadIdx.x_1 + 8), 9))) && (floormod((threadIdx.x_1 + 8), 9) < 8)), data[(((cse_var_2 + (floordiv((threadIdx.x_1 + 980), 9)*7)) + floormod((threadIdx.x_1 + 8), 9)) + 6)], 0f32, dtype=float32)
+ }
attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 98)] = kernel[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 98), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 26), 36), 9)*9)) + (floordiv(floormod((threadIdx.x_2 + 8), 9), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
+ kernel.shared_1[threadIdx.x_2] = kernel[((((((blockIdx.x*18432) + (floordiv(threadIdx.x_2, 48)*4608)) + cse_var_1) + (floordiv(floormod(threadIdx.x_2, 48), 3)*9)) + floormod(threadIdx.x_2, 3)) + 6)]
attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 147)] = kernel[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 147), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 3), 36), 9)*9)) + (floormod((floordiv(threadIdx.x_2, 3) + 1), 3)*3)) + floormod(threadIdx.x_2, 3))]
+ kernel.shared_1[(threadIdx.x_2 + 49)] = kernel[((((((blockIdx.x*18432) + (floordiv((threadIdx.x_2 + 49), 48)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 1), 48), 3)*9)) + floormod((threadIdx.x_2 + 1), 3)) + 6)]
attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 196)] = kernel[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 196), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 16), 36), 9)*9)) + (floordiv(floormod((threadIdx.x_2 + 7), 9), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
+ kernel.shared_1[(threadIdx.x_2 + 98)] = kernel[((((((blockIdx.x*18432) + (floordiv((threadIdx.x_2 + 98), 48)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 2), 48), 3)*9)) + floormod((threadIdx.x_2 + 2), 3)) + 6)]
attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- if @tir.likely((threadIdx.x_2 < 43), dtype=bool) {
- kernel.shared_1[(threadIdx.x_2 + 245)] = kernel[(((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 245), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 29), 36), 9)*9)) + floormod((threadIdx.x_2 + 2), 9))]
+ if @tir.likely((threadIdx.x_2 < 45), dtype=bool) {
+ kernel.shared_1[(threadIdx.x_2 + 147)] = kernel[((((((blockIdx.x*18432) + (floordiv((threadIdx.x_2 + 147), 48)*4608)) + cse_var_1) + (floordiv(threadIdx.x_2, 3)*9)) + floormod(threadIdx.x_2, 3)) + 15)]
}
conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[0]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[36]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[72]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[108]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[144]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[180]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[216]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[252]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[48]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[96]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[144]))
conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[1]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[37]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[73]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[109]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[145]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[181]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[217]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[253]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[49]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[97]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[145]))
conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[2]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[38]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[74]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[110]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[146]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[182]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[218]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[254]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 9)]*kernel.shared_1[3]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 9)]*kernel.shared_1[39]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 9)]*kernel.shared_1[75]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 9)]*kernel.shared_1[111]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 9)]*kernel.shared_1[147]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 9)]*kernel.shared_1[183]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 9)]*kernel.shared_1[219]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 9)]*kernel.shared_1[255]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 10)]*kernel.shared_1[4]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 10)]*kernel.shared_1[40]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 10)]*kernel.shared_1[76]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 10)]*kernel.shared_1[112]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 10)]*kernel.shared_1[148]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 10)]*kernel.shared_1[184]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 10)]*kernel.shared_1[220]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 10)]*kernel.shared_1[256]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 11)]*kernel.shared_1[5]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 11)]*kernel.shared_1[41]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 11)]*kernel.shared_1[77]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 11)]*kernel.shared_1[113]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 11)]*kernel.shared_1[149]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 11)]*kernel.shared_1[185]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 11)]*kernel.shared_1[221]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 11)]*kernel.shared_1[257]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[6]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[42]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[78]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[114]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[150]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[186]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[222]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[258]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[7]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[43]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[79]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[115]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[151]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[187]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[223]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[259]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[8]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[44]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[80]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[116]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[152]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[188]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[224]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[260]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 81)]*kernel.shared_1[9]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 81)]*kernel.shared_1[45]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 81)]*kernel.shared_1[81]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 81)]*kernel.shared_1[117]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 81)]*kernel.shared_1[153]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 81)]*kernel.shared_1[189]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 81)]*kernel.shared_1[225]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 81)]*kernel.shared_1[261]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 82)]*kernel.shared_1[10]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 82)]*kernel.shared_1[46]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 82)]*kernel.shared_1[82]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 82)]*kernel.shared_1[118]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 82)]*kernel.shared_1[154]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 82)]*kernel.shared_1[190]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 82)]*kernel.shared_1[226]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 82)]*kernel.shared_1[262]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 83)]*kernel.shared_1[11]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 83)]*kernel.shared_1[47]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 83)]*kernel.shared_1[83]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 83)]*kernel.shared_1[119]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 83)]*kernel.shared_1[155]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 83)]*kernel.shared_1[191]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 83)]*kernel.shared_1[227]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 83)]*kernel.shared_1[263]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 90)]*kernel.shared_1[12]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 90)]*kernel.shared_1[48]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 90)]*kernel.shared_1[84]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 90)]*kernel.shared_1[120]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 90)]*kernel.shared_1[156]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 90)]*kernel.shared_1[192]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 90)]*kernel.shared_1[228]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 90)]*kernel.shared_1[264]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 91)]*kernel.shared_1[13]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 91)]*kernel.shared_1[49]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 91)]*kernel.shared_1[85]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 91)]*kernel.shared_1[121]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 91)]*kernel.shared_1[157]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 91)]*kernel.shared_1[193]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 91)]*kernel.shared_1[229]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 91)]*kernel.shared_1[265]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 92)]*kernel.shared_1[14]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 92)]*kernel.shared_1[50]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 92)]*kernel.shared_1[86]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 92)]*kernel.shared_1[122]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 92)]*kernel.shared_1[158]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 92)]*kernel.shared_1[194]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 92)]*kernel.shared_1[230]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 92)]*kernel.shared_1[266]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[15]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[51]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[87]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[123]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[159]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[195]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[231]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[267]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[16]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[52]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[88]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[124]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[160]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[196]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[232]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[268]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[17]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[53]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[89]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[125]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[161]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[197]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[233]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[269]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 162)]*kernel.shared_1[18]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 162)]*kernel.shared_1[54]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 162)]*kernel.shared_1[90]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 162)]*kernel.shared_1[126]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 162)]*kernel.shared_1[162]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 162)]*kernel.shared_1[198]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 162)]*kernel.shared_1[234]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 162)]*kernel.shared_1[270]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 163)]*kernel.shared_1[19]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 163)]*kernel.shared_1[55]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 163)]*kernel.shared_1[91]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 163)]*kernel.shared_1[127]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 163)]*kernel.shared_1[163]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 163)]*kernel.shared_1[199]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 163)]*kernel.shared_1[235]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 163)]*kernel.shared_1[271]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 164)]*kernel.shared_1[20]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 164)]*kernel.shared_1[56]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 164)]*kernel.shared_1[92]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 164)]*kernel.shared_1[128]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 164)]*kernel.shared_1[164]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 164)]*kernel.shared_1[200]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 164)]*kernel.shared_1[236]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 164)]*kernel.shared_1[272]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 171)]*kernel.shared_1[21]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 171)]*kernel.shared_1[57]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 171)]*kernel.shared_1[93]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 171)]*kernel.shared_1[129]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 171)]*kernel.shared_1[165]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 171)]*kernel.shared_1[201]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 171)]*kernel.shared_1[237]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 171)]*kernel.shared_1[273]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 172)]*kernel.shared_1[22]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 172)]*kernel.shared_1[58]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 172)]*kernel.shared_1[94]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 172)]*kernel.shared_1[130]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 172)]*kernel.shared_1[166]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 172)]*kernel.shared_1[202]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 172)]*kernel.shared_1[238]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 172)]*kernel.shared_1[274]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 173)]*kernel.shared_1[23]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 173)]*kernel.shared_1[59]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 173)]*kernel.shared_1[95]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 173)]*kernel.shared_1[131]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 173)]*kernel.shared_1[167]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 173)]*kernel.shared_1[203]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 173)]*kernel.shared_1[239]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 173)]*kernel.shared_1[275]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[24]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[60]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[96]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[132]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[168]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[204]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[240]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[276]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[25]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[61]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[97]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[133]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[169]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[205]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[241]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[277]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[26]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[62]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[98]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[134]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[170]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[206]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[242]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[278]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 243)]*kernel.shared_1[27]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 243)]*kernel.shared_1[63]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 243)]*kernel.shared_1[99]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 243)]*kernel.shared_1[135]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 243)]*kernel.shared_1[171]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 243)]*kernel.shared_1[207]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 243)]*kernel.shared_1[243]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 243)]*kernel.shared_1[279]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 244)]*kernel.shared_1[28]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 244)]*kernel.shared_1[64]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 244)]*kernel.shared_1[100]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 244)]*kernel.shared_1[136]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 244)]*kernel.shared_1[172]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 244)]*kernel.shared_1[208]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 244)]*kernel.shared_1[244]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 244)]*kernel.shared_1[280]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 245)]*kernel.shared_1[29]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 245)]*kernel.shared_1[65]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 245)]*kernel.shared_1[101]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 245)]*kernel.shared_1[137]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 245)]*kernel.shared_1[173]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 245)]*kernel.shared_1[209]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 245)]*kernel.shared_1[245]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 245)]*kernel.shared_1[281]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[30]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[66]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[102]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[138]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[174]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[210]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[246]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[282]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[31]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[67]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[103]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[139]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[175]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[211]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[247]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[283]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[32]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[68]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[104]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[140]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[176]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[212]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[248]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[284]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[33]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[69]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[105]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[141]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[177]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[213]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[249]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[285]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[34]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[70]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[106]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[142]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[178]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[214]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[250]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[286]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[35]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[71]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[107]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[143]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[179]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[215]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[251]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[287]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[50]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[98]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[146]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[3]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[51]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[99]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[147]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[4]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[52]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[100]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[148]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[5]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[53]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[101]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[149]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[6]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[54]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[102]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[150]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[7]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[55]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[103]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[151]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[8]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[56]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[104]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[152]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[9]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[57]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[105]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[153]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[10]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[58]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[106]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[154]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[11]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[59]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[107]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[155]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[12]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[60]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[108]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[156]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[13]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[61]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[109]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[157]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[14]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[62]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[110]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[158]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 315)]*kernel.shared_1[15]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 315)]*kernel.shared_1[63]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 315)]*kernel.shared_1[111]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 315)]*kernel.shared_1[159]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 316)]*kernel.shared_1[16]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 316)]*kernel.shared_1[64]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 316)]*kernel.shared_1[112]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 316)]*kernel.shared_1[160]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 317)]*kernel.shared_1[17]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 317)]*kernel.shared_1[65]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 317)]*kernel.shared_1[113]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 317)]*kernel.shared_1[161]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 378)]*kernel.shared_1[18]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 378)]*kernel.shared_1[66]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 378)]*kernel.shared_1[114]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 378)]*kernel.shared_1[162]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 379)]*kernel.shared_1[19]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 379)]*kernel.shared_1[67]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 379)]*kernel.shared_1[115]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 379)]*kernel.shared_1[163]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 380)]*kernel.shared_1[20]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 380)]*kernel.shared_1[68]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 380)]*kernel.shared_1[116]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 380)]*kernel.shared_1[164]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 441)]*kernel.shared_1[21]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 441)]*kernel.shared_1[69]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 441)]*kernel.shared_1[117]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 441)]*kernel.shared_1[165]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 442)]*kernel.shared_1[22]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 442)]*kernel.shared_1[70]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 442)]*kernel.shared_1[118]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 442)]*kernel.shared_1[166]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 443)]*kernel.shared_1[23]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 443)]*kernel.shared_1[71]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 443)]*kernel.shared_1[119]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 443)]*kernel.shared_1[167]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 504)]*kernel.shared_1[24]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 504)]*kernel.shared_1[72]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 504)]*kernel.shared_1[120]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 504)]*kernel.shared_1[168]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 505)]*kernel.shared_1[25]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 505)]*kernel.shared_1[73]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 505)]*kernel.shared_1[121]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 505)]*kernel.shared_1[169]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 506)]*kernel.shared_1[26]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 506)]*kernel.shared_1[74]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 506)]*kernel.shared_1[122]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 506)]*kernel.shared_1[170]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 567)]*kernel.shared_1[27]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 567)]*kernel.shared_1[75]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 567)]*kernel.shared_1[123]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 567)]*kernel.shared_1[171]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 568)]*kernel.shared_1[28]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 568)]*kernel.shared_1[76]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 568)]*kernel.shared_1[124]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 568)]*kernel.shared_1[172]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 569)]*kernel.shared_1[29]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 569)]*kernel.shared_1[77]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 569)]*kernel.shared_1[125]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 569)]*kernel.shared_1[173]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 630)]*kernel.shared_1[30]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 630)]*kernel.shared_1[78]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 630)]*kernel.shared_1[126]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 630)]*kernel.shared_1[174]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 631)]*kernel.shared_1[31]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 631)]*kernel.shared_1[79]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 631)]*kernel.shared_1[127]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 631)]*kernel.shared_1[175]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 632)]*kernel.shared_1[32]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 632)]*kernel.shared_1[80]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 632)]*kernel.shared_1[128]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 632)]*kernel.shared_1[176]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 693)]*kernel.shared_1[33]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 693)]*kernel.shared_1[81]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 693)]*kernel.shared_1[129]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 693)]*kernel.shared_1[177]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 694)]*kernel.shared_1[34]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 694)]*kernel.shared_1[82]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 694)]*kernel.shared_1[130]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 694)]*kernel.shared_1[178]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 695)]*kernel.shared_1[35]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 695)]*kernel.shared_1[83]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 695)]*kernel.shared_1[131]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 695)]*kernel.shared_1[179]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 756)]*kernel.shared_1[36]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 756)]*kernel.shared_1[84]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 756)]*kernel.shared_1[132]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 756)]*kernel.shared_1[180]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 757)]*kernel.shared_1[37]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 757)]*kernel.shared_1[85]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 757)]*kernel.shared_1[133]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 757)]*kernel.shared_1[181]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 758)]*kernel.shared_1[38]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 758)]*kernel.shared_1[86]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 758)]*kernel.shared_1[134]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 758)]*kernel.shared_1[182]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 819)]*kernel.shared_1[39]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 819)]*kernel.shared_1[87]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 819)]*kernel.shared_1[135]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 819)]*kernel.shared_1[183]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 820)]*kernel.shared_1[40]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 820)]*kernel.shared_1[88]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 820)]*kernel.shared_1[136]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 820)]*kernel.shared_1[184]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 821)]*kernel.shared_1[41]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 821)]*kernel.shared_1[89]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 821)]*kernel.shared_1[137]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 821)]*kernel.shared_1[185]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 882)]*kernel.shared_1[42]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 882)]*kernel.shared_1[90]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 882)]*kernel.shared_1[138]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 882)]*kernel.shared_1[186]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 883)]*kernel.shared_1[43]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 883)]*kernel.shared_1[91]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 883)]*kernel.shared_1[139]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 883)]*kernel.shared_1[187]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 884)]*kernel.shared_1[44]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 884)]*kernel.shared_1[92]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 884)]*kernel.shared_1[140]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 884)]*kernel.shared_1[188]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 945)]*kernel.shared_1[45]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 945)]*kernel.shared_1[93]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 945)]*kernel.shared_1[141]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 945)]*kernel.shared_1[189]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 946)]*kernel.shared_1[46]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 946)]*kernel.shared_1[94]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 946)]*kernel.shared_1[142]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 946)]*kernel.shared_1[190]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 947)]*kernel.shared_1[47]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 947)]*kernel.shared_1[95]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 947)]*kernel.shared_1[143]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((floordiv(threadIdx.x, 7)*9) + floormod(threadIdx.x, 7)) + 947)]*kernel.shared_1[191]))
}
}
- compute[((blockIdx.x*392) + threadIdx.x)] = max((conv2d_nchw_1[0] + bias[(blockIdx.x*8)]), 0f32)
- compute[(((blockIdx.x*392) + threadIdx.x) + 49)] = max((conv2d_nchw_1[1] + bias[((blockIdx.x*8) + 1)]), 0f32)
- compute[(((blockIdx.x*392) + threadIdx.x) + 98)] = max((conv2d_nchw_1[2] + bias[((blockIdx.x*8) + 2)]), 0f32)
- compute[(((blockIdx.x*392) + threadIdx.x) + 147)] = max((conv2d_nchw_1[3] + bias[((blockIdx.x*8) + 3)]), 0f32)
- compute[(((blockIdx.x*392) + threadIdx.x) + 196)] = max((conv2d_nchw_1[4] + bias[((blockIdx.x*8) + 4)]), 0f32)
- compute[(((blockIdx.x*392) + threadIdx.x) + 245)] = max((conv2d_nchw_1[5] + bias[((blockIdx.x*8) + 5)]), 0f32)
- compute[(((blockIdx.x*392) + threadIdx.x) + 294)] = max((conv2d_nchw_1[6] + bias[((blockIdx.x*8) + 6)]), 0f32)
- compute[(((blockIdx.x*392) + threadIdx.x) + 343)] = max((conv2d_nchw_1[7] + bias[((blockIdx.x*8) + 7)]), 0f32)
+ for (i1.inner: int32, 0, 4) {
+ compute[(((blockIdx.x*196) + (i1.inner*49)) + threadIdx.x)] = max((conv2d_nchw_1[i1.inner] + bias[((blockIdx.x*4) + i1.inner)]), 0f32)
+ }
}
}
</pre></div>
@@ -870,7 +1277,7 @@ cooperative fetching, unrolling and operator fusion.</p>
<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: 0.223 ms
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Execution time of this operator: 0.268 ms
</pre></div>
</div>
</div>
@@ -899,10 +1306,10 @@ conv2d_nchw_nn_o_i, conv2d_nchw_nn_i = s[conv2d_nchw].split(conv2d_nchw_nn, fact
conv2d_nchw_nn_o_o_i, conv2d_nchw_nn_o_i = s[conv2d_nchw].split(conv2d_nchw_nn_o_i, factor=1)
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_i, conv2d_nchw_ff_i = s[conv2d_nchw].split(conv2d_nchw_ff, factor=4)
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=1)
-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=8)
+conv2d_nchw_ff_o_o_o_o, conv2d_nchw_ff_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_o_o_i, factor=1)
conv2d_nchw_yy_o_i, conv2d_nchw_yy_i = s[conv2d_nchw].split(conv2d_nchw_yy, factor=1)
conv2d_nchw_yy_o_o_i, conv2d_nchw_yy_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_i, factor=1)
conv2d_nchw_yy_o_o_o_i, conv2d_nchw_yy_o_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_o_i, factor=7)
@@ -911,9 +1318,9 @@ conv2d_nchw_xx_o_i, conv2d_nchw_xx_i = s[conv2d_nchw].split(conv2d_nchw_xx, fact
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_i, conv2d_nchw_rc_i = s[conv2d_nchw].split(conv2d_nchw_rc, factor=16)
conv2d_nchw_rc_o_o, conv2d_nchw_rc_o_i = s[conv2d_nchw].split(conv2d_nchw_rc_o_i, factor=1)
-conv2d_nchw_ry_o_i, conv2d_nchw_ry_i = s[conv2d_nchw].split(conv2d_nchw_ry, factor=3)
+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=3)
conv2d_nchw_rx_o_o, conv2d_nchw_rx_o_i = s[conv2d_nchw].split(conv2d_nchw_rx_o_i, factor=1)
@@ -921,9 +1328,9 @@ s[conv2d_nchw].reorder(conv2d_nchw_nn_o_o_o_o, conv2d_nchw_ff_o_o_o_o, 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_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=1)
-compute_i1_o_o_o, compute_i1_o_o_i = s[compute].split(compute_i1_o_o_i, factor=8)
+compute_i1_o_o_o, compute_i1_o_o_i = s[compute].split(compute_i1_o_o_i, factor=1)
compute_i2_o_i, compute_i2_i = s[compute].split(compute_i2, factor=1)
compute_i2_o_o_i, compute_i2_o_i = s[compute].split(compute_i2_o_i, factor=7)
compute_i2_o_o_o, compute_i2_o_o_i = s[compute].split(compute_i2_o_o_i, factor=1)
@@ -951,11 +1358,11 @@ 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=49)
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=2)
+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=49)
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", 512)
+s[conv2d_nchw].pragma(conv2d_nchw_nn_o_o_o_o, "auto_unroll_max_step", 1024)
s[conv2d_nchw].pragma(conv2d_nchw_nn_o_o_o_o, "unroll_explicit", True)
CUDA source code:
@@ -974,337 +1381,687 @@ CUDA source code:
#define uint64_t unsigned long long
#endif
extern "C" __global__ void __launch_bounds__(49) default_function_kernel0(float* __restrict__ data, float* __restrict__ kernel, float* __restrict__ compute, float* __restrict__ bias) {
- float conv2d_nchw[8];
- __shared__ float pad_temp_shared[324];
- __shared__ float kernel_shared[288];
+ float conv2d_nchw[4];
+ __shared__ float pad_temp_shared[1008];
+ __shared__ float kernel_shared[192];
conv2d_nchw[0] = 0.000000e+00f;
conv2d_nchw[1] = 0.000000e+00f;
conv2d_nchw[2] = 0.000000e+00f;
conv2d_nchw[3] = 0.000000e+00f;
- conv2d_nchw[4] = 0.000000e+00f;
- conv2d_nchw[5] = 0.000000e+00f;
- conv2d_nchw[6] = 0.000000e+00f;
- conv2d_nchw[7] = 0.000000e+00f;
- for (int rc_outer_outer = 0; rc_outer_outer < 128; ++rc_outer_outer) {
+ for (int rc_outer_outer = 0; rc_outer_outer < 32; ++rc_outer_outer) {
+ __syncthreads();
+ pad_temp_shared[((int)threadIdx.x)] = ((((9 <= ((int)threadIdx.x)) && (1 <= (((int)threadIdx.x) % 9))) && ((((int)threadIdx.x) % 9) < 8)) ? data[((((rc_outer_outer * 784) + ((((int)threadIdx.x) / 9) * 7)) + (((int)threadIdx.x) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 49)] = ((((9 <= ((((int)threadIdx.x) + 49) % 63)) && (1 <= ((((int)threadIdx.x) + 4) % 9))) && (((((int)threadIdx.x) + 4) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 49) / 9) * 7)) + ((((int)threadIdx.x) + 4) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 98)] = ((((9 <= ((((int)threadIdx.x) + 35) % 63)) && (1 <= ((((int)threadIdx.x) + 8) % 9))) && (((((int)threadIdx.x) + 8) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 98) / 9) * 7)) + ((((int)threadIdx.x) + 8) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 147)] = ((((9 <= ((((int)threadIdx.x) + 21) % 63)) && (1 <= ((((int)threadIdx.x) + 3) % 9))) && (((((int)threadIdx.x) + 3) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 147) / 9) * 7)) + ((((int)threadIdx.x) + 3) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 196)] = ((((2 <= ((int)threadIdx.x)) && (1 <= ((((int)threadIdx.x) + 7) % 9))) && (((((int)threadIdx.x) + 7) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 196) / 9) * 7)) + ((((int)threadIdx.x) + 7) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 245)] = ((((9 <= ((((int)threadIdx.x) + 56) % 63)) && (1 <= ((((int)threadIdx.x) + 2) % 9))) && (((((int)threadIdx.x) + 2) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 245) / 9) * 7)) + ((((int)threadIdx.x) + 2) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 294)] = ((((9 <= ((((int)threadIdx.x) + 42) % 63)) && (1 <= ((((int)threadIdx.x) + 6) % 9))) && (((((int)threadIdx.x) + 6) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 294) / 9) * 7)) + ((((int)threadIdx.x) + 6) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 343)] = ((((9 <= ((((int)threadIdx.x) + 28) % 63)) && (1 <= ((((int)threadIdx.x) + 1) % 9))) && (((((int)threadIdx.x) + 1) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 343) / 9) * 7)) + ((((int)threadIdx.x) + 1) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 392)] = (((1 <= ((((int)threadIdx.x) + 5) % 9)) && (((((int)threadIdx.x) + 5) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 392) / 9) * 7)) + ((((int)threadIdx.x) + 5) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 441)] = ((((9 <= ((int)threadIdx.x)) && (1 <= (((int)threadIdx.x) % 9))) && ((((int)threadIdx.x) % 9) < 8)) ? data[((((rc_outer_outer * 784) + ((((int)threadIdx.x) / 9) * 7)) + (((int)threadIdx.x) % 9)) + 335)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 490)] = ((((9 <= ((((int)threadIdx.x) + 49) % 63)) && (1 <= ((((int)threadIdx.x) + 4) % 9))) && (((((int)threadIdx.x) + 4) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 490) / 9) * 7)) + ((((int)threadIdx.x) + 4) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 539)] = ((((9 <= ((((int)threadIdx.x) + 35) % 63)) && (1 <= ((((int)threadIdx.x) + 8) % 9))) && (((((int)threadIdx.x) + 8) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 539) / 9) * 7)) + ((((int)threadIdx.x) + 8) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 588)] = ((((9 <= ((((int)threadIdx.x) + 21) % 63)) && (1 <= ((((int)threadIdx.x) + 3) % 9))) && (((((int)threadIdx.x) + 3) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 588) / 9) * 7)) + ((((int)threadIdx.x) + 3) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 637)] = ((((2 <= ((int)threadIdx.x)) && (1 <= ((((int)threadIdx.x) + 7) % 9))) && (((((int)threadIdx.x) + 7) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 637) / 9) * 7)) + ((((int)threadIdx.x) + 7) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 686)] = ((((9 <= ((((int)threadIdx.x) + 56) % 63)) && (1 <= ((((int)threadIdx.x) + 2) % 9))) && (((((int)threadIdx.x) + 2) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 686) / 9) * 7)) + ((((int)threadIdx.x) + 2) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 735)] = ((((9 <= ((((int)threadIdx.x) + 42) % 63)) && (1 <= ((((int)threadIdx.x) + 6) % 9))) && (((((int)threadIdx.x) + 6) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 735) / 9) * 7)) + ((((int)threadIdx.x) + 6) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 784)] = ((((9 <= ((((int)threadIdx.x) + 28) % 63)) && (1 <= ((((int)threadIdx.x) + 1) % 9))) && (((((int)threadIdx.x) + 1) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 784) / 9) * 7)) + ((((int)threadIdx.x) + 1) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 833)] = (((1 <= ((((int)threadIdx.x) + 5) % 9)) && (((((int)threadIdx.x) + 5) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 833) / 9) * 7)) + ((((int)threadIdx.x) + 5) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 882)] = ((((9 <= ((int)threadIdx.x)) && (1 <= (((int)threadIdx.x) % 9))) && ((((int)threadIdx.x) % 9) < 8)) ? data[((((rc_outer_outer * 784) + ((((int)threadIdx.x) / 9) * 7)) + (((int)threadIdx.x) % 9)) + 678)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 931)] = ((((9 <= ((((int)threadIdx.x) + 49) % 63)) && (1 <= ((((int)threadIdx.x) + 4) % 9))) && (((((int)threadIdx.x) + 4) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 931) / 9) * 7)) + ((((int)threadIdx.x) + 4) % 9)) - 8)] : 0.000000e+00f);
+ if (((int)threadIdx.x) < 28) {
+ pad_temp_shared[(((int)threadIdx.x) + 980)] = (((1 <= ((((int)threadIdx.x) + 8) % 9)) && (((((int)threadIdx.x) + 8) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 980) / 9) * 7)) + ((((int)threadIdx.x) + 8) % 9)) - 8)] : 0.000000e+00f);
+ }
+ kernel_shared[((int)threadIdx.x)] = kernel[(((((((int)blockIdx.x) * 18432) + ((((int)threadIdx.x) / 48) * 4608)) + (rc_outer_outer * 144)) + (((((int)threadIdx.x) % 48) / 3) * 9)) + (((int)threadIdx.x) % 3))];
+ kernel_shared[(((int)threadIdx.x) + 49)] = kernel[(((((((int)blockIdx.x) * 18432) + (((((int)threadIdx.x) + 49) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 1) % 48) / 3) * 9)) + ((((int)threadIdx.x) + 1) % 3))];
+ kernel_shared[(((int)threadIdx.x) + 98)] = kernel[(((((((int)blockIdx.x) * 18432) + (((((int)threadIdx.x) + 98) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 2) % 48) / 3) * 9)) + ((((int)threadIdx.x) + 2) % 3))];
+ if (((int)threadIdx.x) < 45) {
+ kernel_shared[(((int)threadIdx.x) + 147)] = kernel[((((((((int)blockIdx.x) * 18432) + (((((int)threadIdx.x) + 147) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((int)threadIdx.x) / 3) * 9)) + (((int)threadIdx.x) % 3)) + 9)];
+ }
+ __syncthreads();
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[0]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[48]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[96]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[144]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[1]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[49]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[97]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[145]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[2]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[50]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[98]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[146]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[3]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[51]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[99]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[147]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[4]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[52]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[100]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[148]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[5]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[53]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[101]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[149]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[6]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[54]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[102]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[150]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[7]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[55]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[103]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[151]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[8]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[56]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[104]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[152]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[9]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[57]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[105]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[153]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[10]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[58]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[106]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[154]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[11]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[59]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[107]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[155]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[12]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[60]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[108]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[156]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[13]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[61]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[109]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[157]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[14]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[62]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[110]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[158]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[15]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[63]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[111]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[159]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 316)] * kernel_shared[16]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 316)] * kernel_shared[64]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 316)] * kernel_shared[112]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 316)] * kernel_shared[160]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 317)] * kernel_shared[17]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 317)] * kernel_shared[65]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 317)] * kernel_shared[113]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 317)] * kernel_shared[161]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[18]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[66]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[114]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[162]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 379)] * kernel_shared[19]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 379)] * kernel_shared[67]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 379)] * kernel_shared[115]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 379)] * kernel_shared[163]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 380)] * kernel_shared[20]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 380)] * kernel_shared[68]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 380)] * kernel_shared[116]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 380)] * kernel_shared[164]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[21]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[69]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[117]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[165]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 442)] * kernel_shared[22]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 442)] * kernel_shared[70]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 442)] * kernel_shared[118]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 442)] * kernel_shared[166]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 443)] * kernel_shared[23]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 443)] * kernel_shared[71]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 443)] * kernel_shared[119]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 443)] * kernel_shared[167]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 504)] * kernel_shared[24]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 504)] * kernel_shared[72]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 504)] * kernel_shared[120]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 504)] * kernel_shared[168]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 505)] * kernel_shared[25]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 505)] * kernel_shared[73]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 505)] * kernel_shared[121]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 505)] * kernel_shared[169]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 506)] * kernel_shared[26]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 506)] * kernel_shared[74]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 506)] * kernel_shared[122]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 506)] * kernel_shared[170]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 567)] * kernel_shared[27]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 567)] * kernel_shared[75]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 567)] * kernel_shared[123]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 567)] * kernel_shared[171]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 568)] * kernel_shared[28]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 568)] * kernel_shared[76]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 568)] * kernel_shared[124]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 568)] * kernel_shared[172]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 569)] * kernel_shared[29]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 569)] * kernel_shared[77]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 569)] * kernel_shared[125]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 569)] * kernel_shared[173]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 630)] * kernel_shared[30]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 630)] * kernel_shared[78]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 630)] * kernel_shared[126]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 630)] * kernel_shared[174]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 631)] * kernel_shared[31]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 631)] * kernel_shared[79]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 631)] * kernel_shared[127]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 631)] * kernel_shared[175]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 632)] * kernel_shared[32]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 632)] * kernel_shared[80]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 632)] * kernel_shared[128]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 632)] * kernel_shared[176]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 693)] * kernel_shared[33]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 693)] * kernel_shared[81]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 693)] * kernel_shared[129]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 693)] * kernel_shared[177]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 694)] * kernel_shared[34]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 694)] * kernel_shared[82]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 694)] * kernel_shared[130]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 694)] * kernel_shared[178]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 695)] * kernel_shared[35]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 695)] * kernel_shared[83]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 695)] * kernel_shared[131]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 695)] * kernel_shared[179]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 756)] * kernel_shared[36]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 756)] * kernel_shared[84]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 756)] * kernel_shared[132]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 756)] * kernel_shared[180]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 757)] * kernel_shared[37]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 757)] * kernel_shared[85]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 757)] * kernel_shared[133]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 757)] * kernel_shared[181]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 758)] * kernel_shared[38]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 758)] * kernel_shared[86]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 758)] * kernel_shared[134]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 758)] * kernel_shared[182]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 819)] * kernel_shared[39]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 819)] * kernel_shared[87]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 819)] * kernel_shared[135]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 819)] * kernel_shared[183]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 820)] * kernel_shared[40]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 820)] * kernel_shared[88]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 820)] * kernel_shared[136]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 820)] * kernel_shared[184]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 821)] * kernel_shared[41]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 821)] * kernel_shared[89]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 821)] * kernel_shared[137]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 821)] * kernel_shared[185]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 882)] * kernel_shared[42]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 882)] * kernel_shared[90]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 882)] * kernel_shared[138]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 882)] * kernel_shared[186]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 883)] * kernel_shared[43]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 883)] * kernel_shared[91]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 883)] * kernel_shared[139]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 883)] * kernel_shared[187]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 884)] * kernel_shared[44]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 884)] * kernel_shared[92]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 884)] * kernel_shared[140]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 884)] * kernel_shared[188]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 945)] * kernel_shared[45]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 945)] * kernel_shared[93]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 945)] * kernel_shared[141]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 945)] * kernel_shared[189]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 946)] * kernel_shared[46]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 946)] * kernel_shared[94]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 946)] * kernel_shared[142]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 946)] * kernel_shared[190]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 947)] * kernel_shared[47]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 947)] * kernel_shared[95]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 947)] * kernel_shared[143]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 947)] * kernel_shared[191]));
__syncthreads();
- pad_temp_shared[(((int)threadIdx.x) * 2)] = (((((9 <= ((((int)threadIdx.x) * 2) % 81)) && (((((int)threadIdx.x) * 2) % 81) < 72)) && (1 <= ((((int)threadIdx.x) * 2) % 9))) && (((((int)threadIdx.x) * 2) % 9) < 8)) ? data[(((((rc_outer_outer * 196) + (((((int)threadIdx.x) * 2) / 81) * 49)) + ((((((int)threadIdx.x) * 2) % 81) / 9) * 7)) + ((((int)threadIdx.x) * 2) % 9)) - 8)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 2) + 1)] = (((((9 <= (((((int)threadIdx.x) * 2) + 1) % 81)) && ((((((int)threadIdx.x) * 2) + 1) % 81) < 72)) && (1 <= (((((int)threadIdx.x) * 2) + 1) % 9))) && ((((((int)threadIdx.x) * 2) + 1) % 9) < 8)) ? data[(((((rc_outer_outer * 196) + ((((((int)threadIdx.x) * 2) + 1) / 81) * 49)) + (((((((int)threadIdx.x) * 2) + 1) % 81) / 9) * 7)) + (((((int)threadIdx.x) * 2) + 1) % 9)) - 8)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 2) + 98)] = (((((9 <= (((((int)threadIdx.x) * 2) + 17) % 81)) && ((((((int)threadIdx.x) * 2) + 17) % 81) < 72)) && (1 <= (((((int)threadIdx.x) * 2) + 8) % 9))) && ((((((int)threadIdx.x) * 2) + 8) % 9) < 8)) ? data[(((((rc_outer_outer * 196) + ((((((int)threadIdx.x) * 2) + 98) / 81) * 49)) + (((((((int)threadIdx.x) * 2) + 17) % 81) / 9) * 7)) + (((((int)threadIdx.x) * 2) + 8) % 9)) - 8)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 2) + 99)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 9) + 2) % 9)) && ((((((int)threadIdx.x) * 2) + 18) % 81) < 72)) && (1 <= ((((int)threadIdx.x) * 2) % 9))) && (((((int)threadIdx.x) * 2) % 9) < 8)) ? data[(((((rc_outer_outer * 196) + ((((((int)threadIdx.x) * 2) + 99) / 81) * 49)) + (((((((int)threadIdx.x) * 2) / 9) + 2) % 9) * 7)) + ((((int)threadIdx.x) * 2) % 9)) - 8)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 2) + 196)] = (((((9 <= (((((int)threadIdx.x) * 2) + 34) % 81)) && ((((((int)threadIdx.x) * 2) + 34) % 81) < 72)) && (1 <= (((((int)threadIdx.x) * 2) + 7) % 9))) && ((((((int)threadIdx.x) * 2) + 7) % 9) < 8)) ? data[(((((rc_outer_outer * 196) + ((((((int)threadIdx.x) * 2) + 196) / 81) * 49)) + (((((((int)threadIdx.x) * 2) + 34) % 81) / 9) * 7)) + (((((int)threadIdx.x) * 2) + 7) % 9)) - 8)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 2) + 197)] = (((((9 <= (((((int)threadIdx.x) * 2) + 35) % 81)) && ((((((int)threadIdx.x) * 2) + 35) % 81) < 72)) && (1 <= (((((int)threadIdx.x) * 2) + 8) % 9))) && ((((((int)threadIdx.x) * 2) + 8) % 9) < 8)) ? data[(((((rc_outer_outer * 196) + ((((((int)threadIdx.x) * 2) + 197) / 81) * 49)) + (((((((int)threadIdx.x) * 2) + 35) % 81) / 9) * 7)) + (((((int)threadIdx.x) * 2) + 8) % 9)) - 8)] : 0.000000e+00f);
- if (((int)threadIdx.x) < 15) {
- pad_temp_shared[((((int)threadIdx.x) * 2) + 294)] = ((((((int)threadIdx.x) < 11) && (1 <= (((((int)threadIdx.x) * 2) + 6) % 9))) && ((((((int)threadIdx.x) * 2) + 6) % 9) < 8)) ? data[((((rc_outer_outer * 196) + ((((((int)threadIdx.x) * 2) + 51) / 9) * 7)) + (((((int)threadIdx.x) * 2) + 6) % 9)) + 139)] : 0.000000e+00f);
+ pad_temp_shared[((int)threadIdx.x)] = (((1 <= (((int)threadIdx.x) % 9)) && ((((int)threadIdx.x) % 9) < 8)) ? data[((((rc_outer_outer * 784) + ((((int)threadIdx.x) / 9) * 7)) + (((int)threadIdx.x) % 9)) - 1)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 49)] = (((1 <= ((((int)threadIdx.x) + 4) % 9)) && (((((int)threadIdx.x) + 4) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 49) / 9) * 7)) + ((((int)threadIdx.x) + 4) % 9)) - 1)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 98)] = (((1 <= ((((int)threadIdx.x) + 8) % 9)) && (((((int)threadIdx.x) + 8) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 98) / 9) * 7)) + ((((int)threadIdx.x) + 8) % 9)) - 1)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 147)] = (((1 <= ((((int)threadIdx.x) + 3) % 9)) && (((((int)threadIdx.x) + 3) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 147) / 9) * 7)) + ((((int)threadIdx.x) + 3) % 9)) - 1)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 196)] = (((1 <= ((((int)threadIdx.x) + 7) % 9)) && (((((int)threadIdx.x) + 7) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 196) / 9) * 7)) + ((((int)threadIdx.x) + 7) % 9)) - 1)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 245)] = (((1 <= ((((int)threadIdx.x) + 2) % 9)) && (((((int)threadIdx.x) + 2) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 245) / 9) * 7)) + ((((int)threadIdx.x) + 2) % 9)) - 1)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 294)] = (((1 <= ((((int)threadIdx.x) + 6) % 9)) && (((((int)threadIdx.x) + 6) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 294) / 9) * 7)) + ((((int)threadIdx.x) + 6) % 9)) - 1)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 343)] = (((1 <= ((((int)threadIdx.x) + 1) % 9)) && (((((int)threadIdx.x) + 1) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 343) / 9) * 7)) + ((((int)threadIdx.x) + 1) % 9)) - 1)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 392)] = (((1 <= ((((int)threadIdx.x) + 5) % 9)) && (((((int)threadIdx.x) + 5) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 392) / 9) * 7)) + ((((int)threadIdx.x) + 5) % 9)) - 1)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 441)] = (((1 <= (((int)threadIdx.x) % 9)) && ((((int)threadIdx.x) % 9) < 8)) ? data[((((rc_outer_outer * 784) + ((((int)threadIdx.x) / 9) * 7)) + (((int)threadIdx.x) % 9)) + 342)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 490)] = (((1 <= ((((int)threadIdx.x) + 4) % 9)) && (((((int)threadIdx.x) + 4) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 490) / 9) * 7)) + ((((int)threadIdx.x) + 4) % 9)) - 1)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 539)] = (((1 <= ((((int)threadIdx.x) + 8) % 9)) && (((((int)threadIdx.x) + 8) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 539) / 9) * 7)) + ((((int)threadIdx.x) + 8) % 9)) - 1)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 588)] = (((1 <= ((((int)threadIdx.x) + 3) % 9)) && (((((int)threadIdx.x) + 3) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 588) / 9) * 7)) + ((((int)threadIdx.x) + 3) % 9)) - 1)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 637)] = (((1 <= ((((int)threadIdx.x) + 7) % 9)) && (((((int)threadIdx.x) + 7) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 637) / 9) * 7)) + ((((int)threadIdx.x) + 7) % 9)) - 1)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 686)] = (((1 <= ((((int)threadIdx.x) + 2) % 9)) && (((((int)threadIdx.x) + 2) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 686) / 9) * 7)) + ((((int)threadIdx.x) + 2) % 9)) - 1)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 735)] = (((1 <= ((((int)threadIdx.x) + 6) % 9)) && (((((int)threadIdx.x) + 6) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 735) / 9) * 7)) + ((((int)threadIdx.x) + 6) % 9)) - 1)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 784)] = (((1 <= ((((int)threadIdx.x) + 1) % 9)) && (((((int)threadIdx.x) + 1) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 784) / 9) * 7)) + ((((int)threadIdx.x) + 1) % 9)) - 1)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 833)] = (((1 <= ((((int)threadIdx.x) + 5) % 9)) && (((((int)threadIdx.x) + 5) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 833) / 9) * 7)) + ((((int)threadIdx.x) + 5) % 9)) - 1)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 882)] = (((1 <= (((int)threadIdx.x) % 9)) && ((((int)threadIdx.x) % 9) < 8)) ? data[((((rc_outer_outer * 784) + ((((int)threadIdx.x) / 9) * 7)) + (((int)threadIdx.x) % 9)) + 685)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 931)] = (((1 <= ((((int)threadIdx.x) + 4) % 9)) && (((((int)threadIdx.x) + 4) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 931) / 9) * 7)) + ((((int)threadIdx.x) + 4) % 9)) - 1)] : 0.000000e+00f);
+ if (((int)threadIdx.x) < 28) {
+ pad_temp_shared[(((int)threadIdx.x) + 980)] = (((1 <= ((((int)threadIdx.x) + 8) % 9)) && (((((int)threadIdx.x) + 8) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 980) / 9) * 7)) + ((((int)threadIdx.x) + 8) % 9)) - 1)] : 0.000000e+00f);
+ }
+ kernel_shared[((int)threadIdx.x)] = kernel[((((((((int)blockIdx.x) * 18432) + ((((int)threadIdx.x) / 48) * 4608)) + (rc_outer_outer * 144)) + (((((int)threadIdx.x) % 48) / 3) * 9)) + (((int)threadIdx.x) % 3)) + 3)];
+ kernel_shared[(((int)threadIdx.x) + 49)] = kernel[((((((((int)blockIdx.x) * 18432) + (((((int)threadIdx.x) + 49) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 1) % 48) / 3) * 9)) + ((((int)threadIdx.x) + 1) % 3)) + 3)];
+ kernel_shared[(((int)threadIdx.x) + 98)] = kernel[((((((((int)blockIdx.x) * 18432) + (((((int)threadIdx.x) + 98) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 2) % 48) / 3) * 9)) + ((((int)threadIdx.x) + 2) % 3)) + 3)];
+ if (((int)threadIdx.x) < 45) {
+ kernel_shared[(((int)threadIdx.x) + 147)] = kernel[((((((((int)blockIdx.x) * 18432) + (((((int)threadIdx.x) + 147) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((int)threadIdx.x) / 3) * 9)) + (((int)threadIdx.x) % 3)) + 12)];
}
- if (((int)threadIdx.x) < 15) {
- pad_temp_shared[((((int)threadIdx.x) * 2) + 295)] = ((((((int)threadIdx.x) < 10) && (1 <= (((((int)threadIdx.x) * 2) + 7) % 9))) && ((((((int)threadIdx.x) * 2) + 7) % 9) < 8)) ? data[((((rc_outer_outer * 196) + ((((((int)threadIdx.x) * 2) + 52) / 9) * 7)) + (((((int)threadIdx.x) * 2) + 7) % 9)) + 139)] : 0.000000e+00f);
+ __syncthreads();
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[0]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[48]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[96]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[144]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[1]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[49]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[97]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[145]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[2]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[50]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[98]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[146]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[3]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[51]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[99]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[147]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[4]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[52]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[100]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[148]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[5]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[53]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[101]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[149]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[6]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[54]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[102]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[150]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[7]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[55]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[103]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[151]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[8]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[56]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[104]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[152]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[9]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[57]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[105]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[153]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[10]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[58]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[106]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[154]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[11]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[59]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[107]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[155]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[12]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[60]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[108]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[156]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[13]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[61]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[109]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[157]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[14]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[62]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[110]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[158]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[15]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[63]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[111]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[159]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 316)] * kernel_shared[16]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 316)] * kernel_shared[64]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 316)] * kernel_shared[112]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 316)] * kernel_shared[160]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 317)] * kernel_shared[17]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 317)] * kernel_shared[65]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 317)] * kernel_shared[113]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 317)] * kernel_shared[161]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[18]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[66]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[114]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[162]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 379)] * kernel_shared[19]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 379)] * kernel_shared[67]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 379)] * kernel_shared[115]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 379)] * kernel_shared[163]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 380)] * kernel_shared[20]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 380)] * kernel_shared[68]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 380)] * kernel_shared[116]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 380)] * kernel_shared[164]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[21]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[69]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[117]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[165]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 442)] * kernel_shared[22]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 442)] * kernel_shared[70]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 442)] * kernel_shared[118]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 442)] * kernel_shared[166]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 443)] * kernel_shared[23]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 443)] * kernel_shared[71]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 443)] * kernel_shared[119]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 443)] * kernel_shared[167]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 504)] * kernel_shared[24]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 504)] * kernel_shared[72]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 504)] * kernel_shared[120]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 504)] * kernel_shared[168]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 505)] * kernel_shared[25]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 505)] * kernel_shared[73]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 505)] * kernel_shared[121]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 505)] * kernel_shared[169]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 506)] * kernel_shared[26]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 506)] * kernel_shared[74]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 506)] * kernel_shared[122]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 506)] * kernel_shared[170]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 567)] * kernel_shared[27]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 567)] * kernel_shared[75]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 567)] * kernel_shared[123]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 567)] * kernel_shared[171]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 568)] * kernel_shared[28]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 568)] * kernel_shared[76]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 568)] * kernel_shared[124]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 568)] * kernel_shared[172]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 569)] * kernel_shared[29]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 569)] * kernel_shared[77]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 569)] * kernel_shared[125]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 569)] * kernel_shared[173]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 630)] * kernel_shared[30]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 630)] * kernel_shared[78]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 630)] * kernel_shared[126]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 630)] * kernel_shared[174]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 631)] * kernel_shared[31]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 631)] * kernel_shared[79]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 631)] * kernel_shared[127]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 631)] * kernel_shared[175]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 632)] * kernel_shared[32]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 632)] * kernel_shared[80]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 632)] * kernel_shared[128]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 632)] * kernel_shared[176]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 693)] * kernel_shared[33]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 693)] * kernel_shared[81]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 693)] * kernel_shared[129]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 693)] * kernel_shared[177]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 694)] * kernel_shared[34]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 694)] * kernel_shared[82]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 694)] * kernel_shared[130]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 694)] * kernel_shared[178]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 695)] * kernel_shared[35]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 695)] * kernel_shared[83]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 695)] * kernel_shared[131]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 695)] * kernel_shared[179]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 756)] * kernel_shared[36]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 756)] * kernel_shared[84]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 756)] * kernel_shared[132]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 756)] * kernel_shared[180]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 757)] * kernel_shared[37]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 757)] * kernel_shared[85]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 757)] * kernel_shared[133]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 757)] * kernel_shared[181]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 758)] * kernel_shared[38]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 758)] * kernel_shared[86]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 758)] * kernel_shared[134]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 758)] * kernel_shared[182]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 819)] * kernel_shared[39]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 819)] * kernel_shared[87]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 819)] * kernel_shared[135]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 819)] * kernel_shared[183]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 820)] * kernel_shared[40]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 820)] * kernel_shared[88]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 820)] * kernel_shared[136]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 820)] * kernel_shared[184]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 821)] * kernel_shared[41]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 821)] * kernel_shared[89]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 821)] * kernel_shared[137]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 821)] * kernel_shared[185]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 882)] * kernel_shared[42]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 882)] * kernel_shared[90]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 882)] * kernel_shared[138]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 882)] * kernel_shared[186]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 883)] * kernel_shared[43]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 883)] * kernel_shared[91]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 883)] * kernel_shared[139]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 883)] * kernel_shared[187]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 884)] * kernel_shared[44]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 884)] * kernel_shared[92]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 884)] * kernel_shared[140]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 884)] * kernel_shared[188]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 945)] * kernel_shared[45]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 945)] * kernel_shared[93]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 945)] * kernel_shared[141]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 945)] * kernel_shared[189]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 946)] * kernel_shared[46]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 946)] * kernel_shared[94]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 946)] * kernel_shared[142]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 946)] * kernel_shared[190]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 947)] * kernel_shared[47]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 947)] * kernel_shared[95]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 947)] * kernel_shared[143]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 947)] * kernel_shared[191]));
+ __syncthreads();
+ pad_temp_shared[((int)threadIdx.x)] = (((1 <= (((int)threadIdx.x) % 9)) && ((((int)threadIdx.x) % 9) < 8)) ? data[((((rc_outer_outer * 784) + ((((int)threadIdx.x) / 9) * 7)) + (((int)threadIdx.x) % 9)) + 6)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 49)] = ((((((((int)threadIdx.x) + 49) % 63) < 54) && (1 <= ((((int)threadIdx.x) + 4) % 9))) && (((((int)threadIdx.x) + 4) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 49) / 9) * 7)) + ((((int)threadIdx.x) + 4) % 9)) + 6)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 98)] = ((((((((int)threadIdx.x) + 35) % 63) < 54) && (1 <= ((((int)threadIdx.x) + 8) % 9))) && (((((int)threadIdx.x) + 8) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 98) / 9) * 7)) + ((((int)threadIdx.x) + 8) % 9)) + 6)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 147)] = ((((((((int)threadIdx.x) + 21) % 63) < 54) && (1 <= ((((int)threadIdx.x) + 3) % 9))) && (((((int)threadIdx.x) + 3) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 147) / 9) * 7)) + ((((int)threadIdx.x) + 3) % 9)) + 6)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 196)] = ((((((int)threadIdx.x) < 47) && (1 <= ((((int)threadIdx.x) + 7) % 9))) && (((((int)threadIdx.x) + 7) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 196) / 9) * 7)) + ((((int)threadIdx.x) + 7) % 9)) + 6)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 245)] = ((((((((int)threadIdx.x) + 56) % 63) < 54) && (1 <= ((((int)threadIdx.x) + 2) % 9))) && (((((int)threadIdx.x) + 2) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 245) / 9) * 7)) + ((((int)threadIdx.x) + 2) % 9)) + 6)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 294)] = ((((((((int)threadIdx.x) + 42) % 63) < 54) && (1 <= ((((int)threadIdx.x) + 6) % 9))) && (((((int)threadIdx.x) + 6) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 294) / 9) * 7)) + ((((int)threadIdx.x) + 6) % 9)) + 6)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 343)] = ((((((((int)threadIdx.x) + 28) % 63) < 54) && (1 <= ((((int)threadIdx.x) + 1) % 9))) && (((((int)threadIdx.x) + 1) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 343) / 9) * 7)) + ((((int)threadIdx.x) + 1) % 9)) + 6)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 392)] = ((((((int)threadIdx.x) < 40) && (1 <= ((((int)threadIdx.x) + 5) % 9))) && (((((int)threadIdx.x) + 5) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 392) / 9) * 7)) + ((((int)threadIdx.x) + 5) % 9)) + 6)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 441)] = (((1 <= (((int)threadIdx.x) % 9)) && ((((int)threadIdx.x) % 9) < 8)) ? data[((((rc_outer_outer * 784) + ((((int)threadIdx.x) / 9) * 7)) + (((int)threadIdx.x) % 9)) + 349)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 490)] = ((((((((int)threadIdx.x) + 49) % 63) < 54) && (1 <= ((((int)threadIdx.x) + 4) % 9))) && (((((int)threadIdx.x) + 4) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 490) / 9) * 7)) + ((((int)threadIdx.x) + 4) % 9)) + 6)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 539)] = ((((((((int)threadIdx.x) + 35) % 63) < 54) && (1 <= ((((int)threadIdx.x) + 8) % 9))) && (((((int)threadIdx.x) + 8) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 539) / 9) * 7)) + ((((int)threadIdx.x) + 8) % 9)) + 6)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 588)] = ((((((((int)threadIdx.x) + 21) % 63) < 54) && (1 <= ((((int)threadIdx.x) + 3) % 9))) && (((((int)threadIdx.x) + 3) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 588) / 9) * 7)) + ((((int)threadIdx.x) + 3) % 9)) + 6)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 637)] = ((((((int)threadIdx.x) < 47) && (1 <= ((((int)threadIdx.x) + 7) % 9))) && (((((int)threadIdx.x) + 7) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 637) / 9) * 7)) + ((((int)threadIdx.x) + 7) % 9)) + 6)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 686)] = ((((((((int)threadIdx.x) + 56) % 63) < 54) && (1 <= ((((int)threadIdx.x) + 2) % 9))) && (((((int)threadIdx.x) + 2) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 686) / 9) * 7)) + ((((int)threadIdx.x) + 2) % 9)) + 6)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 735)] = ((((((((int)threadIdx.x) + 42) % 63) < 54) && (1 <= ((((int)threadIdx.x) + 6) % 9))) && (((((int)threadIdx.x) + 6) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 735) / 9) * 7)) + ((((int)threadIdx.x) + 6) % 9)) + 6)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 784)] = ((((((((int)threadIdx.x) + 28) % 63) < 54) && (1 <= ((((int)threadIdx.x) + 1) % 9))) && (((((int)threadIdx.x) + 1) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 784) / 9) * 7)) + ((((int)threadIdx.x) + 1) % 9)) + 6)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 833)] = ((((((int)threadIdx.x) < 40) && (1 <= ((((int)threadIdx.x) + 5) % 9))) && (((((int)threadIdx.x) + 5) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 833) / 9) * 7)) + ((((int)threadIdx.x) + 5) % 9)) + 6)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 882)] = (((1 <= (((int)threadIdx.x) % 9)) && ((((int)threadIdx.x) % 9) < 8)) ? data[((((rc_outer_outer * 784) + ((((int)threadIdx.x) / 9) * 7)) + (((int)threadIdx.x) % 9)) + 692)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 931)] = ((((((((int)threadIdx.x) + 49) % 63) < 54) && (1 <= ((((int)threadIdx.x) + 4) % 9))) && (((((int)threadIdx.x) + 4) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 931) / 9) * 7)) + ((((int)threadIdx.x) + 4) % 9)) + 6)] : 0.000000e+00f);
+ if (((int)threadIdx.x) < 28) {
+ pad_temp_shared[(((int)threadIdx.x) + 980)] = ((((((int)threadIdx.x) < 19) && (1 <= ((((int)threadIdx.x) + 8) % 9))) && (((((int)threadIdx.x) + 8) % 9) < 8)) ? data[((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 980) / 9) * 7)) + ((((int)threadIdx.x) + 8) % 9)) + 6)] : 0.000000e+00f);
}
- kernel_shared[((int)threadIdx.x)] = kernel[((((((int)blockIdx.x) * 36864) + ((((int)threadIdx.x) / 36) * 4608)) + (rc_outer_outer * 36)) + (((int)threadIdx.x) % 36))];
- kernel_shared[(((int)threadIdx.x) + 49)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 49) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 13) % 36) / 9) * 9)) + ((((((int)threadIdx.x) + 4) % 9) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
- kernel_shared[(((int)threadIdx.x) + 98)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 98) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 26) % 36) / 9) * 9)) + ((((((int)threadIdx.x) + 8) % 9) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
- kernel_shared[(((int)threadIdx.x) + 147)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 147) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 3) % 36) / 9) * 9)) + ((((((int)threadIdx.x) / 3) + 1) % 3) * 3)) + (((int)threadIdx.x) % 3))];
- kernel_shared[(((int)threadIdx.x) + 196)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 196) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 16) % 36) / 9) * 9)) + ((((((int)threadIdx.x) + 7) % 9) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
- if (((int)threadIdx.x) < 43) {
- kernel_shared[(((int)threadIdx.x) + 245)] = kernel[(((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 245) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 29) % 36) / 9) * 9)) + ((((int)threadIdx.x) + 2) % 9))];
+ kernel_shared[((int)threadIdx.x)] = kernel[((((((((int)blockIdx.x) * 18432) + ((((int)threadIdx.x) / 48) * 4608)) + (rc_outer_outer * 144)) + (((((int)threadIdx.x) % 48) / 3) * 9)) + (((int)threadIdx.x) % 3)) + 6)];
+ kernel_shared[(((int)threadIdx.x) + 49)] = kernel[((((((((int)blockIdx.x) * 18432) + (((((int)threadIdx.x) + 49) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 1) % 48) / 3) * 9)) + ((((int)threadIdx.x) + 1) % 3)) + 6)];
+ kernel_shared[(((int)threadIdx.x) + 98)] = kernel[((((((((int)blockIdx.x) * 18432) + (((((int)threadIdx.x) + 98) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 2) % 48) / 3) * 9)) + ((((int)threadIdx.x) + 2) % 3)) + 6)];
+ if (((int)threadIdx.x) < 45) {
+ kernel_shared[(((int)threadIdx.x) + 147)] = kernel[((((((((int)blockIdx.x) * 18432) + (((((int)threadIdx.x) + 147) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((int)threadIdx.x) / 3) * 9)) + (((int)threadIdx.x) % 3)) + 15)];
}
__syncthreads();
conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[0]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[36]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[72]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[108]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[144]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[180]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[216]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[252]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[48]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[96]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[144]));
conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[1]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[37]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[73]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[109]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[145]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[181]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[217]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[253]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[49]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[97]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[145]));
conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[2]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[38]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[74]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[110]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[146]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[182]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[218]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[254]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[3]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[39]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[75]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[111]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[147]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[183]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[219]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[255]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[4]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[40]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[76]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[112]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[148]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[184]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[220]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[256]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[5]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[41]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[77]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[113]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[149]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[185]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[221]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[257]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[6]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[42]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[78]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[114]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[150]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[186]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[222]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[258]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[7]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[43]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[79]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[115]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[151]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[187]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[223]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[259]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[8]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[44]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[80]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[116]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[152]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[188]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[224]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[260]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 81)] * kernel_shared[9]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 81)] * kernel_shared[45]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 81)] * kernel_shared[81]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 81)] * kernel_shared[117]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 81)] * kernel_shared[153]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 81)] * kernel_shared[189]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 81)] * kernel_shared[225]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 81)] * kernel_shared[261]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 82)] * kernel_shared[10]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 82)] * kernel_shared[46]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 82)] * kernel_shared[82]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 82)] * kernel_shared[118]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 82)] * kernel_shared[154]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 82)] * kernel_shared[190]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 82)] * kernel_shared[226]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 82)] * kernel_shared[262]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 83)] * kernel_shared[11]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 83)] * kernel_shared[47]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 83)] * kernel_shared[83]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 83)] * kernel_shared[119]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 83)] * kernel_shared[155]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 83)] * kernel_shared[191]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 83)] * kernel_shared[227]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 83)] * kernel_shared[263]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[12]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[48]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[84]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[120]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[156]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[192]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[228]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[264]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[13]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[49]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[85]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[121]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[157]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[193]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[229]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[265]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[14]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[50]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[86]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[122]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[158]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[194]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[230]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[266]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[15]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[51]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[87]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[123]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[159]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[195]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[231]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[267]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[16]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[52]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[88]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[124]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[160]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[196]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[232]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[268]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[17]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[53]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[89]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[125]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[161]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[197]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[233]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[269]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 162)] * kernel_shared[18]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 162)] * kernel_shared[54]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 162)] * kernel_shared[90]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 162)] * kernel_shared[126]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 162)] * kernel_shared[162]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 162)] * kernel_shared[198]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 162)] * kernel_shared[234]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 162)] * kernel_shared[270]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 163)] * kernel_shared[19]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 163)] * kernel_shared[55]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 163)] * kernel_shared[91]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 163)] * kernel_shared[127]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 163)] * kernel_shared[163]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 163)] * kernel_shared[199]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 163)] * kernel_shared[235]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 163)] * kernel_shared[271]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 164)] * kernel_shared[20]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 164)] * kernel_shared[56]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 164)] * kernel_shared[92]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 164)] * kernel_shared[128]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 164)] * kernel_shared[164]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 164)] * kernel_shared[200]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 164)] * kernel_shared[236]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 164)] * kernel_shared[272]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[21]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[57]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[93]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[129]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[165]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[201]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[237]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[273]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[22]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[58]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[94]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[130]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[166]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[202]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[238]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[274]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[23]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[59]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[95]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[131]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[167]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[203]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[239]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[275]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[24]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[60]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[96]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[132]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[168]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[204]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[240]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[276]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[25]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[61]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[97]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[133]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[169]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[205]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[241]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[277]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[26]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[62]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[98]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[134]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[170]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[206]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[242]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[278]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 243)] * kernel_shared[27]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 243)] * kernel_shared[63]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 243)] * kernel_shared[99]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 243)] * kernel_shared[135]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 243)] * kernel_shared[171]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 243)] * kernel_shared[207]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 243)] * kernel_shared[243]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 243)] * kernel_shared[279]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 244)] * kernel_shared[28]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 244)] * kernel_shared[64]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 244)] * kernel_shared[100]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 244)] * kernel_shared[136]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 244)] * kernel_shared[172]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 244)] * kernel_shared[208]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 244)] * kernel_shared[244]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 244)] * kernel_shared[280]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 245)] * kernel_shared[29]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 245)] * kernel_shared[65]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 245)] * kernel_shared[101]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 245)] * kernel_shared[137]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 245)] * kernel_shared[173]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 245)] * kernel_shared[209]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 245)] * kernel_shared[245]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 245)] * kernel_shared[281]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[30]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[66]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[102]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[138]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[174]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[210]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[246]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[282]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[31]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[67]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[103]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[139]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[175]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[211]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[247]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[283]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[32]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[68]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[104]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[140]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[176]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[212]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[248]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[284]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[33]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[69]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[105]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[141]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[177]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[213]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[249]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[285]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[34]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[70]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[106]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[142]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[178]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[214]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[250]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[286]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[35]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[71]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[107]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[143]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[179]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[215]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[251]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[287]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[50]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[98]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[146]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[3]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[51]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[99]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[147]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[4]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[52]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[100]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[148]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[5]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[53]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[101]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[149]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[6]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[54]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[102]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[150]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[7]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[55]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[103]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[151]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[8]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[56]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[104]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[152]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[9]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[57]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[105]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[153]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[10]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[58]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[106]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[154]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[11]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[59]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[107]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[155]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[12]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[60]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[108]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[156]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[13]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[61]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[109]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[157]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[14]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[62]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[110]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[158]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[15]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[63]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[111]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[159]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 316)] * kernel_shared[16]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 316)] * kernel_shared[64]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 316)] * kernel_shared[112]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 316)] * kernel_shared[160]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 317)] * kernel_shared[17]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 317)] * kernel_shared[65]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 317)] * kernel_shared[113]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 317)] * kernel_shared[161]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[18]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[66]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[114]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 378)] * kernel_shared[162]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 379)] * kernel_shared[19]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 379)] * kernel_shared[67]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 379)] * kernel_shared[115]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 379)] * kernel_shared[163]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 380)] * kernel_shared[20]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 380)] * kernel_shared[68]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 380)] * kernel_shared[116]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 380)] * kernel_shared[164]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[21]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[69]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[117]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 441)] * kernel_shared[165]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 442)] * kernel_shared[22]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 442)] * kernel_shared[70]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 442)] * kernel_shared[118]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 442)] * kernel_shared[166]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 443)] * kernel_shared[23]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 443)] * kernel_shared[71]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 443)] * kernel_shared[119]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 443)] * kernel_shared[167]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 504)] * kernel_shared[24]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 504)] * kernel_shared[72]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 504)] * kernel_shared[120]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 504)] * kernel_shared[168]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 505)] * kernel_shared[25]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 505)] * kernel_shared[73]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 505)] * kernel_shared[121]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 505)] * kernel_shared[169]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 506)] * kernel_shared[26]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 506)] * kernel_shared[74]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 506)] * kernel_shared[122]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 506)] * kernel_shared[170]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 567)] * kernel_shared[27]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 567)] * kernel_shared[75]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 567)] * kernel_shared[123]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 567)] * kernel_shared[171]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 568)] * kernel_shared[28]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 568)] * kernel_shared[76]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 568)] * kernel_shared[124]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 568)] * kernel_shared[172]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 569)] * kernel_shared[29]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 569)] * kernel_shared[77]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 569)] * kernel_shared[125]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 569)] * kernel_shared[173]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 630)] * kernel_shared[30]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 630)] * kernel_shared[78]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 630)] * kernel_shared[126]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 630)] * kernel_shared[174]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 631)] * kernel_shared[31]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 631)] * kernel_shared[79]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 631)] * kernel_shared[127]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 631)] * kernel_shared[175]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 632)] * kernel_shared[32]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 632)] * kernel_shared[80]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 632)] * kernel_shared[128]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 632)] * kernel_shared[176]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 693)] * kernel_shared[33]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 693)] * kernel_shared[81]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 693)] * kernel_shared[129]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 693)] * kernel_shared[177]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 694)] * kernel_shared[34]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 694)] * kernel_shared[82]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 694)] * kernel_shared[130]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 694)] * kernel_shared[178]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 695)] * kernel_shared[35]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 695)] * kernel_shared[83]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 695)] * kernel_shared[131]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 695)] * kernel_shared[179]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 756)] * kernel_shared[36]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 756)] * kernel_shared[84]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 756)] * kernel_shared[132]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 756)] * kernel_shared[180]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 757)] * kernel_shared[37]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 757)] * kernel_shared[85]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 757)] * kernel_shared[133]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 757)] * kernel_shared[181]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 758)] * kernel_shared[38]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 758)] * kernel_shared[86]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 758)] * kernel_shared[134]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 758)] * kernel_shared[182]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 819)] * kernel_shared[39]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 819)] * kernel_shared[87]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 819)] * kernel_shared[135]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 819)] * kernel_shared[183]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 820)] * kernel_shared[40]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 820)] * kernel_shared[88]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 820)] * kernel_shared[136]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 820)] * kernel_shared[184]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 821)] * kernel_shared[41]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 821)] * kernel_shared[89]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 821)] * kernel_shared[137]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 821)] * kernel_shared[185]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 882)] * kernel_shared[42]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 882)] * kernel_shared[90]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 882)] * kernel_shared[138]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 882)] * kernel_shared[186]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 883)] * kernel_shared[43]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 883)] * kernel_shared[91]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 883)] * kernel_shared[139]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 883)] * kernel_shared[187]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 884)] * kernel_shared[44]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 884)] * kernel_shared[92]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 884)] * kernel_shared[140]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 884)] * kernel_shared[188]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 945)] * kernel_shared[45]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 945)] * kernel_shared[93]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 945)] * kernel_shared[141]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 945)] * kernel_shared[189]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 946)] * kernel_shared[46]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 946)] * kernel_shared[94]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 946)] * kernel_shared[142]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 946)] * kernel_shared[190]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 947)] * kernel_shared[47]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 947)] * kernel_shared[95]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 947)] * kernel_shared[143]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((((((int)threadIdx.x) / 7) * 9) + (((int)threadIdx.x) % 7)) + 947)] * kernel_shared[191]));
+ }
+ for (int i1_inner = 0; i1_inner < 4; ++i1_inner) {
+ compute[(((((int)blockIdx.x) * 196) + (i1_inner * 49)) + ((int)threadIdx.x))] = max((conv2d_nchw[i1_inner] + bias[((((int)blockIdx.x) * 4) + i1_inner)]), 0.000000e+00f);
}
- compute[((((int)blockIdx.x) * 392) + ((int)threadIdx.x))] = max((conv2d_nchw[0] + bias[(((int)blockIdx.x) * 8)]), 0.000000e+00f);
- compute[(((((int)blockIdx.x) * 392) + ((int)threadIdx.x)) + 49)] = max((conv2d_nchw[1] + bias[((((int)blockIdx.x) * 8) + 1)]), 0.000000e+00f);
- compute[(((((int)blockIdx.x) * 392) + ((int)threadIdx.x)) + 98)] = max((conv2d_nchw[2] + bias[((((int)blockIdx.x) * 8) + 2)]), 0.000000e+00f);
- compute[(((((int)blockIdx.x) * 392) + ((int)threadIdx.x)) + 147)] = max((conv2d_nchw[3] + bias[((((int)blockIdx.x) * 8) + 3)]), 0.000000e+00f);
- compute[(((((int)blockIdx.x) * 392) + ((int)threadIdx.x)) + 196)] = max((conv2d_nchw[4] + bias[((((int)blockIdx.x) * 8) + 4)]), 0.000000e+00f);
- compute[(((((int)blockIdx.x) * 392) + ((int)threadIdx.x)) + 245)] = max((conv2d_nchw[5] + bias[((((int)blockIdx.x) * 8) + 5)]), 0.000000e+00f);
- compute[(((((int)blockIdx.x) * 392) + ((int)threadIdx.x)) + 294)] = max((conv2d_nchw[6] + bias[((((int)blockIdx.x) * 8) + 6)]), 0.000000e+00f);
- compute[(((((int)blockIdx.x) * 392) + ((int)threadIdx.x)) + 343)] = max((conv2d_nchw[7] + bias[((((int)blockIdx.x) * 8) + 7)]), 0.000000e+00f);
}
</pre></div>
</div>
@@ -1340,7 +2097,7 @@ 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> ( 2 minutes 29.143 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 2 minutes 35.377 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 509a27e5f..fd0dc741a 100644
--- a/docs/how_to/tune_with_autoscheduler/tune_network_cuda.html
+++ b/docs/how_to/tune_with_autoscheduler/tune_network_cuda.html
@@ -901,7 +901,7 @@ 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)
- 10.0936 10.0958 10.1276 10.0573 0.0288
+ 9.8384 9.8287 9.8629 9.8236 0.0174
</pre></div>
</div>
</div>
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 f8b2ecc13..160d14bfb 100644
--- a/docs/how_to/tune_with_autoscheduler/tune_network_x86.html
+++ b/docs/how_to/tune_with_autoscheduler/tune_network_x86.html
@@ -920,7 +920,7 @@ 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)
- 751.6743 751.1777 752.8035 751.0416 0.8004
+ 743.2332 743.7093 743.7099 742.2805 0.6737
</pre></div>
</div>
</div>
@@ -942,7 +942,7 @@ 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 20.297 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 19.283 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 99f66ad73..50b773c84 100644
--- a/docs/how_to/tune_with_autoscheduler/tune_sparse_x86.html
+++ b/docs/how_to/tune_with_autoscheduler/tune_sparse_x86.html
@@ -620,32 +620,1298 @@ layout transformation, parallelization, vectorization, unrolling, and operator f
placeholder_4: Buffer(placeholder_14: Pointer(float32), float32, [65536], []),
compute: Buffer(compute_2: Pointer(float32), float32, [65536], [])}
buffer_map = {placeholder_5: placeholder, placeholder_6: placeholder_1, placeholder_7: placeholder_2, placeholder_8: placeholder_3, placeholder_9: placeholder_4, compute_1: compute}
- preflattened_buffer_map = {placeholder_8: placeholder_15: Buffer(placeholder_13, int32, [33], []), placeholder_7: placeholder_16: Buffer(placeholder_12, int32, [4916], []), placeholder_9: placeholder_17: Buffer(placeholder_14, float32, [128, 512], []), placeholder_5: placeholder_18: Buffer(placeholder_10, float32, [128, 256], []), placeholder_6: placeholder_19: Buffer(placeholder_11, float32, [4916, 16, 1], []), compute_1: compute_3: Buffer(compute_2, float32, [128, 512], [])} {
- for (i0.outer.i1.outer.fused: int32, 0, 16) "parallel" {
- allocate(compute_4: Pointer(global float32), float32, [4096]), storage_scope = global {
- for (i.outer.inner: int32, 0, 32) {
- for (nb_j.inner: int32, 0, 2) {
- for (i.inner.init: int32, 0, 4) {
- for (j.init: int32, 0, 16) {
- compute_5: Buffer(compute_4, float32, [4096], [])[((((i.outer.inner*128) + (i.inner.init*32)) + (nb_j.inner*16)) + j.init)] = 0f32
+ preflattened_buffer_map = {placeholder_7: placeholder_15: Buffer(placeholder_12, int32, [4916], []), placeholder_8: placeholder_16: Buffer(placeholder_13, int32, [33], []), placeholder_9: placeholder_17: Buffer(placeholder_14, float32, [128, 512], []), placeholder_5: placeholder_18: Buffer(placeholder_10, float32, [128, 256], []), placeholder_6: placeholder_19: Buffer(placeholder_11, float32, [4916, 16, 1], []), compute_1: compute_3: Buffer(compute_2, float32, [128, 512], [])} {
+ for (i0.outer.i1.outer.fused: int32, 0, 32) "parallel" {
+ allocate(compute_4: Pointer(global float32), float32, [2048]), storage_scope = global {
+ for (i.outer.inner: int32, 0, 8) {
+ let cse_var_1: int32 = (i.outer.inner*256)
+ {
+ compute_5: Buffer(compute_4, float32, [2048], [])[cse_var_1] = 0f32
+ compute_5[(cse_var_1 + 1)] = 0f32
+ compute_5[(cse_var_1 + 2)] = 0f32
+ compute_5[(cse_var_1 + 3)] = 0f32
+ compute_5[(cse_var_1 + 4)] = 0f32
+ compute_5[(cse_var_1 + 5)] = 0f32
+ compute_5[(cse_var_1 + 6)] = 0f32
+ compute_5[(cse_var_1 + 7)] = 0f32
+ compute_5[(cse_var_1 + 8)] = 0f32
+ compute_5[(cse_var_1 + 9)] = 0f32
+ compute_5[(cse_var_1 + 10)] = 0f32
+ compute_5[(cse_var_1 + 11)] = 0f32
+ compute_5[(cse_var_1 + 12)] = 0f32
+ compute_5[(cse_var_1 + 13)] = 0f32
+ compute_5[(cse_var_1 + 14)] = 0f32
+ compute_5[(cse_var_1 + 15)] = 0f32
+ compute_5[(cse_var_1 + 16)] = 0f32
+ compute_5[(cse_var_1 + 17)] = 0f32
+ compute_5[(cse_var_1 + 18)] = 0f32
+ compute_5[(cse_var_1 + 19)] = 0f32
+ compute_5[(cse_var_1 + 20)] = 0f32
+ compute_5[(cse_var_1 + 21)] = 0f32
+ compute_5[(cse_var_1 + 22)] = 0f32
+ compute_5[(cse_var_1 + 23)] = 0f32
+ compute_5[(cse_var_1 + 24)] = 0f32
+ compute_5[(cse_var_1 + 25)] = 0f32
+ compute_5[(cse_var_1 + 26)] = 0f32
+ compute_5[(cse_var_1 + 27)] = 0f32
+ compute_5[(cse_var_1 + 28)] = 0f32
+ compute_5[(cse_var_1 + 29)] = 0f32
+ compute_5[(cse_var_1 + 30)] = 0f32
+ compute_5[(cse_var_1 + 31)] = 0f32
+ compute_5[(cse_var_1 + 32)] = 0f32
+ compute_5[(cse_var_1 + 33)] = 0f32
+ compute_5[(cse_var_1 + 34)] = 0f32
+ compute_5[(cse_var_1 + 35)] = 0f32
+ compute_5[(cse_var_1 + 36)] = 0f32
+ compute_5[(cse_var_1 + 37)] = 0f32
+ compute_5[(cse_var_1 + 38)] = 0f32
+ compute_5[(cse_var_1 + 39)] = 0f32
+ compute_5[(cse_var_1 + 40)] = 0f32
+ compute_5[(cse_var_1 + 41)] = 0f32
+ compute_5[(cse_var_1 + 42)] = 0f32
+ compute_5[(cse_var_1 + 43)] = 0f32
+ compute_5[(cse_var_1 + 44)] = 0f32
+ compute_5[(cse_var_1 + 45)] = 0f32
+ compute_5[(cse_var_1 + 46)] = 0f32
+ compute_5[(cse_var_1 + 47)] = 0f32
+ compute_5[(cse_var_1 + 48)] = 0f32
+ compute_5[(cse_var_1 + 49)] = 0f32
+ compute_5[(cse_var_1 + 50)] = 0f32
+ compute_5[(cse_var_1 + 51)] = 0f32
+ compute_5[(cse_var_1 + 52)] = 0f32
+ compute_5[(cse_var_1 + 53)] = 0f32
+ compute_5[(cse_var_1 + 54)] = 0f32
+ compute_5[(cse_var_1 + 55)] = 0f32
+ compute_5[(cse_var_1 + 56)] = 0f32
+ compute_5[(cse_var_1 + 57)] = 0f32
+ compute_5[(cse_var_1 + 58)] = 0f32
+ compute_5[(cse_var_1 + 59)] = 0f32
+ compute_5[(cse_var_1 + 60)] = 0f32
+ compute_5[(cse_var_1 + 61)] = 0f32
+ compute_5[(cse_var_1 + 62)] = 0f32
+ compute_5[(cse_var_1 + 63)] = 0f32
+ compute_5[(cse_var_1 + 64)] = 0f32
+ compute_5[(cse_var_1 + 65)] = 0f32
+ compute_5[(cse_var_1 + 66)] = 0f32
+ compute_5[(cse_var_1 + 67)] = 0f32
+ compute_5[(cse_var_1 + 68)] = 0f32
+ compute_5[(cse_var_1 + 69)] = 0f32
+ compute_5[(cse_var_1 + 70)] = 0f32
+ compute_5[(cse_var_1 + 71)] = 0f32
+ compute_5[(cse_var_1 + 72)] = 0f32
+ compute_5[(cse_var_1 + 73)] = 0f32
+ compute_5[(cse_var_1 + 74)] = 0f32
+ compute_5[(cse_var_1 + 75)] = 0f32
+ compute_5[(cse_var_1 + 76)] = 0f32
+ compute_5[(cse_var_1 + 77)] = 0f32
+ compute_5[(cse_var_1 + 78)] = 0f32
+ compute_5[(cse_var_1 + 79)] = 0f32
+ compute_5[(cse_var_1 + 80)] = 0f32
+ compute_5[(cse_var_1 + 81)] = 0f32
+ compute_5[(cse_var_1 + 82)] = 0f32
+ compute_5[(cse_var_1 + 83)] = 0f32
+ compute_5[(cse_var_1 + 84)] = 0f32
+ compute_5[(cse_var_1 + 85)] = 0f32
+ compute_5[(cse_var_1 + 86)] = 0f32
+ compute_5[(cse_var_1 + 87)] = 0f32
+ compute_5[(cse_var_1 + 88)] = 0f32
+ compute_5[(cse_var_1 + 89)] = 0f32
+ compute_5[(cse_var_1 + 90)] = 0f32
+ compute_5[(cse_var_1 + 91)] = 0f32
+ compute_5[(cse_var_1 + 92)] = 0f32
+ compute_5[(cse_var_1 + 93)] = 0f32
+ compute_5[(cse_var_1 + 94)] = 0f32
+ compute_5[(cse_var_1 + 95)] = 0f32
+ compute_5[(cse_var_1 + 96)] = 0f32
+ compute_5[(cse_var_1 + 97)] = 0f32
+ compute_5[(cse_var_1 + 98)] = 0f32
+ compute_5[(cse_var_1 + 99)] = 0f32
+ compute_5[(cse_var_1 + 100)] = 0f32
+ compute_5[(cse_var_1 + 101)] = 0f32
+ compute_5[(cse_var_1 + 102)] = 0f32
+ compute_5[(cse_var_1 + 103)] = 0f32
+ compute_5[(cse_var_1 + 104)] = 0f32
+ compute_5[(cse_var_1 + 105)] = 0f32
+ compute_5[(cse_var_1 + 106)] = 0f32
+ compute_5[(cse_var_1 + 107)] = 0f32
+ compute_5[(cse_var_1 + 108)] = 0f32
+ compute_5[(cse_var_1 + 109)] = 0f32
+ compute_5[(cse_var_1 + 110)] = 0f32
+ compute_5[(cse_var_1 + 111)] = 0f32
+ compute_5[(cse_var_1 + 112)] = 0f32
+ compute_5[(cse_var_1 + 113)] = 0f32
+ compute_5[(cse_var_1 + 114)] = 0f32
+ compute_5[(cse_var_1 + 115)] = 0f32
+ compute_5[(cse_var_1 + 116)] = 0f32
+ compute_5[(cse_var_1 + 117)] = 0f32
+ compute_5[(cse_var_1 + 118)] = 0f32
+ compute_5[(cse_var_1 + 119)] = 0f32
+ compute_5[(cse_var_1 + 120)] = 0f32
+ compute_5[(cse_var_1 + 121)] = 0f32
+ compute_5[(cse_var_1 + 122)] = 0f32
+ compute_5[(cse_var_1 + 123)] = 0f32
+ compute_5[(cse_var_1 + 124)] = 0f32
+ compute_5[(cse_var_1 + 125)] = 0f32
+ compute_5[(cse_var_1 + 126)] = 0f32
+ compute_5[(cse_var_1 + 127)] = 0f32
+ compute_5[(cse_var_1 + 128)] = 0f32
+ compute_5[(cse_var_1 + 129)] = 0f32
+ compute_5[(cse_var_1 + 130)] = 0f32
+ compute_5[(cse_var_1 + 131)] = 0f32
+ compute_5[(cse_var_1 + 132)] = 0f32
+ compute_5[(cse_var_1 + 133)] = 0f32
+ compute_5[(cse_var_1 + 134)] = 0f32
+ compute_5[(cse_var_1 + 135)] = 0f32
+ compute_5[(cse_var_1 + 136)] = 0f32
+ compute_5[(cse_var_1 + 137)] = 0f32
+ compute_5[(cse_var_1 + 138)] = 0f32
+ compute_5[(cse_var_1 + 139)] = 0f32
+ compute_5[(cse_var_1 + 140)] = 0f32
+ compute_5[(cse_var_1 + 141)] = 0f32
+ compute_5[(cse_var_1 + 142)] = 0f32
+ compute_5[(cse_var_1 + 143)] = 0f32
+ compute_5[(cse_var_1 + 144)] = 0f32
+ compute_5[(cse_var_1 + 145)] = 0f32
+ compute_5[(cse_var_1 + 146)] = 0f32
+ compute_5[(cse_var_1 + 147)] = 0f32
+ compute_5[(cse_var_1 + 148)] = 0f32
+ compute_5[(cse_var_1 + 149)] = 0f32
+ compute_5[(cse_var_1 + 150)] = 0f32
+ compute_5[(cse_var_1 + 151)] = 0f32
+ compute_5[(cse_var_1 + 152)] = 0f32
+ compute_5[(cse_var_1 + 153)] = 0f32
+ compute_5[(cse_var_1 + 154)] = 0f32
+ compute_5[(cse_var_1 + 155)] = 0f32
+ compute_5[(cse_var_1 + 156)] = 0f32
+ compute_5[(cse_var_1 + 157)] = 0f32
+ compute_5[(cse_var_1 + 158)] = 0f32
+ compute_5[(cse_var_1 + 159)] = 0f32
+ compute_5[(cse_var_1 + 160)] = 0f32
+ compute_5[(cse_var_1 + 161)] = 0f32
+ compute_5[(cse_var_1 + 162)] = 0f32
+ compute_5[(cse_var_1 + 163)] = 0f32
+ compute_5[(cse_var_1 + 164)] = 0f32
+ compute_5[(cse_var_1 + 165)] = 0f32
+ compute_5[(cse_var_1 + 166)] = 0f32
+ compute_5[(cse_var_1 + 167)] = 0f32
+ compute_5[(cse_var_1 + 168)] = 0f32
+ compute_5[(cse_var_1 + 169)] = 0f32
+ compute_5[(cse_var_1 + 170)] = 0f32
+ compute_5[(cse_var_1 + 171)] = 0f32
+ compute_5[(cse_var_1 + 172)] = 0f32
+ compute_5[(cse_var_1 + 173)] = 0f32
+ compute_5[(cse_var_1 + 174)] = 0f32
+ compute_5[(cse_var_1 + 175)] = 0f32
+ compute_5[(cse_var_1 + 176)] = 0f32
+ compute_5[(cse_var_1 + 177)] = 0f32
+ compute_5[(cse_var_1 + 178)] = 0f32
+ compute_5[(cse_var_1 + 179)] = 0f32
+ compute_5[(cse_var_1 + 180)] = 0f32
+ compute_5[(cse_var_1 + 181)] = 0f32
+ compute_5[(cse_var_1 + 182)] = 0f32
+ compute_5[(cse_var_1 + 183)] = 0f32
+ compute_5[(cse_var_1 + 184)] = 0f32
+ compute_5[(cse_var_1 + 185)] = 0f32
+ compute_5[(cse_var_1 + 186)] = 0f32
+ compute_5[(cse_var_1 + 187)] = 0f32
+ compute_5[(cse_var_1 + 188)] = 0f32
+ compute_5[(cse_var_1 + 189)] = 0f32
+ compute_5[(cse_var_1 + 190)] = 0f32
+ compute_5[(cse_var_1 + 191)] = 0f32
+ compute_5[(cse_var_1 + 192)] = 0f32
+ compute_5[(cse_var_1 + 193)] = 0f32
+ compute_5[(cse_var_1 + 194)] = 0f32
+ compute_5[(cse_var_1 + 195)] = 0f32
+ compute_5[(cse_var_1 + 196)] = 0f32
+ compute_5[(cse_var_1 + 197)] = 0f32
+ compute_5[(cse_var_1 + 198)] = 0f32
+ compute_5[(cse_var_1 + 199)] = 0f32
+ compute_5[(cse_var_1 + 200)] = 0f32
+ compute_5[(cse_var_1 + 201)] = 0f32
+ compute_5[(cse_var_1 + 202)] = 0f32
+ compute_5[(cse_var_1 + 203)] = 0f32
+ compute_5[(cse_var_1 + 204)] = 0f32
+ compute_5[(cse_var_1 + 205)] = 0f32
+ compute_5[(cse_var_1 + 206)] = 0f32
+ compute_5[(cse_var_1 + 207)] = 0f32
+ compute_5[(cse_var_1 + 208)] = 0f32
+ compute_5[(cse_var_1 + 209)] = 0f32
+ compute_5[(cse_var_1 + 210)] = 0f32
+ compute_5[(cse_var_1 + 211)] = 0f32
+ compute_5[(cse_var_1 + 212)] = 0f32
+ compute_5[(cse_var_1 + 213)] = 0f32
+ compute_5[(cse_var_1 + 214)] = 0f32
+ compute_5[(cse_var_1 + 215)] = 0f32
+ compute_5[(cse_var_1 + 216)] = 0f32
+ compute_5[(cse_var_1 + 217)] = 0f32
+ compute_5[(cse_var_1 + 218)] = 0f32
+ compute_5[(cse_var_1 + 219)] = 0f32
+ compute_5[(cse_var_1 + 220)] = 0f32
+ compute_5[(cse_var_1 + 221)] = 0f32
+ compute_5[(cse_var_1 + 222)] = 0f32
+ compute_5[(cse_var_1 + 223)] = 0f32
+ compute_5[(cse_var_1 + 224)] = 0f32
+ compute_5[(cse_var_1 + 225)] = 0f32
+ compute_5[(cse_var_1 + 226)] = 0f32
+ compute_5[(cse_var_1 + 227)] = 0f32
+ compute_5[(cse_var_1 + 228)] = 0f32
+ compute_5[(cse_var_1 + 229)] = 0f32
+ compute_5[(cse_var_1 + 230)] = 0f32
+ compute_5[(cse_var_1 + 231)] = 0f32
+ compute_5[(cse_var_1 + 232)] = 0f32
+ compute_5[(cse_var_1 + 233)] = 0f32
+ compute_5[(cse_var_1 + 234)] = 0f32
+ compute_5[(cse_var_1 + 235)] = 0f32
+ compute_5[(cse_var_1 + 236)] = 0f32
+ compute_5[(cse_var_1 + 237)] = 0f32
+ compute_5[(cse_var_1 + 238)] = 0f32
+ compute_5[(cse_var_1 + 239)] = 0f32
+ compute_5[(cse_var_1 + 240)] = 0f32
+ compute_5[(cse_var_1 + 241)] = 0f32
+ compute_5[(cse_var_1 + 242)] = 0f32
+ compute_5[(cse_var_1 + 243)] = 0f32
+ compute_5[(cse_var_1 + 244)] = 0f32
+ compute_5[(cse_var_1 + 245)] = 0f32
+ compute_5[(cse_var_1 + 246)] = 0f32
+ compute_5[(cse_var_1 + 247)] = 0f32
+ compute_5[(cse_var_1 + 248)] = 0f32
+ compute_5[(cse_var_1 + 249)] = 0f32
+ compute_5[(cse_var_1 + 250)] = 0f32
+ compute_5[(cse_var_1 + 251)] = 0f32
+ compute_5[(cse_var_1 + 252)] = 0f32
+ compute_5[(cse_var_1 + 253)] = 0f32
+ compute_5[(cse_var_1 + 254)] = 0f32
+ compute_5[(cse_var_1 + 255)] = 0f32
+ for (elem_idx: int32, 0, (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])) {
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ compute_5[cse_var_1] = (compute_5[cse_var_1] + (placeholder_1[((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16))]*max(placeholder[((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)])], 0f32)))
}
- }
- for (elem_idx: int32, 0, let cse_var_1: int32 = ((i0.outer.i1.outer.fused*2) + nb_j.inner) in (placeholder_3[(cse_var_1 + 1)] - placeholder_3[cse_var_1])) {
- for (i.inner: int32, 0, 4) {
- for (j: int32, 0, 16) {
- let cse_var_3: int32 = ((i0.outer.i1.outer.fused*2) + nb_j.inner)
- let cse_var_2: int32 = ((((i.outer.inner*128) + (i.inner*32)) + (nb_j.inner*16)) + j)
- compute_5[cse_var_2] = (compute_5[cse_var_2] + (placeholder_1[(((placeholder_3[cse_var_3]*16) + (elem_idx*16)) + j)]*max(placeholder[(((i.outer.inner*1024) + (i.inner*256)) + placeholder_2[(placeholder_3[cse_var_3] + elem_idx)])], 0f32)))
- }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_2: int32 = (cse_var_1 + 1)
+ compute_5[cse_var_2] = (compute_5[cse_var_2] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 1)]*max(placeholder[((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)])], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_3: int32 = (cse_var_1 + 2)
+ compute_5[cse_var_3] = (compute_5[cse_var_3] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 2)]*max(placeholder[((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)])], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_4: int32 = (cse_var_1 + 3)
+ compute_5[cse_var_4] = (compute_5[cse_var_4] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 3)]*max(placeholder[((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)])], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_5: int32 = (cse_var_1 + 4)
+ compute_5[cse_var_5] = (compute_5[cse_var_5] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 4)]*max(placeholder[((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)])], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_6: int32 = (cse_var_1 + 5)
+ compute_5[cse_var_6] = (compute_5[cse_var_6] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 5)]*max(placeholder[((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)])], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_7: int32 = (cse_var_1 + 6)
+ compute_5[cse_var_7] = (compute_5[cse_var_7] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 6)]*max(placeholder[((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)])], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_8: int32 = (cse_var_1 + 7)
+ compute_5[cse_var_8] = (compute_5[cse_var_8] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 7)]*max(placeholder[((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)])], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_9: int32 = (cse_var_1 + 8)
+ compute_5[cse_var_9] = (compute_5[cse_var_9] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 8)]*max(placeholder[((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)])], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_10: int32 = (cse_var_1 + 9)
+ compute_5[cse_var_10] = (compute_5[cse_var_10] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 9)]*max(placeholder[((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)])], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_11: int32 = (cse_var_1 + 10)
+ compute_5[cse_var_11] = (compute_5[cse_var_11] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 10)]*max(placeholder[((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)])], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_12: int32 = (cse_var_1 + 11)
+ compute_5[cse_var_12] = (compute_5[cse_var_12] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 11)]*max(placeholder[((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)])], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_13: int32 = (cse_var_1 + 12)
+ compute_5[cse_var_13] = (compute_5[cse_var_13] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 12)]*max(placeholder[((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)])], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_14: int32 = (cse_var_1 + 13)
+ compute_5[cse_var_14] = (compute_5[cse_var_14] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 13)]*max(placeholder[((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)])], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_15: int32 = (cse_var_1 + 14)
+ compute_5[cse_var_15] = (compute_5[cse_var_15] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 14)]*max(placeholder[((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)])], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_16: int32 = (cse_var_1 + 15)
+ compute_5[cse_var_16] = (compute_5[cse_var_16] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 15)]*max(placeholder[((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)])], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_17: int32 = (cse_var_1 + 16)
+ compute_5[cse_var_17] = (compute_5[cse_var_17] + (placeholder_1[((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16))]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 256)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_18: int32 = (cse_var_1 + 17)
+ compute_5[cse_var_18] = (compute_5[cse_var_18] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 1)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 256)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_19: int32 = (cse_var_1 + 18)
+ compute_5[cse_var_19] = (compute_5[cse_var_19] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 2)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 256)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_20: int32 = (cse_var_1 + 19)
+ compute_5[cse_var_20] = (compute_5[cse_var_20] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 3)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 256)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_21: int32 = (cse_var_1 + 20)
+ compute_5[cse_var_21] = (compute_5[cse_var_21] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 4)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 256)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_22: int32 = (cse_var_1 + 21)
+ compute_5[cse_var_22] = (compute_5[cse_var_22] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 5)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 256)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_23: int32 = (cse_var_1 + 22)
+ compute_5[cse_var_23] = (compute_5[cse_var_23] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 6)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 256)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_24: int32 = (cse_var_1 + 23)
+ compute_5[cse_var_24] = (compute_5[cse_var_24] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 7)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 256)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_25: int32 = (cse_var_1 + 24)
+ compute_5[cse_var_25] = (compute_5[cse_var_25] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 8)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 256)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_26: int32 = (cse_var_1 + 25)
+ compute_5[cse_var_26] = (compute_5[cse_var_26] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 9)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 256)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_27: int32 = (cse_var_1 + 26)
+ compute_5[cse_var_27] = (compute_5[cse_var_27] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 10)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 256)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_28: int32 = (cse_var_1 + 27)
+ compute_5[cse_var_28] = (compute_5[cse_var_28] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 11)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 256)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_29: int32 = (cse_var_1 + 28)
+ compute_5[cse_var_29] = (compute_5[cse_var_29] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 12)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 256)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_30: int32 = (cse_var_1 + 29)
+ compute_5[cse_var_30] = (compute_5[cse_var_30] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 13)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 256)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_31: int32 = (cse_var_1 + 30)
+ compute_5[cse_var_31] = (compute_5[cse_var_31] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 14)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 256)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_32: int32 = (cse_var_1 + 31)
+ compute_5[cse_var_32] = (compute_5[cse_var_32] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 15)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 256)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_33: int32 = (cse_var_1 + 32)
+ compute_5[cse_var_33] = (compute_5[cse_var_33] + (placeholder_1[((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16))]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 512)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_34: int32 = (cse_var_1 + 33)
+ compute_5[cse_var_34] = (compute_5[cse_var_34] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 1)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 512)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_35: int32 = (cse_var_1 + 34)
+ compute_5[cse_var_35] = (compute_5[cse_var_35] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 2)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 512)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_36: int32 = (cse_var_1 + 35)
+ compute_5[cse_var_36] = (compute_5[cse_var_36] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 3)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 512)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_37: int32 = (cse_var_1 + 36)
+ compute_5[cse_var_37] = (compute_5[cse_var_37] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 4)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 512)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_38: int32 = (cse_var_1 + 37)
+ compute_5[cse_var_38] = (compute_5[cse_var_38] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 5)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 512)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_39: int32 = (cse_var_1 + 38)
+ compute_5[cse_var_39] = (compute_5[cse_var_39] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 6)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 512)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_40: int32 = (cse_var_1 + 39)
+ compute_5[cse_var_40] = (compute_5[cse_var_40] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 7)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 512)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_41: int32 = (cse_var_1 + 40)
+ compute_5[cse_var_41] = (compute_5[cse_var_41] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 8)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 512)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_42: int32 = (cse_var_1 + 41)
+ compute_5[cse_var_42] = (compute_5[cse_var_42] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 9)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 512)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_43: int32 = (cse_var_1 + 42)
+ compute_5[cse_var_43] = (compute_5[cse_var_43] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 10)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 512)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_44: int32 = (cse_var_1 + 43)
+ compute_5[cse_var_44] = (compute_5[cse_var_44] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 11)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 512)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_45: int32 = (cse_var_1 + 44)
+ compute_5[cse_var_45] = (compute_5[cse_var_45] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 12)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 512)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_46: int32 = (cse_var_1 + 45)
+ compute_5[cse_var_46] = (compute_5[cse_var_46] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 13)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 512)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_47: int32 = (cse_var_1 + 46)
+ compute_5[cse_var_47] = (compute_5[cse_var_47] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 14)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 512)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_48: int32 = (cse_var_1 + 47)
+ compute_5[cse_var_48] = (compute_5[cse_var_48] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 15)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 512)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_49: int32 = (cse_var_1 + 48)
+ compute_5[cse_var_49] = (compute_5[cse_var_49] + (placeholder_1[((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16))]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 768)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_50: int32 = (cse_var_1 + 49)
+ compute_5[cse_var_50] = (compute_5[cse_var_50] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 1)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 768)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_51: int32 = (cse_var_1 + 50)
+ compute_5[cse_var_51] = (compute_5[cse_var_51] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 2)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 768)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_52: int32 = (cse_var_1 + 51)
+ compute_5[cse_var_52] = (compute_5[cse_var_52] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 3)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 768)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_53: int32 = (cse_var_1 + 52)
+ compute_5[cse_var_53] = (compute_5[cse_var_53] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 4)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 768)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_54: int32 = (cse_var_1 + 53)
+ compute_5[cse_var_54] = (compute_5[cse_var_54] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 5)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 768)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_55: int32 = (cse_var_1 + 54)
+ compute_5[cse_var_55] = (compute_5[cse_var_55] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 6)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 768)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_56: int32 = (cse_var_1 + 55)
+ compute_5[cse_var_56] = (compute_5[cse_var_56] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 7)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 768)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_57: int32 = (cse_var_1 + 56)
+ compute_5[cse_var_57] = (compute_5[cse_var_57] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 8)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 768)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_58: int32 = (cse_var_1 + 57)
+ compute_5[cse_var_58] = (compute_5[cse_var_58] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 9)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 768)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_59: int32 = (cse_var_1 + 58)
+ compute_5[cse_var_59] = (compute_5[cse_var_59] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 10)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 768)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_60: int32 = (cse_var_1 + 59)
+ compute_5[cse_var_60] = (compute_5[cse_var_60] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 11)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 768)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_61: int32 = (cse_var_1 + 60)
+ compute_5[cse_var_61] = (compute_5[cse_var_61] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 12)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 768)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_62: int32 = (cse_var_1 + 61)
+ compute_5[cse_var_62] = (compute_5[cse_var_62] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 13)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 768)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_63: int32 = (cse_var_1 + 62)
+ compute_5[cse_var_63] = (compute_5[cse_var_63] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 14)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 768)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_64: int32 = (cse_var_1 + 63)
+ compute_5[cse_var_64] = (compute_5[cse_var_64] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 15)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 768)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_65: int32 = (cse_var_1 + 64)
+ compute_5[cse_var_65] = (compute_5[cse_var_65] + (placeholder_1[((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16))]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1024)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_66: int32 = (cse_var_1 + 65)
+ compute_5[cse_var_66] = (compute_5[cse_var_66] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 1)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1024)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_67: int32 = (cse_var_1 + 66)
+ compute_5[cse_var_67] = (compute_5[cse_var_67] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 2)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1024)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_68: int32 = (cse_var_1 + 67)
+ compute_5[cse_var_68] = (compute_5[cse_var_68] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 3)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1024)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_69: int32 = (cse_var_1 + 68)
+ compute_5[cse_var_69] = (compute_5[cse_var_69] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 4)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1024)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_70: int32 = (cse_var_1 + 69)
+ compute_5[cse_var_70] = (compute_5[cse_var_70] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 5)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1024)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_71: int32 = (cse_var_1 + 70)
+ compute_5[cse_var_71] = (compute_5[cse_var_71] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 6)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1024)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_72: int32 = (cse_var_1 + 71)
+ compute_5[cse_var_72] = (compute_5[cse_var_72] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 7)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1024)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_73: int32 = (cse_var_1 + 72)
+ compute_5[cse_var_73] = (compute_5[cse_var_73] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 8)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1024)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_74: int32 = (cse_var_1 + 73)
+ compute_5[cse_var_74] = (compute_5[cse_var_74] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 9)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1024)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_75: int32 = (cse_var_1 + 74)
+ compute_5[cse_var_75] = (compute_5[cse_var_75] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 10)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1024)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_76: int32 = (cse_var_1 + 75)
+ compute_5[cse_var_76] = (compute_5[cse_var_76] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 11)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1024)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_77: int32 = (cse_var_1 + 76)
+ compute_5[cse_var_77] = (compute_5[cse_var_77] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 12)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1024)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_78: int32 = (cse_var_1 + 77)
+ compute_5[cse_var_78] = (compute_5[cse_var_78] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 13)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1024)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_79: int32 = (cse_var_1 + 78)
+ compute_5[cse_var_79] = (compute_5[cse_var_79] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 14)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1024)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_80: int32 = (cse_var_1 + 79)
+ compute_5[cse_var_80] = (compute_5[cse_var_80] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 15)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1024)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_81: int32 = (cse_var_1 + 80)
+ compute_5[cse_var_81] = (compute_5[cse_var_81] + (placeholder_1[((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16))]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1280)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_82: int32 = (cse_var_1 + 81)
+ compute_5[cse_var_82] = (compute_5[cse_var_82] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 1)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1280)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_83: int32 = (cse_var_1 + 82)
+ compute_5[cse_var_83] = (compute_5[cse_var_83] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 2)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1280)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_84: int32 = (cse_var_1 + 83)
+ compute_5[cse_var_84] = (compute_5[cse_var_84] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 3)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1280)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_85: int32 = (cse_var_1 + 84)
+ compute_5[cse_var_85] = (compute_5[cse_var_85] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 4)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1280)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_86: int32 = (cse_var_1 + 85)
+ compute_5[cse_var_86] = (compute_5[cse_var_86] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 5)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1280)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_87: int32 = (cse_var_1 + 86)
+ compute_5[cse_var_87] = (compute_5[cse_var_87] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 6)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1280)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_88: int32 = (cse_var_1 + 87)
+ compute_5[cse_var_88] = (compute_5[cse_var_88] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 7)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1280)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_89: int32 = (cse_var_1 + 88)
+ compute_5[cse_var_89] = (compute_5[cse_var_89] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 8)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1280)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_90: int32 = (cse_var_1 + 89)
+ compute_5[cse_var_90] = (compute_5[cse_var_90] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 9)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1280)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_91: int32 = (cse_var_1 + 90)
+ compute_5[cse_var_91] = (compute_5[cse_var_91] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 10)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1280)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_92: int32 = (cse_var_1 + 91)
+ compute_5[cse_var_92] = (compute_5[cse_var_92] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 11)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1280)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_93: int32 = (cse_var_1 + 92)
+ compute_5[cse_var_93] = (compute_5[cse_var_93] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 12)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1280)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_94: int32 = (cse_var_1 + 93)
+ compute_5[cse_var_94] = (compute_5[cse_var_94] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 13)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1280)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_95: int32 = (cse_var_1 + 94)
+ compute_5[cse_var_95] = (compute_5[cse_var_95] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 14)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1280)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_96: int32 = (cse_var_1 + 95)
+ compute_5[cse_var_96] = (compute_5[cse_var_96] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 15)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1280)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_97: int32 = (cse_var_1 + 96)
+ compute_5[cse_var_97] = (compute_5[cse_var_97] + (placeholder_1[((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16))]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1536)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_98: int32 = (cse_var_1 + 97)
+ compute_5[cse_var_98] = (compute_5[cse_var_98] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 1)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1536)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_99: int32 = (cse_var_1 + 98)
+ compute_5[cse_var_99] = (compute_5[cse_var_99] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 2)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1536)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_100: int32 = (cse_var_1 + 99)
+ compute_5[cse_var_100] = (compute_5[cse_var_100] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 3)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1536)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_101: int32 = (cse_var_1 + 100)
+ compute_5[cse_var_101] = (compute_5[cse_var_101] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 4)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1536)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_102: int32 = (cse_var_1 + 101)
+ compute_5[cse_var_102] = (compute_5[cse_var_102] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 5)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1536)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_103: int32 = (cse_var_1 + 102)
+ compute_5[cse_var_103] = (compute_5[cse_var_103] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 6)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1536)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_104: int32 = (cse_var_1 + 103)
+ compute_5[cse_var_104] = (compute_5[cse_var_104] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 7)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1536)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_105: int32 = (cse_var_1 + 104)
+ compute_5[cse_var_105] = (compute_5[cse_var_105] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 8)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1536)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_106: int32 = (cse_var_1 + 105)
+ compute_5[cse_var_106] = (compute_5[cse_var_106] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 9)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1536)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_107: int32 = (cse_var_1 + 106)
+ compute_5[cse_var_107] = (compute_5[cse_var_107] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 10)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1536)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_108: int32 = (cse_var_1 + 107)
+ compute_5[cse_var_108] = (compute_5[cse_var_108] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 11)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1536)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_109: int32 = (cse_var_1 + 108)
+ compute_5[cse_var_109] = (compute_5[cse_var_109] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 12)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1536)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_110: int32 = (cse_var_1 + 109)
+ compute_5[cse_var_110] = (compute_5[cse_var_110] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 13)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1536)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_111: int32 = (cse_var_1 + 110)
+ compute_5[cse_var_111] = (compute_5[cse_var_111] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 14)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1536)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_112: int32 = (cse_var_1 + 111)
+ compute_5[cse_var_112] = (compute_5[cse_var_112] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 15)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1536)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_113: int32 = (cse_var_1 + 112)
+ compute_5[cse_var_113] = (compute_5[cse_var_113] + (placeholder_1[((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16))]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1792)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_114: int32 = (cse_var_1 + 113)
+ compute_5[cse_var_114] = (compute_5[cse_var_114] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 1)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1792)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_115: int32 = (cse_var_1 + 114)
+ compute_5[cse_var_115] = (compute_5[cse_var_115] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 2)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1792)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_116: int32 = (cse_var_1 + 115)
+ compute_5[cse_var_116] = (compute_5[cse_var_116] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 3)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1792)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_117: int32 = (cse_var_1 + 116)
+ compute_5[cse_var_117] = (compute_5[cse_var_117] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 4)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1792)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_118: int32 = (cse_var_1 + 117)
+ compute_5[cse_var_118] = (compute_5[cse_var_118] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 5)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1792)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_119: int32 = (cse_var_1 + 118)
+ compute_5[cse_var_119] = (compute_5[cse_var_119] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 6)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1792)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_120: int32 = (cse_var_1 + 119)
+ compute_5[cse_var_120] = (compute_5[cse_var_120] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 7)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1792)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_121: int32 = (cse_var_1 + 120)
+ compute_5[cse_var_121] = (compute_5[cse_var_121] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 8)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1792)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_122: int32 = (cse_var_1 + 121)
+ compute_5[cse_var_122] = (compute_5[cse_var_122] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 9)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1792)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_123: int32 = (cse_var_1 + 122)
+ compute_5[cse_var_123] = (compute_5[cse_var_123] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 10)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1792)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_124: int32 = (cse_var_1 + 123)
+ compute_5[cse_var_124] = (compute_5[cse_var_124] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 11)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1792)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_125: int32 = (cse_var_1 + 124)
+ compute_5[cse_var_125] = (compute_5[cse_var_125] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 12)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1792)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_126: int32 = (cse_var_1 + 125)
+ compute_5[cse_var_126] = (compute_5[cse_var_126] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 13)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1792)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_127: int32 = (cse_var_1 + 126)
+ compute_5[cse_var_127] = (compute_5[cse_var_127] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 14)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1792)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_128: int32 = (cse_var_1 + 127)
+ compute_5[cse_var_128] = (compute_5[cse_var_128] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 15)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 1792)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_129: int32 = (cse_var_1 + 128)
+ compute_5[cse_var_129] = (compute_5[cse_var_129] + (placeholder_1[((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16))]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2048)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_130: int32 = (cse_var_1 + 129)
+ compute_5[cse_var_130] = (compute_5[cse_var_130] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 1)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2048)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_131: int32 = (cse_var_1 + 130)
+ compute_5[cse_var_131] = (compute_5[cse_var_131] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 2)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2048)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_132: int32 = (cse_var_1 + 131)
+ compute_5[cse_var_132] = (compute_5[cse_var_132] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 3)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2048)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_133: int32 = (cse_var_1 + 132)
+ compute_5[cse_var_133] = (compute_5[cse_var_133] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 4)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2048)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_134: int32 = (cse_var_1 + 133)
+ compute_5[cse_var_134] = (compute_5[cse_var_134] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 5)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2048)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_135: int32 = (cse_var_1 + 134)
+ compute_5[cse_var_135] = (compute_5[cse_var_135] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 6)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2048)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_136: int32 = (cse_var_1 + 135)
+ compute_5[cse_var_136] = (compute_5[cse_var_136] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 7)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2048)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_137: int32 = (cse_var_1 + 136)
+ compute_5[cse_var_137] = (compute_5[cse_var_137] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 8)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2048)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_138: int32 = (cse_var_1 + 137)
+ compute_5[cse_var_138] = (compute_5[cse_var_138] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 9)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2048)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_139: int32 = (cse_var_1 + 138)
+ compute_5[cse_var_139] = (compute_5[cse_var_139] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 10)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2048)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_140: int32 = (cse_var_1 + 139)
+ compute_5[cse_var_140] = (compute_5[cse_var_140] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 11)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2048)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_141: int32 = (cse_var_1 + 140)
+ compute_5[cse_var_141] = (compute_5[cse_var_141] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 12)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2048)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_142: int32 = (cse_var_1 + 141)
+ compute_5[cse_var_142] = (compute_5[cse_var_142] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 13)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2048)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_143: int32 = (cse_var_1 + 142)
+ compute_5[cse_var_143] = (compute_5[cse_var_143] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 14)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2048)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_144: int32 = (cse_var_1 + 143)
+ compute_5[cse_var_144] = (compute_5[cse_var_144] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 15)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2048)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_145: int32 = (cse_var_1 + 144)
+ compute_5[cse_var_145] = (compute_5[cse_var_145] + (placeholder_1[((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16))]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2304)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_146: int32 = (cse_var_1 + 145)
+ compute_5[cse_var_146] = (compute_5[cse_var_146] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 1)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2304)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_147: int32 = (cse_var_1 + 146)
+ compute_5[cse_var_147] = (compute_5[cse_var_147] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 2)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2304)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_148: int32 = (cse_var_1 + 147)
+ compute_5[cse_var_148] = (compute_5[cse_var_148] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 3)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2304)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_149: int32 = (cse_var_1 + 148)
+ compute_5[cse_var_149] = (compute_5[cse_var_149] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 4)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2304)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_150: int32 = (cse_var_1 + 149)
+ compute_5[cse_var_150] = (compute_5[cse_var_150] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 5)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2304)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_151: int32 = (cse_var_1 + 150)
+ compute_5[cse_var_151] = (compute_5[cse_var_151] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 6)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2304)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_152: int32 = (cse_var_1 + 151)
+ compute_5[cse_var_152] = (compute_5[cse_var_152] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 7)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2304)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_153: int32 = (cse_var_1 + 152)
+ compute_5[cse_var_153] = (compute_5[cse_var_153] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 8)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2304)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_154: int32 = (cse_var_1 + 153)
+ compute_5[cse_var_154] = (compute_5[cse_var_154] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 9)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2304)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_155: int32 = (cse_var_1 + 154)
+ compute_5[cse_var_155] = (compute_5[cse_var_155] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 10)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2304)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_156: int32 = (cse_var_1 + 155)
+ compute_5[cse_var_156] = (compute_5[cse_var_156] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 11)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2304)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_157: int32 = (cse_var_1 + 156)
+ compute_5[cse_var_157] = (compute_5[cse_var_157] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 12)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2304)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_158: int32 = (cse_var_1 + 157)
+ compute_5[cse_var_158] = (compute_5[cse_var_158] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 13)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2304)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_159: int32 = (cse_var_1 + 158)
+ compute_5[cse_var_159] = (compute_5[cse_var_159] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 14)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2304)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_160: int32 = (cse_var_1 + 159)
+ compute_5[cse_var_160] = (compute_5[cse_var_160] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 15)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2304)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_161: int32 = (cse_var_1 + 160)
+ compute_5[cse_var_161] = (compute_5[cse_var_161] + (placeholder_1[((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16))]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2560)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_162: int32 = (cse_var_1 + 161)
+ compute_5[cse_var_162] = (compute_5[cse_var_162] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 1)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2560)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_163: int32 = (cse_var_1 + 162)
+ compute_5[cse_var_163] = (compute_5[cse_var_163] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 2)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2560)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_164: int32 = (cse_var_1 + 163)
+ compute_5[cse_var_164] = (compute_5[cse_var_164] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 3)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2560)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_165: int32 = (cse_var_1 + 164)
+ compute_5[cse_var_165] = (compute_5[cse_var_165] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 4)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2560)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_166: int32 = (cse_var_1 + 165)
+ compute_5[cse_var_166] = (compute_5[cse_var_166] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 5)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2560)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_167: int32 = (cse_var_1 + 166)
+ compute_5[cse_var_167] = (compute_5[cse_var_167] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 6)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2560)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_168: int32 = (cse_var_1 + 167)
+ compute_5[cse_var_168] = (compute_5[cse_var_168] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 7)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2560)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_169: int32 = (cse_var_1 + 168)
+ compute_5[cse_var_169] = (compute_5[cse_var_169] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 8)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2560)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_170: int32 = (cse_var_1 + 169)
+ compute_5[cse_var_170] = (compute_5[cse_var_170] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 9)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2560)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_171: int32 = (cse_var_1 + 170)
+ compute_5[cse_var_171] = (compute_5[cse_var_171] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 10)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2560)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_172: int32 = (cse_var_1 + 171)
+ compute_5[cse_var_172] = (compute_5[cse_var_172] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 11)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2560)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_173: int32 = (cse_var_1 + 172)
+ compute_5[cse_var_173] = (compute_5[cse_var_173] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 12)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2560)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_174: int32 = (cse_var_1 + 173)
+ compute_5[cse_var_174] = (compute_5[cse_var_174] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 13)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2560)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_175: int32 = (cse_var_1 + 174)
+ compute_5[cse_var_175] = (compute_5[cse_var_175] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 14)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2560)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_176: int32 = (cse_var_1 + 175)
+ compute_5[cse_var_176] = (compute_5[cse_var_176] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 15)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2560)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_177: int32 = (cse_var_1 + 176)
+ compute_5[cse_var_177] = (compute_5[cse_var_177] + (placeholder_1[((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16))]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2816)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_178: int32 = (cse_var_1 + 177)
+ compute_5[cse_var_178] = (compute_5[cse_var_178] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 1)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2816)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_179: int32 = (cse_var_1 + 178)
+ compute_5[cse_var_179] = (compute_5[cse_var_179] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 2)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2816)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_180: int32 = (cse_var_1 + 179)
+ compute_5[cse_var_180] = (compute_5[cse_var_180] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 3)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2816)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_181: int32 = (cse_var_1 + 180)
+ compute_5[cse_var_181] = (compute_5[cse_var_181] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 4)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2816)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_182: int32 = (cse_var_1 + 181)
+ compute_5[cse_var_182] = (compute_5[cse_var_182] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 5)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2816)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_183: int32 = (cse_var_1 + 182)
+ compute_5[cse_var_183] = (compute_5[cse_var_183] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 6)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2816)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_184: int32 = (cse_var_1 + 183)
+ compute_5[cse_var_184] = (compute_5[cse_var_184] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 7)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2816)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_185: int32 = (cse_var_1 + 184)
+ compute_5[cse_var_185] = (compute_5[cse_var_185] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 8)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2816)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_186: int32 = (cse_var_1 + 185)
+ compute_5[cse_var_186] = (compute_5[cse_var_186] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 9)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2816)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_187: int32 = (cse_var_1 + 186)
+ compute_5[cse_var_187] = (compute_5[cse_var_187] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 10)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2816)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_188: int32 = (cse_var_1 + 187)
+ compute_5[cse_var_188] = (compute_5[cse_var_188] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 11)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2816)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_189: int32 = (cse_var_1 + 188)
+ compute_5[cse_var_189] = (compute_5[cse_var_189] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 12)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2816)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_190: int32 = (cse_var_1 + 189)
+ compute_5[cse_var_190] = (compute_5[cse_var_190] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 13)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2816)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_191: int32 = (cse_var_1 + 190)
+ compute_5[cse_var_191] = (compute_5[cse_var_191] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 14)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2816)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_192: int32 = (cse_var_1 + 191)
+ compute_5[cse_var_192] = (compute_5[cse_var_192] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 15)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 2816)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_193: int32 = (cse_var_1 + 192)
+ compute_5[cse_var_193] = (compute_5[cse_var_193] + (placeholder_1[((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16))]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3072)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_194: int32 = (cse_var_1 + 193)
+ compute_5[cse_var_194] = (compute_5[cse_var_194] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 1)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3072)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_195: int32 = (cse_var_1 + 194)
+ compute_5[cse_var_195] = (compute_5[cse_var_195] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 2)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3072)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_196: int32 = (cse_var_1 + 195)
+ compute_5[cse_var_196] = (compute_5[cse_var_196] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 3)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3072)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_197: int32 = (cse_var_1 + 196)
+ compute_5[cse_var_197] = (compute_5[cse_var_197] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 4)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3072)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_198: int32 = (cse_var_1 + 197)
+ compute_5[cse_var_198] = (compute_5[cse_var_198] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 5)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3072)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_199: int32 = (cse_var_1 + 198)
+ compute_5[cse_var_199] = (compute_5[cse_var_199] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 6)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3072)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_200: int32 = (cse_var_1 + 199)
+ compute_5[cse_var_200] = (compute_5[cse_var_200] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 7)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3072)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_201: int32 = (cse_var_1 + 200)
+ compute_5[cse_var_201] = (compute_5[cse_var_201] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 8)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3072)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_202: int32 = (cse_var_1 + 201)
+ compute_5[cse_var_202] = (compute_5[cse_var_202] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 9)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3072)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_203: int32 = (cse_var_1 + 202)
+ compute_5[cse_var_203] = (compute_5[cse_var_203] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 10)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3072)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_204: int32 = (cse_var_1 + 203)
+ compute_5[cse_var_204] = (compute_5[cse_var_204] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 11)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3072)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_205: int32 = (cse_var_1 + 204)
+ compute_5[cse_var_205] = (compute_5[cse_var_205] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 12)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3072)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_206: int32 = (cse_var_1 + 205)
+ compute_5[cse_var_206] = (compute_5[cse_var_206] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 13)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3072)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_207: int32 = (cse_var_1 + 206)
+ compute_5[cse_var_207] = (compute_5[cse_var_207] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 14)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3072)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_208: int32 = (cse_var_1 + 207)
+ compute_5[cse_var_208] = (compute_5[cse_var_208] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 15)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3072)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_209: int32 = (cse_var_1 + 208)
+ compute_5[cse_var_209] = (compute_5[cse_var_209] + (placeholder_1[((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16))]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3328)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_210: int32 = (cse_var_1 + 209)
+ compute_5[cse_var_210] = (compute_5[cse_var_210] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 1)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3328)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_211: int32 = (cse_var_1 + 210)
+ compute_5[cse_var_211] = (compute_5[cse_var_211] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 2)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3328)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_212: int32 = (cse_var_1 + 211)
+ compute_5[cse_var_212] = (compute_5[cse_var_212] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 3)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3328)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_213: int32 = (cse_var_1 + 212)
+ compute_5[cse_var_213] = (compute_5[cse_var_213] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 4)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3328)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_214: int32 = (cse_var_1 + 213)
+ compute_5[cse_var_214] = (compute_5[cse_var_214] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 5)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3328)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_215: int32 = (cse_var_1 + 214)
+ compute_5[cse_var_215] = (compute_5[cse_var_215] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 6)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3328)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_216: int32 = (cse_var_1 + 215)
+ compute_5[cse_var_216] = (compute_5[cse_var_216] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 7)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3328)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_217: int32 = (cse_var_1 + 216)
+ compute_5[cse_var_217] = (compute_5[cse_var_217] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 8)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3328)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_218: int32 = (cse_var_1 + 217)
+ compute_5[cse_var_218] = (compute_5[cse_var_218] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 9)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3328)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_219: int32 = (cse_var_1 + 218)
+ compute_5[cse_var_219] = (compute_5[cse_var_219] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 10)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3328)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_220: int32 = (cse_var_1 + 219)
+ compute_5[cse_var_220] = (compute_5[cse_var_220] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 11)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3328)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_221: int32 = (cse_var_1 + 220)
+ compute_5[cse_var_221] = (compute_5[cse_var_221] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 12)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3328)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_222: int32 = (cse_var_1 + 221)
+ compute_5[cse_var_222] = (compute_5[cse_var_222] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 13)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3328)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_223: int32 = (cse_var_1 + 222)
+ compute_5[cse_var_223] = (compute_5[cse_var_223] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 14)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3328)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_224: int32 = (cse_var_1 + 223)
+ compute_5[cse_var_224] = (compute_5[cse_var_224] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 15)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3328)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_225: int32 = (cse_var_1 + 224)
+ compute_5[cse_var_225] = (compute_5[cse_var_225] + (placeholder_1[((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16))]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3584)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_226: int32 = (cse_var_1 + 225)
+ compute_5[cse_var_226] = (compute_5[cse_var_226] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 1)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3584)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_227: int32 = (cse_var_1 + 226)
+ compute_5[cse_var_227] = (compute_5[cse_var_227] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 2)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3584)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_228: int32 = (cse_var_1 + 227)
+ compute_5[cse_var_228] = (compute_5[cse_var_228] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 3)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3584)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_229: int32 = (cse_var_1 + 228)
+ compute_5[cse_var_229] = (compute_5[cse_var_229] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 4)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3584)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_230: int32 = (cse_var_1 + 229)
+ compute_5[cse_var_230] = (compute_5[cse_var_230] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 5)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3584)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_231: int32 = (cse_var_1 + 230)
+ compute_5[cse_var_231] = (compute_5[cse_var_231] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 6)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3584)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_232: int32 = (cse_var_1 + 231)
+ compute_5[cse_var_232] = (compute_5[cse_var_232] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 7)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3584)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_233: int32 = (cse_var_1 + 232)
+ compute_5[cse_var_233] = (compute_5[cse_var_233] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 8)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3584)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_234: int32 = (cse_var_1 + 233)
+ compute_5[cse_var_234] = (compute_5[cse_var_234] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 9)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3584)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_235: int32 = (cse_var_1 + 234)
+ compute_5[cse_var_235] = (compute_5[cse_var_235] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 10)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3584)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_236: int32 = (cse_var_1 + 235)
+ compute_5[cse_var_236] = (compute_5[cse_var_236] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 11)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3584)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_237: int32 = (cse_var_1 + 236)
+ compute_5[cse_var_237] = (compute_5[cse_var_237] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 12)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3584)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_238: int32 = (cse_var_1 + 237)
+ compute_5[cse_var_238] = (compute_5[cse_var_238] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 13)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3584)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_239: int32 = (cse_var_1 + 238)
+ compute_5[cse_var_239] = (compute_5[cse_var_239] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 14)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3584)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_240: int32 = (cse_var_1 + 239)
+ compute_5[cse_var_240] = (compute_5[cse_var_240] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 15)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3584)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_241: int32 = (cse_var_1 + 240)
+ compute_5[cse_var_241] = (compute_5[cse_var_241] + (placeholder_1[((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16))]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3840)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_242: int32 = (cse_var_1 + 241)
+ compute_5[cse_var_242] = (compute_5[cse_var_242] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 1)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3840)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_243: int32 = (cse_var_1 + 242)
+ compute_5[cse_var_243] = (compute_5[cse_var_243] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 2)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3840)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_244: int32 = (cse_var_1 + 243)
+ compute_5[cse_var_244] = (compute_5[cse_var_244] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 3)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3840)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_245: int32 = (cse_var_1 + 244)
+ compute_5[cse_var_245] = (compute_5[cse_var_245] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 4)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3840)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_246: int32 = (cse_var_1 + 245)
+ compute_5[cse_var_246] = (compute_5[cse_var_246] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 5)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3840)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_247: int32 = (cse_var_1 + 246)
+ compute_5[cse_var_247] = (compute_5[cse_var_247] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 6)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3840)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_248: int32 = (cse_var_1 + 247)
+ compute_5[cse_var_248] = (compute_5[cse_var_248] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 7)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3840)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_249: int32 = (cse_var_1 + 248)
+ compute_5[cse_var_249] = (compute_5[cse_var_249] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 8)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3840)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_250: int32 = (cse_var_1 + 249)
+ compute_5[cse_var_250] = (compute_5[cse_var_250] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 9)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3840)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_251: int32 = (cse_var_1 + 250)
+ compute_5[cse_var_251] = (compute_5[cse_var_251] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 10)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3840)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_252: int32 = (cse_var_1 + 251)
+ compute_5[cse_var_252] = (compute_5[cse_var_252] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 11)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3840)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_253: int32 = (cse_var_1 + 252)
+ compute_5[cse_var_253] = (compute_5[cse_var_253] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 12)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3840)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_254: int32 = (cse_var_1 + 253)
+ compute_5[cse_var_254] = (compute_5[cse_var_254] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 13)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3840)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_255: int32 = (cse_var_1 + 254)
+ compute_5[cse_var_255] = (compute_5[cse_var_255] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 14)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3840)], 0f32)))
+ }
+ if @tir.likely((elem_idx < (placeholder_3[(i0.outer.i1.outer.fused + 1)] - placeholder_3[i0.outer.i1.outer.fused])), dtype=bool) {
+ let cse_var_256: int32 = (cse_var_1 + 255)
+ compute_5[cse_var_256] = (compute_5[cse_var_256] + (placeholder_1[(((placeholder_3[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + 15)]*max(placeholder[(((i.outer.inner*4096) + placeholder_2[(placeholder_3[i0.outer.i1.outer.fused] + elem_idx)]) + 3840)], 0f32)))
}
}
}
}
for (i0.inner: int32, 0, 128) {
- for (i1.inner: int32, 0, 32) {
- let cse_var_4: int32 = (((i0.inner*512) + (i0.outer.i1.outer.fused*32)) + i1.inner)
- compute[cse_var_4] = max((compute_5[((i0.inner*32) + i1.inner)] + placeholder_4[cse_var_4]), 0f32)
- }
+ let cse_var_257: int32 = ((i0.inner*512) + (i0.outer.i1.outer.fused*16))
+ compute[ramp(cse_var_257, 1, 16)] = max((compute_5[ramp((i0.inner*16), 1, 16)] + placeholder_4[ramp(cse_var_257, 1, 16)]), broadcast(0f32, 16))
}
}
}
@@ -683,7 +1949,7 @@ layout transformation, parallelization, vectorization, unrolling, and operator f
<span class="p">)</span>
</pre></div>
</div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Execution time of this operator: 1.443 ms
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Execution time of this operator: 3.119 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 bcb2bd228..05b15cb7c 100644
--- a/docs/how_to/tune_with_autotvm/sg_execution_times.html
+++ b/docs/how_to/tune_with_autotvm/sg_execution_times.html
@@ -322,7 +322,7 @@
<div class="section" id="computation-times">
<span id="sphx-glr-how-to-tune-with-autotvm-sg-execution-times"></span><h1>Computation times<a class="headerlink" href="#computation-times" title="Permalink to this headline">¶</a></h1>
-<p><strong>00:43.512</strong> total execution time for <strong>how_to_tune_with_autotvm</strong> files:</p>
+<p><strong>00:43.808</strong> total execution time for <strong>how_to_tune_with_autotvm</strong> files:</p>
<table class="docutils align-default">
<colgroup>
<col style="width: 84%" />
@@ -331,7 +331,7 @@
</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>00:43.476</p></td>
+<td><p>00:43.772</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>
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 48d6598f1..6552b04d4 100644
--- a/docs/how_to/tune_with_autotvm/tune_conv2d_cuda.html
+++ b/docs/how_to/tune_with_autotvm/tune_conv2d_cuda.html
@@ -1167,8 +1167,8 @@ Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 871, 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, 32]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 1, 128]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,2885496
-No: 6 GFLOPS: 42.45/42.45 result: MeasureResult(costs=(0.005453405210526316,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.6411683559417725, timestamp=1657505886.9299095) [('tile_f', [-1, 1, 1, 1]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 4, 4]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,3754080
-No: 7 GFLOPS: 0.00/42.45 result: Traceback (most recent call last):
+No: 6 GFLOPS: 104.00/104.00 result: MeasureResult(costs=(0.002225995604166667,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.636479139328003, timestamp=1657533659.40923) [('tile_f', [-1, 1, 1, 1]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 4, 4]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,3754080
+No: 7 GFLOPS: 0.00/104.00 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 588, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 540, in _build_func_common
@@ -1291,7 +1291,7 @@ Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 871, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 1, 16, 32]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 256, 1]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,6225319
-No: 8 GFLOPS: 0.00/42.45 result: Traceback (most recent call last):
+No: 8 GFLOPS: 0.00/104.00 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 588, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 540, in _build_func_common
@@ -1414,7 +1414,7 @@ Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 871, 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, 32]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 8, 64]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,943546
-No: 9 GFLOPS: 0.00/42.45 result: Traceback (most recent call last):
+No: 9 GFLOPS: 0.00/104.00 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 588, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 540, in _build_func_common
@@ -1537,7 +1537,7 @@ Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 871, 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, 4]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 16, 32]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,2868708
-No: 10 GFLOPS: 0.00/42.45 result: Traceback (most recent call last):
+No: 10 GFLOPS: 0.00/104.00 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 142, in build
res = future.result()
File "/usr/lib/python3.7/concurrent/futures/_base.py", line 435, in result
@@ -1555,7 +1555,7 @@ No: 10 GFLOPS: 0.00/42.45 result: Traceback (most recent call last):
TimeoutError
[('tile_f', [-1, 32, 2, 4]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 4, 2]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,4691833
-No: 11 GFLOPS: 0.00/42.45 result: Traceback (most recent call last):
+No: 11 GFLOPS: 0.00/104.00 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 588, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 540, in _build_func_common
@@ -1678,7 +1678,7 @@ Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 871, 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, 64]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 4, 4]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,1042124
-No: 12 GFLOPS: 0.00/42.45 result: Traceback (most recent call last):
+No: 12 GFLOPS: 0.00/104.00 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 588, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 540, in _build_func_common
@@ -1801,7 +1801,7 @@ Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 871, 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, 4]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 32, 16]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,10013405
-No: 13 GFLOPS: 0.00/42.45 result: Traceback (most recent call last):
+No: 13 GFLOPS: 0.00/104.00 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 588, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 540, in _build_func_common
@@ -1924,7 +1924,7 @@ Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 871, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 8, 8, 2]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 4, 32]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,6732082
-No: 14 GFLOPS: 0.00/42.45 result: Traceback (most recent call last):
+No: 14 GFLOPS: 0.00/104.00 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 588, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 540, in _build_func_common
@@ -2047,7 +2047,7 @@ Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 871, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 4, 32]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 4, 128]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,7536735
-No: 15 GFLOPS: 0.00/42.45 result: Traceback (most recent call last):
+No: 15 GFLOPS: 0.00/104.00 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 588, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 540, in _build_func_common
@@ -2170,7 +2170,7 @@ Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 871, 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, 4]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 128, 4]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,482121
-No: 16 GFLOPS: 0.00/42.45 result: Traceback (most recent call last):
+No: 16 GFLOPS: 0.00/104.00 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 588, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 540, in _build_func_common
@@ -2293,7 +2293,7 @@ Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 871, 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, 16]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 32, 8]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,2824525
-No: 17 GFLOPS: 0.00/42.45 result: Traceback (most recent call last):
+No: 17 GFLOPS: 0.00/104.00 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 588, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 540, in _build_func_common
@@ -2416,7 +2416,7 @@ Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 871, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 64, 1, 1]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 8, 8]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,4559286
-No: 18 GFLOPS: 0.00/42.45 result: Traceback (most recent call last):
+No: 18 GFLOPS: 0.00/104.00 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 588, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 540, in _build_func_common
@@ -2539,7 +2539,7 @@ Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 871, 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, 32, 16]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 1, 512]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,9677544
-No: 19 GFLOPS: 0.00/42.45 result: Traceback (most recent call last):
+No: 19 GFLOPS: 0.00/104.00 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 738, in __call__
yield remote, remote.load_module(os.path.split(build_result.filename)[1])
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 702, in run_through_rpc
@@ -2627,7 +2627,7 @@ tvm._ffi.base.TVMError: Traceback (most recent call last):
15: _PyEval_EvalFrameDefault
14: 0x0000000000537c30
13: _PyObject_FastCallKeywords
- 12: 0x00007ff304e22fa2
+ 12: 0x00007f0ddc2e4fa2
11: _ctypes_callproc
10: ffi_call
9: ffi_call_unix64
@@ -2692,7 +2692,7 @@ Traceback (most recent call last):
21: _PyFunction_FastCallKeywords
20: _PyEval_EvalFrameDefault
19: _PyFunction_FastCall [('tile_f', [-1, 8, 2, 16]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 1, 1]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,6390073
-No: 20 GFLOPS: 142.88/142.88 result: MeasureResult(costs=(0.00162021567,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.442824363708496, timestamp=1657505912.7780535) [('tile_f', [-1, 1, 4, 1]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 4, 1]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,9881539
+No: 20 GFLOPS: 143.59/143.59 result: MeasureResult(costs=(0.00161221691,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.4332330226898193, timestamp=1657533685.933291) [('tile_f', [-1, 1, 4, 1]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 4, 1]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,9881539
</pre></div>
</div>
<p>Finally we can inspect the best config from log file, check correctness,
@@ -2733,7 +2733,7 @@ and measure running time.</p>
Best config:
[('tile_f', [-1, 1, 4, 1]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 4, 1]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,9881539
Finish loading 20 records
-Time cost of this operator: 0.002013
+Time cost of this operator: 0.001964
</pre></div>
</div>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-tune-with-autotvm-tune-conv2d-cuda-py">
diff --git a/docs/how_to/work_with_microtvm/micro_autotune.html b/docs/how_to/work_with_microtvm/micro_autotune.html
index 8ec38124c..e514ad729 100644
--- a/docs/how_to/work_with_microtvm/micro_autotune.html
+++ b/docs/how_to/work_with_microtvm/micro_autotune.html
@@ -578,10 +578,10 @@ the tuned operator.</p>
########## Build without Autotuning ##########
Node Name Ops Time(us) Time(%) Shape Inputs Outputs Measurements(us)
--------- --- -------- ------- ----- ------ ------- ----------------
-tvmgen_default_fused_nn_contrib_conv2d_NCHWc tvmgen_default_fused_nn_contrib_conv2d_NCHWc 315.1 98.729 (1, 2, 10, 10, 3) 2 1 [315.1]
-tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 3.064 0.96 (1, 6, 10, 10) 1 1 [3.064]
-tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.994 0.311 (1, 1, 10, 10, 3) 1 1 [0.994]
-Total_time - 319.158 - - - - -
+tvmgen_default_fused_nn_contrib_conv2d_NCHWc tvmgen_default_fused_nn_contrib_conv2d_NCHWc 308.2 98.734 (1, 2, 10, 10, 3) 2 1 [308.2]
+tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 2.983 0.956 (1, 6, 10, 10) 1 1 [2.983]
+tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.969 0.31 (1, 1, 10, 10, 3) 1 1 [0.969]
+Total_time - 312.152 - - - - -
</pre></div>
</div>
</div>
@@ -634,10 +634,10 @@ Total_time -
########## 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 209.7 98.55 (1, 1, 10, 10, 6) 2 1 [209.7]
-tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 2.21 1.039 (1, 6, 10, 10) 1 1 [2.21]
-tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.875 0.411 (1, 3, 10, 10, 1) 1 1 [0.875]
-Total_time - 212.785 - - - - -
+tvmgen_default_fused_nn_contrib_conv2d_NCHWc tvmgen_default_fused_nn_contrib_conv2d_NCHWc 79.562 96.684 (1, 6, 10, 10, 1) 2 1 [79.562]
+tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 1.713 2.082 (1, 6, 10, 10) 1 1 [1.713]
+tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 1.015 1.234 (1, 1, 10, 10, 3) 1 1 [1.015]
+Total_time - 82.291 - - - - -
</pre></div>
</div>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-work-with-microtvm-micro-autotune-py">
diff --git a/docs/how_to/work_with_microtvm/micro_train.html b/docs/how_to/work_with_microtvm/micro_train.html
index 509976b09..92976ef84 100644
--- a/docs/how_to/work_with_microtvm/micro_train.html
+++ b/docs/how_to/work_with_microtvm/micro_train.html
@@ -510,7 +510,7 @@ take about <strong>2 minutes</strong> to download the Stanford Cars, while COCO
<a href="https://docs.python.org/3/library/shutil.html#shutil.move" title="shutil.move" class="sphx-glr-backref-module-shutil sphx-glr-backref-type-py-function"><span class="n">shutil</span><span class="o">.</span><span class="n">move</span></a><span class="p">(</span><span class="sa">f</span><span class="s2">"</span><span class="si">{</span><a href="https://docs.python.org/3/library/stdtypes.html#str" title="builtins.str" class="sphx-glr-backref-module-builtins sphx-glr-backref-typ [...]
</pre></div>
</div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>'/tmp/tmpcd8odi49/images/random'
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>'/tmp/tmpj7dioiiz/images/random'
</pre></div>
</div>
</div>
@@ -570,8 +570,8 @@ objects to other stuff? We can display some examples from our datasets using <co
<span class="n">plt</span><span class="o">.</span><span class="n">axis</span><span class="p">(</span><span class="s2">"off"</span><span class="p">)</span>
</pre></div>
</div>
-<img src="../../_images/sphx_glr_micro_train_001.png" srcset="../../_images/sphx_glr_micro_train_001.png" alt="[1.0, 0.0], [1.0, 0.0], [1.0, 0.0], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0], [1.0, 0.0], [0.0, 1.0], [1.0, 0.0]" class = "sphx-glr-single-img"/><div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>/tmp/tmpcd8odi49/images/target contains 8144 images
-/tmp/tmpcd8odi49/images/random contains 5000 images
+<img src="../../_images/sphx_glr_micro_train_001.png" srcset="../../_images/sphx_glr_micro_train_001.png" alt="[1.0, 0.0], [1.0, 0.0], [1.0, 0.0], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0], [1.0, 0.0], [0.0, 1.0], [1.0, 0.0]" class = "sphx-glr-single-img"/><div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>/tmp/tmpj7dioiiz/images/target contains 8144 images
+/tmp/tmpj7dioiiz/images/random contains 5000 images
</pre></div>
</div>
</div>
@@ -683,13 +683,13 @@ the time on our validation set).</p>
</pre></div>
</div>
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Epoch 1/3
-328/328 - 55s - loss: 0.2250 - accuracy: 0.9231 - val_loss: 0.1352 - val_accuracy: 0.9588
+328/328 - 53s - loss: 0.2160 - accuracy: 0.9251 - val_loss: 0.1530 - val_accuracy: 0.9494
Epoch 2/3
-328/328 - 53s - loss: 0.0951 - accuracy: 0.9639 - val_loss: 0.1336 - val_accuracy: 0.9622
+328/328 - 50s - loss: 0.0966 - accuracy: 0.9638 - val_loss: 0.1054 - val_accuracy: 0.9641
Epoch 3/3
-328/328 - 53s - loss: 0.0654 - accuracy: 0.9757 - val_loss: 0.1372 - val_accuracy: 0.9558
+328/328 - 50s - loss: 0.0623 - accuracy: 0.9769 - val_loss: 0.1160 - val_accuracy: 0.9637
-<keras.callbacks.History object at 0x7f486eff1790>
+<keras.callbacks.History object at 0x7fd72d0ebe10>
</pre></div>
</div>
</div>
@@ -951,7 +951,7 @@ as intended.</p>
<p>From here, we could modify the model to read live images from the camera - we have another
Arduino tutorial for how to do that <a class="reference external" href="https://github.com/guberti/tvm-arduino-demos/tree/master/examples/person_detection">on GitHub</a>. Alternatively, we could also
<a class="reference external" href="https://tvm.apache.org/docs/how_to/work_with_microtvm/micro_autotune.html">use TVM’s autotuning capabilities</a> to dramatically improve the model’s performance.</p>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 4 minutes 51.961 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 4 minutes 50.207 seconds)</p>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-work-with-microtvm-micro-train-py">
<div class="sphx-glr-download sphx-glr-download-python docutils container">
<p><a class="reference download internal" download="" href="../../_downloads/b52cec46baf4f78d6bcd94cbe269c8a6/micro_train.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">micro_train.py</span></code></a></p>
diff --git a/docs/how_to/work_with_microtvm/sg_execution_times.html b/docs/how_to/work_with_microtvm/sg_execution_times.html
index 59636e6c8..c0815f7d9 100644
--- a/docs/how_to/work_with_microtvm/sg_execution_times.html
+++ b/docs/how_to/work_with_microtvm/sg_execution_times.html
@@ -322,7 +322,7 @@
<div class="section" id="computation-times">
<span id="sphx-glr-how-to-work-with-microtvm-sg-execution-times"></span><h1>Computation times<a class="headerlink" href="#computation-times" title="Permalink to this headline">¶</a></h1>
-<p><strong>05:39.131</strong> total execution time for <strong>how_to_work_with_microtvm</strong> files:</p>
+<p><strong>05:36.322</strong> total execution time for <strong>how_to_work_with_microtvm</strong> files:</p>
<table class="docutils align-default">
<colgroup>
<col style="width: 83%" />
@@ -331,15 +331,15 @@
</colgroup>
<tbody>
<tr class="row-odd"><td><p><a class="reference internal" href="micro_train.html#sphx-glr-how-to-work-with-microtvm-micro-train-py"><span class="std std-ref">Training Vision Models for microTVM on Arduino</span></a> (<code class="docutils literal notranslate"><span class="pre">micro_train.py</span></code>)</p></td>
-<td><p>04:51.961</p></td>
+<td><p>04:50.207</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="micro_autotune.html#sphx-glr-how-to-work-with-microtvm-micro-autotune-py"><span class="std std-ref">Autotuning with microTVM</span></a> (<code class="docutils literal notranslate"><span class="pre">micro_autotune.py</span></code>)</p></td>
-<td><p>00:43.801</p></td>
+<td><p>00:42.750</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="micro_tflite.html#sphx-glr-how-to-work-with-microtvm-micro-tflite-py"><span class="std std-ref">microTVM with TFLite Models</span></a> (<code class="docutils literal notranslate"><span class="pre">micro_tflite.py</span></code>)</p></td>
-<td><p>00:03.368</p></td>
+<td><p>00:03.362</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="micro_ethosu.html#sphx-glr-how-to-work-with-microtvm-micro-ethosu-py"><span class="std std-ref">Running TVM on bare metal Arm(R) Cortex(R)-M55 CPU and Ethos(TM)-U55 NPU with CMSIS-NN</span></a> (<code class="docutils literal notranslate"><span class="pre">micro_ethosu.py</span></code>)</p></td>
diff --git a/docs/how_to/work_with_relay/sg_execution_times.html b/docs/how_to/work_with_relay/sg_execution_times.html
index c267b2ff5..feec505c6 100644
--- a/docs/how_to/work_with_relay/sg_execution_times.html
+++ b/docs/how_to/work_with_relay/sg_execution_times.html
@@ -322,7 +322,7 @@
<div class="section" id="computation-times">
<span id="sphx-glr-how-to-work-with-relay-sg-execution-times"></span><h1>Computation times<a class="headerlink" href="#computation-times" title="Permalink to this headline">¶</a></h1>
-<p><strong>00:11.307</strong> total execution time for <strong>how_to_work_with_relay</strong> files:</p>
+<p><strong>00:11.582</strong> total execution time for <strong>how_to_work_with_relay</strong> files:</p>
<table class="docutils align-default">
<colgroup>
<col style="width: 83%" />
@@ -331,11 +331,11 @@
</colgroup>
<tbody>
<tr class="row-odd"><td><p><a class="reference internal" href="using_external_lib.html#sphx-glr-how-to-work-with-relay-using-external-lib-py"><span class="std std-ref">Using External Libraries in Relay</span></a> (<code class="docutils literal notranslate"><span class="pre">using_external_lib.py</span></code>)</p></td>
-<td><p>00:09.789</p></td>
+<td><p>00:10.125</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="build_gcn.html#sphx-glr-how-to-work-with-relay-build-gcn-py"><span class="std std-ref">Building a Graph Convolutional Network</span></a> (<code class="docutils literal notranslate"><span class="pre">build_gcn.py</span></code>)</p></td>
-<td><p>00:01.512</p></td>
+<td><p>00:01.451</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="using_relay_viz.html#sphx-glr-how-to-work-with-relay-using-relay-viz-py"><span class="std std-ref">Use Relay Visualizer to Visualize Relay</span></a> (<code class="docutils literal notranslate"><span class="pre">using_relay_viz.py</span></code>)</p></td>
diff --git a/docs/how_to/work_with_schedules/intrin_math.html b/docs/how_to/work_with_schedules/intrin_math.html
index 24ac0d764..44807889c 100644
--- a/docs/how_to/work_with_schedules/intrin_math.html
+++ b/docs/how_to/work_with_schedules/intrin_math.html
@@ -517,7 +517,7 @@ The following example customizes CUDA lowering rule for <code class="code docuti
<a href="../../reference/api/python/ir.html#tvm.ir.register_intrin_lowering" title="tvm.ir.register_intrin_lowering" class="sphx-glr-backref-module-tvm-ir sphx-glr-backref-type-py-function"><span class="n">register_intrin_lowering</span></a><span class="p">(</span><span class="s2">"tir.exp"</span><span class="p">,</span> <span class="n">target</span><span class="o">=</span><span class="s2">"cuda"</span><span class="p">,</span> <span class="n">f</span><span class="o">= [...]
</pre></div>
</div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span><function my_cuda_math_rule at 0x7f4969c349e0>
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span><function my_cuda_math_rule at 0x7fd7381487a0>
</pre></div>
</div>
<p>Register the rule to TVM with override option to override existing rule.
diff --git a/docs/how_to/work_with_schedules/sg_execution_times.html b/docs/how_to/work_with_schedules/sg_execution_times.html
index cfbcbcc17..306b2e97c 100644
--- a/docs/how_to/work_with_schedules/sg_execution_times.html
+++ b/docs/how_to/work_with_schedules/sg_execution_times.html
@@ -322,7 +322,7 @@
<div class="section" id="computation-times">
<span id="sphx-glr-how-to-work-with-schedules-sg-execution-times"></span><h1>Computation times<a class="headerlink" href="#computation-times" title="Permalink to this headline">¶</a></h1>
-<p><strong>00:04.024</strong> total execution time for <strong>how_to_work_with_schedules</strong> files:</p>
+<p><strong>00:04.105</strong> total execution time for <strong>how_to_work_with_schedules</strong> files:</p>
<table class="docutils align-default">
<colgroup>
<col style="width: 83%" />
@@ -331,31 +331,31 @@
</colgroup>
<tbody>
<tr class="row-odd"><td><p><a class="reference internal" href="intrin_math.html#sphx-glr-how-to-work-with-schedules-intrin-math-py"><span class="std std-ref">Intrinsics and Math Functions</span></a> (<code class="docutils literal notranslate"><span class="pre">intrin_math.py</span></code>)</p></td>
-<td><p>00:01.875</p></td>
+<td><p>00:01.904</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="tensorize.html#sphx-glr-how-to-work-with-schedules-tensorize-py"><span class="std std-ref">Use Tensorize to Leverage Hardware Intrinsics</span></a> (<code class="docutils literal notranslate"><span class="pre">tensorize.py</span></code>)</p></td>
-<td><p>00:00.944</p></td>
+<td><p>00:00.973</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="reduction.html#sphx-glr-how-to-work-with-schedules-reduction-py"><span class="std std-ref">Reduction</span></a> (<code class="docutils literal notranslate"><span class="pre">reduction.py</span></code>)</p></td>
-<td><p>00:00.522</p></td>
+<td><p>00:00.531</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="scan.html#sphx-glr-how-to-work-with-schedules-scan-py"><span class="std std-ref">Scan and Recurrent Kernel</span></a> (<code class="docutils literal notranslate"><span class="pre">scan.py</span></code>)</p></td>
-<td><p>00:00.506</p></td>
+<td><p>00:00.517</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="extern_op.html#sphx-glr-how-to-work-with-schedules-extern-op-py"><span class="std std-ref">External Tensor Functions</span></a> (<code class="docutils literal notranslate"><span class="pre">extern_op.py</span></code>)</p></td>
-<td><p>00:00.100</p></td>
+<td><p>00:00.102</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="schedule_primitives.html#sphx-glr-how-to-work-with-schedules-schedule-primitives-py"><span class="std std-ref">Schedule Primitives in TVM</span></a> (<code class="docutils literal notranslate"><span class="pre">schedule_primitives.py</span></code>)</p></td>
-<td><p>00:00.035</p></td>
+<td><p>00:00.036</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="tedd.html#sphx-glr-how-to-work-with-schedules-tedd-py"><span class="std std-ref">Use Tensor Expression Debug Display (TEDD) for Visualization</span></a> (<code class="docutils literal notranslate"><span class="pre">tedd.py</span></code>)</p></td>
-<td><p>00:00.027</p></td>
+<td><p>00:00.028</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="tuple_inputs.html#sphx-glr-how-to-work-with-schedules-tuple-inputs-py"><span class="std std-ref">Compute and Reduce with Tuple Inputs</span></a> (<code class="docutils literal notranslate"><span class="pre">tuple_inputs.py</span></code>)</p></td>
diff --git a/docs/how_to/work_with_schedules/tensorize.html b/docs/how_to/work_with_schedules/tensorize.html
index 0b92b2956..ef2786835 100644
--- a/docs/how_to/work_with_schedules/tensorize.html
+++ b/docs/how_to/work_with_schedules/tensorize.html
@@ -572,7 +572,7 @@ The importing needs to happen before the tensorized GEMV being executed.</p>
C: Buffer(C_2: Pointer(float32), float32, [524288], [])}
buffer_map = {A_1: A, B_1: B, C_1: C}
preflattened_buffer_map = {A_1: A_3: Buffer(A_2, float32, [1024, 64], []), B_1: B_3: Buffer(B_2, float32, [512, 64], []), C_1: C_3: Buffer(C_2, float32, [1024, 512], [])} {
- attr [IterVar(i: int32, (nullptr), "DataPar", "")] "pragma_import_llvm" = "; ModuleID = '/tmp/tmpi6dlqhp0/input0.cc'\nsource_filename = \"/tmp/tmpi6dlqhp0/input0.cc\"\ntarget datalayout = \"e-m:e-i64:64-f80:128-n8:16:32:64-S128\"\ntarget triple = \"x86_64-pc-linux-gnu\"\n\n; Function Attrs: noinline nounwind optnone uwtable\ndefine dso_local i32 @gemv_update(float*, float*, float*, i32, i32, i32) #0 {\n %7 = allo [...]
+ attr [IterVar(i: int32, (nullptr), "DataPar", "")] "pragma_import_llvm" = "; ModuleID = '/tmp/tmp91ffoxoj/input0.cc'\nsource_filename = \"/tmp/tmp91ffoxoj/input0.cc\"\ntarget datalayout = \"e-m:e-i64:64-f80:128-n8:16:32:64-S128\"\ntarget triple = \"x86_64-pc-linux-gnu\"\n\n; Function Attrs: noinline nounwind optnone uwtable\ndefine dso_local i32 @gemv_update(float*, float*, float*, i32, i32, i32) #0 {\n %7 = allo [...]
for (i, 0, 1024) {
for (j.outer: int32, 0, 32) {
@tir.call_extern("gemv_update", @tir.tvm_access_ptr(@tir.type_annotation(, dtype=float32), C_2, ((i*512) + (j.outer*16)), 16, 2, dtype=handle), @tir.tvm_access_ptr(@tir.type_annotation(, dtype=float32), A_2, (i*64), 64, 1, dtype=handle), @tir.tvm_access_ptr(@tir.type_annotation(, dtype=float32), B_2, (j.outer*1024), 1024, 1, dtype=handle), 16, 64, 64, dtype=int32)
diff --git a/docs/install/from_source.html b/docs/install/from_source.html
index 217c51cc7..a7a0db46d 100644
--- a/docs/install/from_source.html
+++ b/docs/install/from_source.html
@@ -623,7 +623,7 @@ like <code class="docutils literal notranslate"><span class="pre">virtualenv</sp
<ul class="simple">
<li><p>If you want to use auto-tuning module</p></li>
</ul>
-<div class="highlight-bash notranslate"><div class="highlight"><pre><span></span>pip3 install --user tornado psutil xgboost cloudpickle
+<div class="highlight-bash notranslate"><div class="highlight"><pre><span></span>pip3 install --user tornado psutil <span class="s1">'xgboost<1.6.0'</span> cloudpickle
</pre></div>
</div>
</div></blockquote>
@@ -639,7 +639,7 @@ pip install pybind11 cython pythran
pip install scipy --no-use-pep517
-pip install xgboost
+pip install <span class="s1">'xgboost<1.6.0'</span>
</pre></div>
</div>
</div></blockquote>
diff --git a/docs/reference/api/python/auto_scheduler.html b/docs/reference/api/python/auto_scheduler.html
index 358b7a97b..9089e6bb2 100644
--- a/docs/reference/api/python/auto_scheduler.html
+++ b/docs/reference/api/python/auto_scheduler.html
@@ -1597,7 +1597,7 @@ history states as starting point to perform Evolutionary Search).</p></li>
... 2918 lines suppressed ...