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 2022/03/31 07:42:52 UTC
[tvm-site] branch asf-site updated: deploying docs (apache/tvm@42985c372681bc1915a6be1be917aa45f8f94a55)
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 3fabc37 deploying docs (apache/tvm@42985c372681bc1915a6be1be917aa45f8f94a55)
3fabc37 is described below
commit 3fabc372617eb4829d42d47304a2b89210ad7741
Author: tvm-bot <95...@users.noreply.github.com>
AuthorDate: Thu Mar 31 07:42:45 2022 +0000
deploying docs (apache/tvm@42985c372681bc1915a6be1be917aa45f8f94a55)
---
.../how_to/compile_models/from_mxnet.rst.txt | 2 +-
.../how_to/compile_models/from_paddle.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 | 20 +-
.../deploy_models/deploy_model_on_android.rst.txt | 2 +-
.../deploy_object_detection_pytorch.rst.txt | 11 +-
.../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 | 16 +-
.../extend_tvm/bring_your_own_datatypes.rst.txt | 2 +-
.../how_to/extend_tvm/sg_execution_times.rst.txt | 10 +-
.../how_to/extend_tvm/use_pass_instrument.rst.txt | 16 +-
.../optimize_operators/opt_conv_cuda.rst.txt | 2 +-
.../optimize_operators/opt_conv_tensorcore.rst.txt | 2 +-
.../how_to/optimize_operators/opt_gemm.rst.txt | 16 +-
.../optimize_operators/sg_execution_times.rst.txt | 8 +-
.../sg_execution_times.rst.txt | 16 +-
.../tune_conv2d_layer_cuda.rst.txt | 2394 ++++++++++++++++++--
.../tune_network_cuda.rst.txt | 2 +-
.../tune_network_x86.rst.txt | 4 +-
.../tune_sparse_x86.rst.txt | 175 +-
.../tune_with_autotvm/sg_execution_times.rst.txt | 12 +-
.../tune_with_autotvm/tune_conv2d_cuda.rst.txt | 34 +-
.../work_with_microtvm/micro_autotune.rst.txt | 16 +-
.../work_with_microtvm/sg_execution_times.rst.txt | 12 +-
.../work_with_relay/sg_execution_times.rst.txt | 8 +-
.../work_with_schedules/sg_execution_times.rst.txt | 18 +-
.../how_to/work_with_schedules/tensorize.rst.txt | 4 +-
.../tutorials/autotvm/sg_execution_times.rst.txt | 6 +-
.../frontend/deploy_classification.rst.txt | 2 +-
.../tutorials/frontend/deploy_detection.rst.txt | 2 +-
.../tutorials/frontend/sg_execution_times.rst.txt | 6 +-
.../tutorials/optimize/sg_execution_times.rst.txt | 6 +-
.../topic/vta/tutorials/sg_execution_times.rst.txt | 6 +-
.../tutorial/auto_scheduler_matmul_x86.rst.txt | 9 +-
docs/_sources/tutorial/autotvm_relay_x86.rst.txt | 62 +-
.../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 | 40 +-
docs/commit_hash | 2 +-
docs/how_to/compile_models/from_mxnet.html | 2 +-
docs/how_to/compile_models/from_paddle.html | 2 +-
docs/how_to/compile_models/from_pytorch.html | 29 +-
docs/how_to/compile_models/from_tensorflow.html | 2 +-
docs/how_to/compile_models/sg_execution_times.html | 20 +-
.../deploy_models/deploy_model_on_android.html | 2 +-
.../deploy_object_detection_pytorch.html | 157 +-
docs/how_to/deploy_models/deploy_prequantized.html | 11 +-
.../deploy_models/deploy_prequantized_tflite.html | 4 +-
docs/how_to/deploy_models/deploy_quantized.html | 2 +-
docs/how_to/deploy_models/deploy_ssd_gluoncv.html | 37 +-
docs/how_to/deploy_models/sg_execution_times.html | 16 +-
.../extend_tvm/bring_your_own_datatypes.html | 2 +-
docs/how_to/extend_tvm/sg_execution_times.html | 10 +-
docs/how_to/extend_tvm/use_pass_instrument.html | 16 +-
docs/how_to/optimize_operators/opt_conv_cuda.html | 2 +-
.../optimize_operators/opt_conv_tensorcore.html | 2 +-
docs/how_to/optimize_operators/opt_gemm.html | 16 +-
.../optimize_operators/sg_execution_times.html | 8 +-
.../sg_execution_times.html | 14 +-
.../tune_conv2d_layer_cuda.html | 2394 ++++++++++++++++++--
.../tune_with_autoscheduler/tune_network_cuda.html | 2 +-
.../tune_with_autoscheduler/tune_network_x86.html | 4 +-
.../tune_with_autoscheduler/tune_sparse_x86.html | 175 +-
.../tune_with_autotvm/sg_execution_times.html | 12 +-
.../how_to/tune_with_autotvm/tune_conv2d_cuda.html | 34 +-
docs/how_to/work_with_microtvm/micro_autotune.html | 16 +-
.../work_with_microtvm/sg_execution_times.html | 12 +-
.../how_to/work_with_relay/sg_execution_times.html | 8 +-
.../work_with_schedules/sg_execution_times.html | 18 +-
docs/how_to/work_with_schedules/tensorize.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 | 6 +-
.../tutorials/frontend/deploy_classification.html | 2 +-
.../vta/tutorials/frontend/deploy_detection.html | 2 +-
.../vta/tutorials/frontend/sg_execution_times.html | 6 +-
.../vta/tutorials/optimize/sg_execution_times.html | 6 +-
docs/topic/vta/tutorials/sg_execution_times.html | 6 +-
docs/tutorial/auto_scheduler_matmul_x86.html | 4 +-
docs/tutorial/autotvm_relay_x86.html | 172 +-
docs/tutorial/cross_compilation_and_rpc.html | 2 +-
docs/tutorial/intro_topi.html | 2 +-
docs/tutorial/sg_execution_times.html | 22 +-
docs/tutorial/tensor_expr_get_started.html | 40 +-
113 files changed, 5539 insertions(+), 1217 deletions(-)
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 ea7e47b..212d0c6 100644
--- a/docs/_sources/how_to/compile_models/from_mxnet.rst.txt
+++ b/docs/_sources/how_to/compile_models/from_mxnet.rst.txt
@@ -98,7 +98,7 @@ In this section, we download a pretrained imagenet model and classify an image.
.. code-block:: none
- Downloading /workspace/.mxnet/models/resnet18_v1-a0666292.zipcad3e097-2ad2-47c3-8ea4-462b69f1426b from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/resnet18_v1-a0666292.zip...
+ Downloading /workspace/.mxnet/models/resnet18_v1-a0666292.zip8c4ab368-acda-4b07-b69a-0498f0305aba 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_paddle.rst.txt b/docs/_sources/how_to/compile_models/from_paddle.rst.txt
index bfffb3b..9affcd7 100644
--- a/docs/_sources/how_to/compile_models/from_paddle.rst.txt
+++ b/docs/_sources/how_to/compile_models/from_paddle.rst.txt
@@ -182,7 +182,7 @@ Look up prediction top 1 index in 1000 class synset.
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 5.435 seconds)
+ **Total running time of the script:** ( 1 minutes 5.921 seconds)
.. _sphx_glr_download_how_to_compile_models_from_paddle.py:
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 a09925b..2e8ba57 100644
--- a/docs/_sources/how_to/compile_models/from_pytorch.rst.txt
+++ b/docs/_sources/how_to/compile_models/from_pytorch.rst.txt
@@ -79,7 +79,7 @@ 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]
6%|6 | 2.81M/44.7M [00:00<00:01, 29.4MB/s]
13%|#2 | 5.62M/44.7M [00:00<00:03, 11.5MB/s]
16%|#6 | 7.23M/44.7M [00:00<00:03, 12.5MB/s]
20%|#9 | 8.81M/44.7M [00:00<00:02, 13.2MB/s]
26%|##5 | 11.5M/44.7M [00:00<00:02, 17.1MB/s]
30%|### | 13.4M/44.7M [00:01<00:02, 13.5MB/s]
34%|###3 | 15.0M/44.7M [00:01<00:02, 13.8MB/s]
37%|###6 | 16.5M/44.7M [00:01<00:02, 12.8MB/s]
40%|###9 | 17.9M/44.7M [00:01<00:02, 11.7MB/s]
43%|####2 | 19.1M/44.7M [00:01<00:02, 11.3MB/s]
45%|####5 | 20.2M/44.7M [00:01<00:02, 9.06MB/s]
48%|####7 | 21.3M/44.7M [00:01<00:02, 9.26MB/s]
50%|####9 | 22.3M/44.7M [00:01<00:02, 9.45MB/s]
54%|#####4 | 24.3M/44.7M [00:02<00:01, 12.0MB/s]
59%|#####8 | 26.2M/44.7M [00:02<00:01, 13.9MB/s]
63%|######2 | 28.1M/44.7M [00:02<00:01, 15.5MB/s]
66%|######6 | 29.7M/44.7M [00
:02<00:01, 14.9MB/s]
70%|######9 | 31.1M/44.7M [00:02<00:01, 9.33MB/s]
73%|#######3 | 32.7M/44.7M [00:02<00:01, 10.5MB/s]
78%|#######7 | 34.7M/44.7M [00:03<00:00, 11.1MB/s]
82%|########2 | 36.7M/44.7M [00:03<00:00, 13.1MB/s]
85%|########5 | 38.2M/44.7M [00:03<00:00, 13.4MB/s]
90%|######### | 40.3M/44.7M [00:03<00:00, 15.5MB/s]
95%|#########4| 42.4M/44.7M [00:03<00:00, 17.2MB/s]
100%|##########| 44.7M/44.7M [00:03<00:00, 13.4MB/s]
+
0%| | 0.00/44.7M [00:00<?, ?B/s]
21%|##1 | 9.52M/44.7M [00:00<00:00, 99.8MB/s]
57%|#####7 | 25.5M/44.7M [00:00<00:00, 140MB/s]
93%|#########3| 41.8M/44.7M [00:00<00:00, 153MB/s]
100%|##########| 44.7M/44.7M [00:00<00:00, 147MB/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 daa79dc..0d9766e 100644
--- a/docs/_sources/how_to/compile_models/from_tensorflow.rst.txt
+++ b/docs/_sources/how_to/compile_models/from_tensorflow.rst.txt
@@ -380,7 +380,7 @@ Run the corresponding model on tensorflow
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 11.242 seconds)
+ **Total running time of the script:** ( 1 minutes 8.286 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 30d23bf..46ffecd 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
@@ -5,14 +5,14 @@
Computation times
=================
-**04:55.760** total execution time for **how_to_compile_models** files:
+**04:51.558** total execution time for **how_to_compile_models** files:
-- **01:11.242**: :ref:`sphx_glr_how_to_compile_models_from_tensorflow.py` (``from_tensorflow.py``)
-- **01:05.435**: :ref:`sphx_glr_how_to_compile_models_from_paddle.py` (``from_paddle.py``)
-- **00:54.383**: :ref:`sphx_glr_how_to_compile_models_from_darknet.py` (``from_darknet.py``)
-- **00:24.480**: :ref:`sphx_glr_how_to_compile_models_from_tflite.py` (``from_tflite.py``)
-- **00:23.067**: :ref:`sphx_glr_how_to_compile_models_from_pytorch.py` (``from_pytorch.py``)
-- **00:20.618**: :ref:`sphx_glr_how_to_compile_models_from_mxnet.py` (``from_mxnet.py``)
-- **00:20.223**: :ref:`sphx_glr_how_to_compile_models_from_coreml.py` (``from_coreml.py``)
-- **00:13.886**: :ref:`sphx_glr_how_to_compile_models_from_keras.py` (``from_keras.py``)
-- **00:02.426**: :ref:`sphx_glr_how_to_compile_models_from_onnx.py` (``from_onnx.py``)
+- **01:08.286**: :ref:`sphx_glr_how_to_compile_models_from_tensorflow.py` (``from_tensorflow.py``)
+- **01:05.921**: :ref:`sphx_glr_how_to_compile_models_from_paddle.py` (``from_paddle.py``)
+- **00:55.651**: :ref:`sphx_glr_how_to_compile_models_from_darknet.py` (``from_darknet.py``)
+- **00:25.350**: :ref:`sphx_glr_how_to_compile_models_from_tflite.py` (``from_tflite.py``)
+- **00:20.574**: :ref:`sphx_glr_how_to_compile_models_from_mxnet.py` (``from_mxnet.py``)
+- **00:20.378**: :ref:`sphx_glr_how_to_compile_models_from_coreml.py` (``from_coreml.py``)
+- **00:18.481**: :ref:`sphx_glr_how_to_compile_models_from_pytorch.py` (``from_pytorch.py``)
+- **00:13.972**: :ref:`sphx_glr_how_to_compile_models_from_keras.py` (``from_keras.py``)
+- **00:02.945**: :ref:`sphx_glr_how_to_compile_models_from_onnx.py` (``from_onnx.py``)
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 416ebc4..5a3b0fb 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
@@ -393,7 +393,7 @@ Execute on TVM
Evaluate inference time cost...
Execution time summary:
mean (ms) median (ms) max (ms) min (ms) std (ms)
- 15.4084 15.4311 15.5087 15.2838 0.0684
+ 16.0157 15.9729 16.5046 15.8410 0.1828
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 8661248..3a7dc07 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
@@ -108,7 +108,7 @@ 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]
0%| | 824k/170M [00:00<00:21, 8.14MB/s]
1%|1 | 2.38M/170M [00:00<00:13, 12.9MB/s]
3%|2 | 5.06M/170M [00:00<00:09, 18.5MB/s]
4%|4 | 7.24M/170M [00:00<00:09, 18.1MB/s]
5%|5 | 8.96M/170M [00:00<00:11, 14.8MB/s]
6%|6 | 10.4M/170M [00:00<00:14, 11.8MB/s]
7%|6 | 11.9M/170M [00:00<00:13, 12.4MB/s]
8%|7 | 13.6M/170M [00:01<00:12, 13.6MB/s]
9%|8 | 15.1M/170M [00:01<00:11, 14.1MB/s]
10%|9 | 16.6M/170M [00:01<00:11, 14.2MB/s]
11%|# | 18.0M/170M [00:01<00:11, 13.7MB/s]
12%|#2 | 20.5M/170M [00:01<00:09, 16.8MB/s]
13%|#3 | 22.5M/170M [00:01<00:08, 18.0MB/s]
14%|#4 | 24.2M/170M [00:01<00:08, 17.0MB/s]
15%|#5 | 25.9M/170M [00:01<00:09, 15.1MB/s]
16%|#6 | 27.4M/170M [00:01<00:11, 13.6MB/s]
17%|#7 | 29.1M/170M [00:02<00:11, 12.9MB/s
]
18%|#8 | 31.0M/170M [00:02<00:10, 14.5MB/s]
20%|#9 | 33.5M/170M [00:02<00:08, 17.4MB/s]
21%|## | 35.3M/170M [00:02<00:08, 17.4MB/s]
22%|##1 | 37.0M/170M [00:02<00:09, 15.3MB/s]
23%|##2 | 38.5M/170M [00:02<00:09, 15.1MB/s]
24%|##3 | 40.7M/170M [00:02<00:08, 16.9MB/s]
25%|##4 | 42.4M/170M [00:02<00:08, 15.6MB/s]
26%|##6 | 44.2M/170M [00:03<00:07, 16.6MB/s]
27%|##7 | 45.9M/170M [00:03<00:08, 15.8MB/s]
28%|##7 | 47.4M/170M [00:03<00:08, 15.8MB/s]
29%|##8 | 48.9M/170M [00:03<00:11, 11.0MB/s]
30%|##9 | 50.4M/170M [00:03<00:10, 11.8MB/s]
31%|###1 | 53.3M/170M [00:03<00:07, 15.6MB/s]
33%|###2 | 55.4M/170M [00:03<00:06, 17.4MB/s]
34%|###4 | 57.9M/170M [00:03<00:06, 19.2MB/s]
35%|###5 | 59.9M/170M [00:04<00:07, 16.5MB/s]
36%|###6 | 61.6M/170M [00:04<00:08, 13.7MB/s]
37%|###7 | 63.3M/170M [00:04<00:0
7, 14.6MB/s]
39%|###8 | 65.5M/170M [00:04<00:08, 13.2MB/s]
40%|###9 | 67.7M/170M [00:04<00:07, 14.9MB/s]
42%|####1 | 70.5M/170M [00:04<00:05, 17.5MB/s]
43%|####2 | 72.9M/170M [00:04<00:05, 19.0MB/s]
44%|####4 | 75.1M/170M [00:05<00:04, 20.1MB/s]
45%|####5 | 77.1M/170M [00:05<00:04, 20.1MB/s]
47%|####6 | 79.2M/170M [00:05<00:04, 20.1MB/s]
48%|####7 | 81.1M/170M [00:05<00:05, 17.6MB/s]
49%|####8 | 82.9M/170M [00:05<00:05, 16.8MB/s]
50%|####9 | 84.6M/170M [00:05<00:06, 13.5MB/s]
51%|##### | 86.0M/170M [00:05<00:06, 12.8MB/s]
51%|#####1 | 87.3M/170M [00:06<00:07, 12.3MB/s]
52%|#####2 | 88.7M/170M [00:06<00:06, 12.7MB/s]
53%|#####3 | 90.7M/170M [00:06<00:05, 14.2MB/s]
55%|#####4 | 93.2M/170M [00:06<00:04, 17.3MB/s]
56%|#####6 | 95.9M/170M [00:06<00:03, 20.2MB/s]
58%|#####7 | 97.9M/170M [00:06<00:04, 18.0MB/s]
59%|#####8 | 99.7M/170M
[00:06<00:04, 15.5MB/s]
60%|#####9 | 101M/170M [00:06<00:04, 15.2MB/s]
61%|###### | 103M/170M [00:06<00:04, 16.1MB/s]
62%|######1 | 105M/170M [00:07<00:04, 16.5MB/s]
63%|######3 | 107M/170M [00:07<00:03, 17.4MB/s]
64%|######4 | 109M/170M [00:07<00:03, 18.8MB/s]
66%|######6 | 113M/170M [00:07<00:02, 23.2MB/s]
68%|######7 | 115M/170M [00:07<00:02, 22.7MB/s]
69%|######9 | 118M/170M [00:07<00:02, 24.9MB/s]
71%|####### | 120M/170M [00:07<00:02, 21.8MB/s]
72%|#######2 | 122M/170M [00:07<00:02, 21.6MB/s]
73%|#######3 | 125M/170M [00:07<00:02, 21.1MB/s]
75%|#######4 | 127M/170M [00:08<00:02, 19.7MB/s]
76%|#######5 | 129M/170M [00:08<00:02, 20.4MB/s]
77%|#######7 | 131M/170M [00:08<00:02, 16.9MB/s]
78%|#######8 | 133M/170M [00:08<00:02, 14.9MB/s]
79%|#######9 | 134M/170M [00:08<00:02, 14.6MB/s]
80%|#######9 | 136M/170M [00:08<00:02, 13.5MB/s]
81%|######## | 137M/170M [00:09
<00:03, 10.9MB/s]
81%|########1 | 138M/170M [00:09<00:03, 10.9MB/s]
82%|########2 | 139M/170M [00:09<00:03, 9.60MB/s]
83%|########2 | 141M/170M [00:09<00:02, 10.5MB/s]
84%|########3 | 142M/170M [00:09<00:02, 11.6MB/s]
85%|########4 | 144M/170M [00:09<00:01, 13.5MB/s]
86%|########5 | 146M/170M [00:09<00:01, 14.4MB/s]
87%|########6 | 147M/170M [00:09<00:01, 13.6MB/s]
88%|########7 | 149M/170M [00:09<00:01, 14.1MB/s]
89%|########8 | 151M/170M [00:10<00:01, 15.7MB/s]
90%|########9 | 152M/170M [00:10<00:01, 13.6MB/s]
91%|#########1| 155M/170M [00:10<00:00, 17.3MB/s]
93%|#########2| 157M/170M [00:10<00:00, 18.9MB/s]
94%|#########3| 159M/170M [00:10<00:00, 19.6MB/s]
95%|#########5| 161M/170M [00:10<00:00, 19.4MB/s]
96%|#########6| 163M/170M [00:10<00:00, 18.9MB/s]
97%|#########7| 165M/170M [00:10<00:00, 17.8MB/s]
98%|#########8| 167M/170M [00:11<00:00, 16.8MB/s]
99%|#########9| 168M/170M [00:11<00:00,
13.8MB/s]
100%|##########| 170M/170M [00:11<00:00, 15.7MB/s]
+
0%| | 0.00/170M [00:00<?, ?B/s]
3%|2 | 4.56M/170M [00:00<00:03, 46.9MB/s]
5%|5 | 9.04M/170M [00:00<00:04, 33.7MB/s]
7%|7 | 12.5M/170M [00:00<00:06, 24.7MB/s]
10%|# | 17.4M/170M [00:00<00:04, 32.1MB/s]
13%|#2 | 22.0M/170M [00:00<00:04, 35.7MB/s]
15%|#5 | 25.8M/170M [00:00<00:04, 35.4MB/s]
18%|#7 | 30.0M/170M [00:00<00:03, 37.8MB/s]
20%|#9 | 33.8M/170M [00:01<00:03, 36.8MB/s]
22%|##2 | 37.4M/170M [00:01<00:04, 30.3MB/s]
24%|##3 | 40.5M/170M [00:01<00:04, 29.0MB/s]
26%|##5 | 43.5M/170M [00:01<00:04, 28.9MB/s]
27%|##7 | 46.3M/170M [00:01<00:05, 23.4MB/s]
30%|### | 51.0M/170M [00:01<00:04, 29.2MB/s]
33%|###2 | 55.5M/170M [00:01<00:03, 33.4MB/s]
35%|###4 | 58.9M/170M [00:02<00:04, 27.7MB/s]
37%|###6 | 62.3M/170M [00:02<00:03, 29.4MB/s]
39%|###8 | 65.4M/170M [00:02<00:03, 29.9MB/
s]
40%|#### | 68.5M/170M [00:02<00:03, 30.2MB/s]
42%|####2 | 71.5M/170M [00:02<00:03, 26.9MB/s]
44%|####3 | 74.2M/170M [00:02<00:04, 24.1MB/s]
45%|####5 | 76.7M/170M [00:02<00:04, 19.5MB/s]
47%|####7 | 80.6M/170M [00:02<00:03, 24.1MB/s]
49%|####8 | 83.2M/170M [00:03<00:03, 22.9MB/s]
51%|##### | 86.0M/170M [00:03<00:03, 24.0MB/s]
53%|#####2 | 89.8M/170M [00:03<00:03, 27.9MB/s]
55%|#####4 | 92.6M/170M [00:03<00:02, 28.3MB/s]
56%|#####6 | 95.5M/170M [00:03<00:03, 22.1MB/s]
58%|#####7 | 97.9M/170M [00:03<00:03, 19.9MB/s]
59%|#####9 | 100M/170M [00:03<00:03, 21.2MB/s]
61%|###### | 103M/170M [00:03<00:03, 22.5MB/s]
62%|######2 | 106M/170M [00:04<00:02, 24.0MB/s]
64%|######3 | 108M/170M [00:04<00:02, 23.9MB/s]
66%|######6 | 112M/170M [00:04<00:02, 28.3MB/s]
68%|######7 | 115M/170M [00:04<00:02, 26.3MB/s]
69%|######9 | 118M/170M [00:04<00:02, 24
.3MB/s]
71%|####### | 120M/170M [00:04<00:02, 24.4MB/s]
72%|#######2 | 123M/170M [00:05<00:03, 13.9MB/s]
73%|#######3 | 124M/170M [00:05<00:03, 14.3MB/s]
75%|#######4 | 127M/170M [00:05<00:02, 15.6MB/s]
76%|#######6 | 130M/170M [00:05<00:02, 16.6MB/s]
78%|#######7 | 132M/170M [00:05<00:02, 15.6MB/s]
79%|#######9 | 135M/170M [00:05<00:01, 18.6MB/s]
80%|######## | 137M/170M [00:05<00:02, 16.4MB/s]
81%|########1 | 138M/170M [00:05<00:01, 17.0MB/s]
83%|########2 | 140M/170M [00:06<00:02, 15.6MB/s]
84%|########3 | 142M/170M [00:06<00:01, 17.2MB/s]
85%|########5 | 145M/170M [00:06<00:01, 18.8MB/s]
86%|########6 | 146M/170M [00:06<00:01, 14.3MB/s]
88%|########7 | 149M/170M [00:06<00:01, 17.5MB/s]
89%|########9 | 152M/170M [00:06<00:00, 19.4MB/s]
91%|######### | 154M/170M [00:06<00:00, 19.6MB/s]
92%|#########1| 156M/170M [00:06<00:00, 20.0MB/s]
94%|#########3| 160M/170M [00:07<00:00, 24.7MB/s]
95%|#########5| 162M/170M [00:07<00:00, 21.2MB/s]
97%|#########6| 164M/170M [00:07<00:00, 20.6MB/s]
98%|#########8| 167M/170M [00:07<00:00, 23.2MB/s]
100%|#########9| 170M/170M [00:07<00:00, 21.3MB/s]
100%|##########| 170M/170M [00:07<00:00, 23.4MB/s]
/usr/local/lib/python3.7/dist-packages/torch/functional.py:445: UserWarning: torch.meshgrid: in an upcoming release, it will be required to pass the indexing argument. (Triggered internally at ../aten/src/ATen/native/TensorShape.cpp:2157.)
return _VF.meshgrid(tensors, **kwargs) # type: ignore[attr-defined]
/usr/local/lib/python3.7/dist-packages/torch/nn/functional.py:3702: 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).
@@ -129,13 +129,6 @@ Load pre-trained maskrcnn from torchvision and do tracing
for s, s_orig in zip(new_size, original_size)
/usr/local/lib/python3.7/dist-packages/torchvision/models/detection/roi_heads.py:372: 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).
return torch.tensor(M + 2 * padding).to(torch.float32) / torch.tensor(M).to(torch.float32)
- /usr/local/lib/python3.7/dist-packages/torch/jit/_trace.py:991: TracerWarning: Output nr 2. of the traced function does not match the corresponding output of the Python function. Detailed error:
- Tensor-likes are not close!
-
- Mismatched elements: 1 / 2 (50.0%)
- Greatest absolute difference: 0.9288840740919113 at index 1 (up to 1e-05 allowed)
- Greatest relative difference: 13.16154513835738 at index 1 (up to 1e-05 allowed)
- _module_class,
@@ -262,7 +255,7 @@ Get boxes with score larger than 0.9
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 3 minutes 6.085 seconds)
+ **Total running time of the script:** ( 3 minutes 7.676 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 f83ca9d..ff56614 100644
--- a/docs/_sources/how_to/deploy_models/deploy_prequantized.rst.txt
+++ b/docs/_sources/how_to/deploy_models/deploy_prequantized.rst.txt
@@ -187,7 +187,7 @@ 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]
100%|##########| 13.6M/13.6M [00:00<00:00, 144MB/s]
+
0%| | 0.00/13.6M [00:00<?, ?B/s]
9%|8 | 1.16M/13.6M [00:00<00:01, 11.9MB/s]
22%|##2 | 3.01M/13.6M [00:00<00:00, 16.2MB/s]
45%|####5 | 6.13M/13.6M [00:00<00:00, 23.2MB/s]
61%|######1 | 8.34M/13.6M [00:00<00:00, 19.7MB/s]
90%|########9 | 12.1M/13.6M [00:00<00:00, 26.1MB/s]
100%|##########| 13.6M/13.6M [00:00<00:00, 23.6MB/s]
@@ -348,7 +348,7 @@ 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)
- 89.3021 89.9408 90.7087 88.0098 0.8485
+ 90.1617 90.1035 93.2273 89.9587 0.3468
@@ -388,7 +388,7 @@ TODO
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 2.571 seconds)
+ **Total running time of the script:** ( 1 minutes 3.920 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 89f81e9..2226258 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
@@ -351,7 +351,7 @@ 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)
- 118.1848 118.1044 126.6462 116.9886 0.9616
+ 119.4513 119.4375 122.0499 118.4966 0.4518
@@ -385,7 +385,7 @@ 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:** ( 1 minutes 52.560 seconds)
+ **Total running time of the script:** ( 1 minutes 58.957 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 e3dd405..52ae7fb 100644
--- a/docs/_sources/how_to/deploy_models/deploy_quantized.rst.txt
+++ b/docs/_sources/how_to/deploy_models/deploy_quantized.rst.txt
@@ -221,7 +221,7 @@ We create a Relay VM to build and execute the model.
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 15.134 seconds)
+ **Total running time of the script:** ( 1 minutes 11.281 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 830fe54..972cdd4 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
@@ -137,7 +137,7 @@ Convert and compile model for CPU.
data: None
input_sym_arg_type = in_param.infer_type()[0]
Downloading /workspace/.mxnet/models/ssd_512_resnet50_v1_voc-9c8b225a.zip from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/ssd_512_resnet50_v1_voc-9c8b225a.zip...
-
0%| | 0/132723 [00:00<?, ?KB/s]
4%|3 | 4830/132723 [00:00<00:02, 48286.23KB/s]
9%|9 | 12337/132723 [00:00<00:01, 64027.72KB/s]
15%|#4 | 19833/132723 [00:00<00:01, 69014.41KB/s]
21%|## | 27444/132723 [00:00<00:01, 71813.69KB/s]
27%|##6 | 35206/132723 [00:00<00:01, 73904.50KB/s]
32%|###2 | 42597/132723 [00:00<00:01, 68105.15KB/s]
37%|###7 | 49540/132723 [00:00<00:01, 68513.57KB/s]
43%|####3 | 57099/132723 [00:00<00:01, 70679.52KB/s]
49%|####8 | 64719/132723 [00:00<00:00, 72357.98KB/s]
54%|#####4 | 72138/132723 [00:01<00:00, 72911.50KB/s]
60%|###### | 79790/132723 [00:01<00:00, 74000.67KB/s]
66%|######5 | 87391/132723 [00:01<00:00, 74604.83KB/s]
72%|#######1 | 95004/132723 [00:01<00:00, 75062.31KB/s]
77%|#######7 | 102552/132723 [00:01<00:00, 75182.93KB/s]
83%|########3 | 110227/132723 [00:01<00:00, 75652.81KB/s]
89%|########8
| 117841/132723 [00:01<00:00, 75794.22KB/s]
95%|#########4| 125424/132723 [00:01<00:00, 75772.01KB/s]
100%|##########| 132723/132723 [00:01<00:00, 72865.56KB/s]
+
0%| | 0/132723 [00:00<?, ?KB/s]
4%|3 | 5041/132723 [00:00<00:02, 50394.55KB/s]
10%|9 | 12799/132723 [00:00<00:01, 66382.15KB/s]
16%|#5 | 20593/132723 [00:00<00:01, 71654.23KB/s]
21%|##1 | 28437/132723 [00:00<00:01, 74330.43KB/s]
27%|##7 | 36266/132723 [00:00<00:01, 75722.39KB/s]
33%|###3 | 44150/132723 [00:00<00:01, 76780.15KB/s]
39%|###9 | 52193/132723 [00:00<00:01, 77971.38KB/s]
45%|####5 | 60362/132723 [00:00<00:00, 79152.79KB/s]
52%|#####1 | 68530/132723 [00:00<00:00, 79939.76KB/s]
58%|#####7 | 76637/132723 [00:01<00:00, 80287.36KB/s]
64%|######3 | 84666/132723 [00:01<00:00, 77459.33KB/s]
70%|######9 | 92788/132723 [00:01<00:00, 78571.90KB/s]
76%|#######5 | 100869/132723 [00:01<00:00, 79233.85KB/s]
82%|########2 | 108983/132723 [00:01<00:00, 79801.12KB/s]
88%|########8 | 117095/132723 [00:01<00:00, 80192.06KB/s]
94%|########
#4| 125195/132723 [00:01<00:00, 80429.95KB/s]
100%|##########| 132723/132723 [00:01<00:00, 77817.54KB/s]
@@ -202,7 +202,7 @@ Display result
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 2 minutes 21.793 seconds)
+ **Total running time of the script:** ( 2 minutes 20.271 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 70c4fa7..b883abf 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
@@ -5,13 +5,13 @@
Computation times
=================
-**10:25.692** total execution time for **how_to_deploy_models** files:
+**10:30.903** total execution time for **how_to_deploy_models** files:
-- **03:06.085**: :ref:`sphx_glr_how_to_deploy_models_deploy_object_detection_pytorch.py` (``deploy_object_detection_pytorch.py``)
-- **02:21.793**: :ref:`sphx_glr_how_to_deploy_models_deploy_ssd_gluoncv.py` (``deploy_ssd_gluoncv.py``)
-- **01:52.560**: :ref:`sphx_glr_how_to_deploy_models_deploy_prequantized_tflite.py` (``deploy_prequantized_tflite.py``)
-- **01:15.134**: :ref:`sphx_glr_how_to_deploy_models_deploy_quantized.py` (``deploy_quantized.py``)
-- **01:02.571**: :ref:`sphx_glr_how_to_deploy_models_deploy_prequantized.py` (``deploy_prequantized.py``)
-- **00:26.628**: :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_android.py` (``deploy_model_on_android.py``)
-- **00:20.715**: :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_rasp.py` (``deploy_model_on_rasp.py``)
+- **03:07.676**: :ref:`sphx_glr_how_to_deploy_models_deploy_object_detection_pytorch.py` (``deploy_object_detection_pytorch.py``)
+- **02:20.271**: :ref:`sphx_glr_how_to_deploy_models_deploy_ssd_gluoncv.py` (``deploy_ssd_gluoncv.py``)
+- **01:58.957**: :ref:`sphx_glr_how_to_deploy_models_deploy_prequantized_tflite.py` (``deploy_prequantized_tflite.py``)
+- **01:11.281**: :ref:`sphx_glr_how_to_deploy_models_deploy_quantized.py` (``deploy_quantized.py``)
+- **01:03.920**: :ref:`sphx_glr_how_to_deploy_models_deploy_prequantized.py` (``deploy_prequantized.py``)
+- **00:26.856**: :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_android.py` (``deploy_model_on_android.py``)
+- **00:21.739**: :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_rasp.py` (``deploy_model_on_rasp.py``)
- **00:00.204**: :ref:`sphx_glr_how_to_deploy_models_deploy_sparse.py` (``deploy_sparse.py``)
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 5e13cb5..c0df344 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
@@ -423,7 +423,7 @@ 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.zip0bee6f03-5151-42e6-bd49-fb366a42751b from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/mobilenet0.25-9f83e440.zip...
+ Downloading /workspace/.mxnet/models/mobilenet0.25-9f83e440.zipaa47f14e-527d-4ca7-8541-3bd59544493d from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/mobilenet0.25-9f83e440.zip...
diff --git a/docs/_sources/how_to/extend_tvm/sg_execution_times.rst.txt b/docs/_sources/how_to/extend_tvm/sg_execution_times.rst.txt
index 27a82cd..21c3f5b 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
@@ -5,9 +5,9 @@
Computation times
=================
-**00:38.327** total execution time for **how_to_extend_tvm** files:
+**00:38.274** total execution time for **how_to_extend_tvm** files:
-- **00:34.744**: :ref:`sphx_glr_how_to_extend_tvm_bring_your_own_datatypes.py` (``bring_your_own_datatypes.py``)
-- **00:02.288**: :ref:`sphx_glr_how_to_extend_tvm_use_pass_instrument.py` (``use_pass_instrument.py``)
-- **00:01.083**: :ref:`sphx_glr_how_to_extend_tvm_use_pass_infra.py` (``use_pass_infra.py``)
-- **00:00.211**: :ref:`sphx_glr_how_to_extend_tvm_low_level_custom_pass.py` (``low_level_custom_pass.py``)
+- **00:34.679**: :ref:`sphx_glr_how_to_extend_tvm_bring_your_own_datatypes.py` (``bring_your_own_datatypes.py``)
+- **00:02.282**: :ref:`sphx_glr_how_to_extend_tvm_use_pass_instrument.py` (``use_pass_instrument.py``)
+- **00:01.094**: :ref:`sphx_glr_how_to_extend_tvm_use_pass_infra.py` (``use_pass_infra.py``)
+- **00:00.219**: :ref:`sphx_glr_how_to_extend_tvm_low_level_custom_pass.py` (``low_level_custom_pass.py``)
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 407d8a8..a9988f2 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
@@ -199,10 +199,10 @@ profile the execution time of each passes.
.. code-block:: none
Printing results of timing profile...
- InferType: 6122us [6122us] (45.27%; 45.27%)
- FoldScaleAxis: 7401us [2us] (54.73%; 54.73%)
- FoldConstant: 7399us [1546us] (54.71%; 99.97%)
- InferType: 5852us [5852us] (43.28%; 79.10%)
+ InferType: 5982us [5982us] (45.45%; 45.45%)
+ FoldScaleAxis: 7180us [2us] (54.55%; 54.55%)
+ FoldConstant: 7178us [1474us] (54.54%; 99.97%)
+ InferType: 5703us [5703us] (43.33%; 79.46%)
@@ -239,10 +239,10 @@ Refer to following sections and :py:func:`tvm.instrument.pass_instrument` for th
.. code-block:: none
Printing results of timing profile...
- InferType: 5938us [5938us] (44.77%; 44.77%)
- FoldScaleAxis: 7325us [2us] (55.23%; 55.23%)
- FoldConstant: 7323us [1522us] (55.21%; 99.98%)
- InferType: 5800us [5800us] (43.74%; 79.21%)
+ InferType: 5767us [5767us] (44.83%; 44.83%)
+ FoldScaleAxis: 7097us [2us] (55.17%; 55.17%)
+ FoldConstant: 7095us [1478us] (55.16%; 99.97%)
+ InferType: 5617us [5617us] (43.67%; 79.17%)
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 0ad07e9..b15302a 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
@@ -295,7 +295,7 @@ latency of convolution.
.. code-block:: none
- Convolution: 45.131203 ms
+ Convolution: 35.525673 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 33112dd..1384cf1 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
@@ -626,7 +626,7 @@ be able to run on our build server
.. code-block:: none
- conv2d with tensor core: 10.130518 ms
+ conv2d with tensor core: 11.048099 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 3b02956..f587ee8 100644
--- a/docs/_sources/how_to/optimize_operators/opt_gemm.rst.txt
+++ b/docs/_sources/how_to/optimize_operators/opt_gemm.rst.txt
@@ -118,8 +118,8 @@ Then we write a baseline implementation, the simplest way to write a matrix mult
.. code-block:: none
- Numpy running time: 0.017912
- Baseline: 3.381993
+ Numpy running time: 0.018033
+ Baseline: 3.296873
@@ -209,7 +209,7 @@ fill 32 * 32 * sizeof(float) which is 4KB in the cache whose total size is 32KB
.. code-block:: none
- Opt1: 0.308665
+ Opt1: 0.297950
@@ -307,7 +307,7 @@ In this tutorial, we chose to vectorize the inner loop row data since it is cach
.. code-block:: none
- Opt2: 0.328278
+ Opt2: 0.327325
@@ -398,7 +398,7 @@ the access pattern for A matrix is more cache friendly.
.. code-block:: none
- Opt3: 0.118575
+ Opt3: 0.116769
@@ -516,7 +516,7 @@ flattening.
.. code-block:: none
- Opt4: 0.111692
+ Opt4: 0.111140
@@ -633,7 +633,7 @@ write to C when all the block results are ready.
.. code-block:: none
- Opt5: 0.111826
+ Opt5: 0.111151
@@ -753,7 +753,7 @@ Futhermore, we can also utilize multi-core processors to do the thread-level par
.. code-block:: none
- Opt6: 0.144389
+ Opt6: 0.144714
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 b261207..7d8fdbb 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
@@ -5,8 +5,8 @@
Computation times
=================
-**00:35.018** total execution time for **how_to_optimize_operators** files:
+**00:34.816** total execution time for **how_to_optimize_operators** files:
-- **00:32.353**: :ref:`sphx_glr_how_to_optimize_operators_opt_gemm.py` (``opt_gemm.py``)
-- **00:01.463**: :ref:`sphx_glr_how_to_optimize_operators_opt_conv_tensorcore.py` (``opt_conv_tensorcore.py``)
-- **00:01.202**: :ref:`sphx_glr_how_to_optimize_operators_opt_conv_cuda.py` (``opt_conv_cuda.py``)
+- **00:31.900**: :ref:`sphx_glr_how_to_optimize_operators_opt_gemm.py` (``opt_gemm.py``)
+- **00:01.573**: :ref:`sphx_glr_how_to_optimize_operators_opt_conv_tensorcore.py` (``opt_conv_tensorcore.py``)
+- **00:01.343**: :ref:`sphx_glr_how_to_optimize_operators_opt_conv_cuda.py` (``opt_conv_cuda.py``)
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 1dabdb8..f60f82c 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
@@ -5,11 +5,11 @@
Computation times
=================
-**04:54.312** total execution time for **how_to_tune_with_autoscheduler** files:
-
-- **02:21.084**: :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_conv2d_layer_cuda.py` (``tune_conv2d_layer_cuda.py``)
-- **01:19.840**: :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_x86.py` (``tune_network_x86.py``)
-- **00:40.131**: :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_cuda.py` (``tune_network_cuda.py``)
-- **00:16.340**: :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_sparse_x86.py` (``tune_sparse_x86.py``)
-- **00:08.640**: :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_mali.py` (``tune_network_mali.py``)
-- **00:08.278**: :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_arm.py` (``tune_network_arm.py``)
+**04:56.572** total execution time for **how_to_tune_with_autoscheduler** files:
+
+- **02:21.977**: :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_conv2d_layer_cuda.py` (``tune_conv2d_layer_cuda.py``)
+- **01:19.340**: :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_x86.py` (``tune_network_x86.py``)
+- **00:40.052**: :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_cuda.py` (``tune_network_cuda.py``)
+- **00:18.821**: :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_sparse_x86.py` (``tune_sparse_x86.py``)
+- **00:08.276**: :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_mali.py` (``tune_network_mali.py``)
+- **00:08.106**: :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_arm.py` (``tune_network_arm.py``)
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 3ac182f..47f0077 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
@@ -221,74 +221,1163 @@ cooperative fetching, unrolling and operator fusion.
bias: Buffer(bias_2: Pointer(float32), float32, [512], []),
compute: Buffer(compute_2: Pointer(float32), float32, [25088], [])}
buffer_map = {data_1: data, kernel_1: kernel, bias_1: bias, compute_1: compute} {
- attr [IterVar(blockIdx.x: int32, (nullptr), "ThreadIndex", "blockIdx.x")] "thread_extent" = 112;
- allocate(conv2d_nchw: Pointer(local float32), float32, [4]), storage_scope = local;
- allocate(pad_temp.shared: Pointer(shared float32), float32, [72]), storage_scope = shared;
- allocate(kernel.shared: Pointer(shared float32), float32, [768]), storage_scope = shared;
- attr [IterVar(threadIdx.x: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 56 {
- conv2d_nchw_1: Buffer(conv2d_nchw, float32, [4], [], scope="local", align=16)[0] = 0f32
+ attr [IterVar(blockIdx.x: int32, (nullptr), "ThreadIndex", "blockIdx.x")] "thread_extent" = 8;
+ allocate(conv2d_nchw: Pointer(local float32), float32, [28]), storage_scope = local;
+ allocate(pad_temp.shared: Pointer(shared float32), float32, [648]), storage_scope = shared;
+ allocate(kernel.shared: Pointer(shared float32), float32, [4608]), storage_scope = shared;
+ attr [IterVar(threadIdx.x: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112 {
+ conv2d_nchw_1: Buffer(conv2d_nchw, float32, [196], [], scope="local", align=32)[0] = 0f32
+ conv2d_nchw_1[14] = 0f32
conv2d_nchw_1[1] = 0f32
+ conv2d_nchw_1[15] = 0f32
conv2d_nchw_1[2] = 0f32
+ conv2d_nchw_1[16] = 0f32
conv2d_nchw_1[3] = 0f32
+ conv2d_nchw_1[17] = 0f32
+ conv2d_nchw_1[4] = 0f32
+ conv2d_nchw_1[18] = 0f32
+ conv2d_nchw_1[5] = 0f32
+ conv2d_nchw_1[19] = 0f32
+ conv2d_nchw_1[6] = 0f32
+ conv2d_nchw_1[20] = 0f32
+ conv2d_nchw_1[7] = 0f32
+ conv2d_nchw_1[21] = 0f32
+ conv2d_nchw_1[8] = 0f32
+ conv2d_nchw_1[22] = 0f32
+ conv2d_nchw_1[9] = 0f32
+ conv2d_nchw_1[23] = 0f32
+ conv2d_nchw_1[10] = 0f32
+ conv2d_nchw_1[24] = 0f32
+ conv2d_nchw_1[11] = 0f32
+ conv2d_nchw_1[25] = 0f32
+ conv2d_nchw_1[12] = 0f32
+ conv2d_nchw_1[26] = 0f32
+ conv2d_nchw_1[13] = 0f32
+ conv2d_nchw_1[27] = 0f32
for (rc.outer.outer: int32, 0, 64) {
- for (rx.outer.outer: int32, 0, 3) {
- let cse_var_1: int32 = (rc.outer.outer*72)
- {
- attr [IterVar(threadIdx.x_1: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 56;
- pad_temp.shared_1: Buffer(pad_temp.shared, float32, [72], [], scope="shared")[threadIdx.x_1] = @tir.if_then_else(((((1 <= floormod(threadIdx.x_1, 9)) && (floormod(threadIdx.x_1, 9) < 8)) && (1 <= (rx.outer.outer + floormod(blockIdx.x, 7)))) && ((rx.outer.outer + floormod(blockIdx.x, 7)) < 8)), data[((((((rc.outer.outer*392) + (floordiv(threadIdx.x_1, 9)*49)) + (floormod(threadIdx.x_1, 9)*7)) + rx.outer.outer) + floormod(blockIdx.x, 7)) - 8)], 0f32, dtype=float32)
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 56;
- if @tir.likely((threadIdx.x_1 < 16), dtype=bool) {
- pad_temp.shared_1[(threadIdx.x_1 + 56)] = @tir.if_then_else(((((1 <= floormod((threadIdx.x_1 + 2), 9)) && (floormod((threadIdx.x_1 + 2), 9) < 8)) && (1 <= (rx.outer.outer + floormod(blockIdx.x, 7)))) && ((rx.outer.outer + floormod(blockIdx.x, 7)) < 8)), data[((((((rc.outer.outer*392) + (floordiv((threadIdx.x_1 + 56), 9)*49)) + (floormod((threadIdx.x_1 + 2), 9)*7)) + rx.outer.outer) + floormod(blockIdx.x, 7)) - 8)], 0f32, dtype=float32)
- }
- attr [IterVar(threadIdx.x_2: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 56;
- kernel.shared_1: Buffer(kernel.shared, float32, [768], [], scope="shared")[threadIdx.x_2] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv(threadIdx.x_2, 24)*4608)) + cse_var_1) + (floormod(threadIdx.x_2, 24)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 56;
- kernel.shared_1[(threadIdx.x_2 + 56)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((floordiv(threadIdx.x_2, 8) + 7), 3)*4608)) + cse_var_1) + (floormod((threadIdx.x_2 + 8), 24)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 56;
- kernel.shared_1[(threadIdx.x_2 + 112)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((floordiv(threadIdx.x_2, 8) + 14), 3)*4608)) + cse_var_1) + (floormod((threadIdx.x_2 + 16), 24)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 56;
- kernel.shared_1[(threadIdx.x_2 + 168)] = kernel[((((((floordiv(blockIdx.x, 7)*147456) + (floordiv(floordiv(threadIdx.x_2, 8), 3)*4608)) + cse_var_1) + (floormod(threadIdx.x_2, 24)*3)) + rx.outer.outer) + 32256)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 56;
- kernel.shared_1[(threadIdx.x_2 + 224)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((floordiv(threadIdx.x_2, 8) + 28), 3)*4608)) + cse_var_1) + (floormod((threadIdx.x_2 + 8), 24)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 56;
- kernel.shared_1[(threadIdx.x_2 + 280)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((floordiv(threadIdx.x_2, 8) + 35), 3)*4608)) + cse_var_1) + (floormod((threadIdx.x_2 + 16), 24)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 56;
- kernel.shared_1[(threadIdx.x_2 + 336)] = kernel[((((((floordiv(blockIdx.x, 7)*147456) + (floordiv(floordiv(threadIdx.x_2, 8), 3)*4608)) + cse_var_1) + (floormod(threadIdx.x_2, 24)*3)) + rx.outer.outer) + 64512)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 56;
- kernel.shared_1[(threadIdx.x_2 + 392)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((floordiv(threadIdx.x_2, 8) + 49), 3)*4608)) + cse_var_1) + (floormod((threadIdx.x_2 + 8), 24)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 56;
- kernel.shared_1[(threadIdx.x_2 + 448)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((floordiv(threadIdx.x_2, 8) + 56), 3)*4608)) + cse_var_1) + (floormod((threadIdx.x_2 + 16), 24)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 56;
- kernel.shared_1[(threadIdx.x_2 + 504)] = kernel[((((((floordiv(blockIdx.x, 7)*147456) + (floordiv(floordiv(threadIdx.x_2, 8), 3)*4608)) + cse_var_1) + (floormod(threadIdx.x_2, 24)*3)) + rx.outer.outer) + 96768)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 56;
- kernel.shared_1[(threadIdx.x_2 + 560)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((floordiv(threadIdx.x_2, 8) + 70), 3)*4608)) + cse_var_1) + (floormod((threadIdx.x_2 + 8), 24)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 56;
- kernel.shared_1[(threadIdx.x_2 + 616)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((floordiv(threadIdx.x_2, 8) + 77), 3)*4608)) + cse_var_1) + (floormod((threadIdx.x_2 + 16), 24)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 56;
- kernel.shared_1[(threadIdx.x_2 + 672)] = kernel[((((((floordiv(blockIdx.x, 7)*147456) + (floordiv(floordiv(threadIdx.x_2, 8), 3)*4608)) + cse_var_1) + (floormod(threadIdx.x_2, 24)*3)) + rx.outer.outer) + 129024)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 56;
- if @tir.likely((threadIdx.x_2 < 40), dtype=bool) {
- kernel.shared_1[(threadIdx.x_2 + 728)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((floordiv(threadIdx.x_2, 8) + 91), 3)*4608)) + cse_var_1) + (floormod((threadIdx.x_2 + 8), 24)*3)) + rx.outer.outer)]
- }
- for (rc.outer.inner: int32, 0, 8) {
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((rc.outer.inner*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[((floordiv(threadIdx.x, 7)*96) + (rc.outer.inner*3))]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*96) + (rc.outer.inner*3)) + 1)]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*96) + (rc.outer.inner*3)) + 2)]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((rc.outer.inner*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*96) + (rc.outer.inner*3)) + 24)]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*96) + (rc.outer.inner*3)) + 25)]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*96) + (rc.outer.inner*3)) + 26)]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((rc.outer.inner*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*96) + (rc.outer.inner*3)) + 48)]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*96) + (rc.outer.inner*3)) + 49)]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*96) + (rc.outer.inner*3)) + 50)]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((rc.outer.inner*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*96) + (rc.outer.inner*3)) + 72)]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*96) + (rc.outer.inner*3)) + 73)]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*96) + (rc.outer.inner*3)) + 74)]))
- }
+ let cse_var_2: int32 = (rc.outer.outer*392)
+ let cse_var_1: int32 = (rc.outer.outer*72)
+ {
+ attr [IterVar(threadIdx.x_1: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ pad_temp.shared_1: Buffer(pad_temp.shared, float32, [648], [], scope="shared")[threadIdx.x_1] = @tir.if_then_else(((((9 <= floormod(threadIdx.x_1, 81)) && (floormod(threadIdx.x_1, 81) < 72)) && (1 <= floormod(threadIdx.x_1, 9))) && (floormod(threadIdx.x_1, 9) < 8)), data[((((cse_var_2 + (floordiv(threadIdx.x_1, 81)*49)) + (floordiv(floormod(threadIdx.x_1, 81), 9)*7)) + floormod(threadIdx.x_1, 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ pad_temp.shared_1[(threadIdx.x_1 + 112)] = @tir.if_then_else(((((9 <= floormod((threadIdx.x_1 + 112), 81)) && (floormod((threadIdx.x_1 + 31), 81) < 72)) && (1 <= floormod((threadIdx.x_1 + 4), 9))) && (floormod((threadIdx.x_1 + 4), 9) < 8)), data[((((cse_var_2 + (floordiv((threadIdx.x_1 + 112), 81)*49)) + (floordiv(floormod((threadIdx.x_1 + 112), 81), 9)*7)) + floormod((threadIdx.x_1 + 4), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ pad_temp.shared_1[(threadIdx.x_1 + 224)] = @tir.if_then_else(((((9 <= floormod((threadIdx.x_1 + 224), 81)) && (floormod((threadIdx.x_1 + 62), 81) < 72)) && (1 <= floormod((threadIdx.x_1 + 8), 9))) && (floormod((threadIdx.x_1 + 8), 9) < 8)), data[((((cse_var_2 + (floordiv((threadIdx.x_1 + 224), 81)*49)) + (floordiv(floormod((threadIdx.x_1 + 224), 81), 9)*7)) + floormod((threadIdx.x_1 + 8), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ pad_temp.shared_1[(threadIdx.x_1 + 336)] = @tir.if_then_else(((((9 <= floormod((threadIdx.x_1 + 336), 81)) && (floormod((threadIdx.x_1 + 12), 81) < 72)) && (1 <= floormod((threadIdx.x_1 + 3), 9))) && (floormod((threadIdx.x_1 + 3), 9) < 8)), data[((((cse_var_2 + (floordiv((threadIdx.x_1 + 336), 81)*49)) + (floordiv(floormod((threadIdx.x_1 + 336), 81), 9)*7)) + floormod((threadIdx.x_1 + 3), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ pad_temp.shared_1[(threadIdx.x_1 + 448)] = @tir.if_then_else(((((9 <= floormod((threadIdx.x_1 + 448), 81)) && (floormod((threadIdx.x_1 + 43), 81) < 72)) && (1 <= floormod((threadIdx.x_1 + 7), 9))) && (floormod((threadIdx.x_1 + 7), 9) < 8)), data[((((cse_var_2 + (floordiv((threadIdx.x_1 + 448), 81)*49)) + (floordiv(floormod((threadIdx.x_1 + 448), 81), 9)*7)) + floormod((threadIdx.x_1 + 7), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ if @tir.likely((threadIdx.x_1 < 88), dtype=bool) {
+ pad_temp.shared_1[(threadIdx.x_1 + 560)] = @tir.if_then_else(((((9 <= floormod((threadIdx.x_1 + 560), 81)) && (floormod((threadIdx.x_1 + 74), 81) < 72)) && (1 <= floormod((threadIdx.x_1 + 2), 9))) && (floormod((threadIdx.x_1 + 2), 9) < 8)), data[((((cse_var_2 + (floordiv((threadIdx.x_1 + 560), 81)*49)) + (floordiv(floormod((threadIdx.x_1 + 560), 81), 9)*7)) + floormod((threadIdx.x_1 + 2), 9)) - 8)], 0f32, dtype=float32)
+ }
+ attr [IterVar(threadIdx.x_2: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1: Buffer(kernel.shared, float32, [4608], [], scope="shared")[threadIdx.x_2] = kernel[((((blockIdx.x*294912) + (floordiv(threadIdx.x_2, 72)*4608)) + cse_var_1) + floormod(threadIdx.x_2, 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 112)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 14), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 40), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 224)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 28), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 8), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 336)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 42), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 48), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 448)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 56), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 16), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 560)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 70), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 56), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 672)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 84), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 24), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 784)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 98), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 64), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 896)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 112), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 32), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 1008)] = kernel[(((((blockIdx.x*294912) + (floordiv(floordiv(threadIdx.x_2, 8), 9)*4608)) + cse_var_1) + floormod(threadIdx.x_2, 72)) + 64512)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 1120)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 140), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 40), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 1232)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 154), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 8), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 1344)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 168), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 48), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 1456)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 182), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 16), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 1568)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 196), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 56), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 1680)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 210), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 24), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 1792)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 224), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 64), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 1904)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 238), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 32), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 2016)] = kernel[(((((blockIdx.x*294912) + (floordiv(floordiv(threadIdx.x_2, 8), 9)*4608)) + cse_var_1) + floormod(threadIdx.x_2, 72)) + 129024)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 2128)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 266), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 40), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 2240)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 280), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 8), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 2352)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 294), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 48), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 2464)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 308), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 16), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 2576)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 322), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 56), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 2688)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 336), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 24), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 2800)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 350), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 64), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 2912)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 364), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 32), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 3024)] = kernel[(((((blockIdx.x*294912) + (floordiv(floordiv(threadIdx.x_2, 8), 9)*4608)) + cse_var_1) + floormod(threadIdx.x_2, 72)) + 193536)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 3136)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 392), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 40), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 3248)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 406), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 8), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 3360)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 420), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 48), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 3472)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 434), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 16), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 3584)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 448), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 56), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 3696)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 462), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 24), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 3808)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 476), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 64), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 3920)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 490), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 32), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 4032)] = kernel[(((((blockIdx.x*294912) + (floordiv(floordiv(threadIdx.x_2, 8), 9)*4608)) + cse_var_1) + floormod(threadIdx.x_2, 72)) + 258048)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 4144)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 518), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 40), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 4256)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 532), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 8), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 4368)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 546), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 48), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 4480)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 560), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 16), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ if @tir.likely((threadIdx.x_2 < 16), dtype=bool) {
+ kernel.shared_1[(threadIdx.x_2 + 4592)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 574), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 56), 72))]
+ }
+ for (rc.outer.inner: int32, 0, 2) {
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((rc.outer.inner*324) + floormod(threadIdx.x, 7))]*kernel.shared_1[((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36))]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[((rc.outer.inner*324) + floormod(threadIdx.x, 7))]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2304)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 1)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2305)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2306)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 81)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 9)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 81)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2313)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 82)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 10)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 82)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2314)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 83)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 11)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 83)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2315)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 162)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 18)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 162)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2322)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 163)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 19)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 163)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2323)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 164)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 20)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 164)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2324)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 243)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 27)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 243)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2331)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 244)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 28)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 244)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2332)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 245)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 29)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 245)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2333)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 9)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36))]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 9)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2304)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 10)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 1)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 10)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2305)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 11)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 11)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2306)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 90)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 9)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 90)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2313)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 91)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 10)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 91)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2314)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 92)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 11)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 92)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2315)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 171)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 18)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 171)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2322)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 172)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 19)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 172)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2323)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 173)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 20)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 173)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2324)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 27)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2331)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 28)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2332)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 29)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2333)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36))]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2304)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 1)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2305)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2306)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 9)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2313)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 10)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2314)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 11)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2315)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 18)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2322)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 19)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2323)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 20)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2324)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 27)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2331)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 28)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2332)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 29)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2333)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 27)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36))]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 27)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2304)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 28)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 1)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 28)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2305)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 29)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 29)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2306)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 108)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 9)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 108)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2313)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 109)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 10)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 109)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2314)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 110)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 11)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 110)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2315)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 18)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2322)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 19)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2323)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 20)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2324)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 270)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 27)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 270)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2331)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 271)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 28)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 271)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2332)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 272)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 29)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 272)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2333)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 36)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36))]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 36)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2304)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 37)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 1)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 37)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2305)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 38)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 38)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2306)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 117)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 9)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 117)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2313)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 118)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 10)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 118)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2314)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 119)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 11)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 119)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2315)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 198)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 18)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 198)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2322)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 199)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 19)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 199)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2323)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 200)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 20)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 200)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2324)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 279)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 27)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 279)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2331)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 280)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 28)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 280)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2332)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 281)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 29)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 281)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2333)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 45)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36))]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 45)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2304)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 46)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 1)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 46)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2305)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 47)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 47)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2306)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 9)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2313)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 10)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2314)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 11)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2315)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 207)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 18)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 207)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2322)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 208)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 19)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 208)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2323)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 209)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 20)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 209)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2324)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 288)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 27)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 288)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2331)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 289)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 28)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 289)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2332)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 290)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 29)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 290)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2333)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 54)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36))]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 54)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2304)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 55)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 1)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 55)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2305)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 56)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 56)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2306)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 135)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 9)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 135)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2313)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 136)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 10)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 136)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2314)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 137)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 11)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 137)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2315)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 216)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 18)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 216)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2322)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 217)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 19)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 217)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2323)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 218)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 20)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 218)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2324)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 297)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 27)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 297)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2331)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 298)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 28)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 298)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2332)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 299)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 29)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 299)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2333)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((rc.outer.inner*324) + floormod(threadIdx.x, 7))]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 72)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[((rc.outer.inner*324) + floormod(threadIdx.x, 7))]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2376)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 73)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2377)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 74)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2378)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 81)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 81)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 81)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2385)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 82)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 82)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 82)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2386)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 83)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 83)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 83)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2387)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 162)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 90)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 162)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2394)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 163)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 91)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 163)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2395)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 164)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 92)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 164)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2396)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 243)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 99)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 243)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2403)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 244)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 100)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 244)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2404)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 245)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 101)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 245)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2405)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 9)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 72)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 9)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2376)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 10)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 73)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 10)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2377)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 11)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 74)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 11)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2378)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 90)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 81)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 90)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2385)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 91)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 82)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 91)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2386)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 92)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 83)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 92)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2387)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 171)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 90)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 171)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2394)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 172)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 91)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 172)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2395)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 173)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 92)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 173)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2396)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 99)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2403)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 100)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2404)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 101)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2405)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 72)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2376)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 73)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2377)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 74)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2378)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 81)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2385)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 82)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2386)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 83)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2387)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 90)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2394)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 91)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2395)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 92)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2396)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 99)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2403)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 100)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2404)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 101)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2405)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 27)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 72)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 27)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2376)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 28)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 73)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 28)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2377)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 29)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 74)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 29)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2378)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 108)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 81)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 108)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2385)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 109)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 82)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 109)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2386)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 110)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 83)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 110)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2387)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 90)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2394)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 91)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2395)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 92)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2396)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 270)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 99)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 270)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2403)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 271)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 100)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 271)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2404)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 272)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 101)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 272)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2405)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 36)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 72)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 36)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2376)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 37)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 73)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 37)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2377)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 38)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 74)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 38)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2378)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 117)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 81)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 117)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2385)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 118)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 82)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 118)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2386)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 119)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 83)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 119)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2387)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 198)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 90)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 198)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2394)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 199)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 91)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 199)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2395)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 200)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 92)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 200)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2396)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 279)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 99)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 279)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2403)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 280)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 100)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 280)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2404)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 281)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 101)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 281)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2405)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 45)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 72)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 45)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2376)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 46)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 73)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 46)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2377)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 47)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 74)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 47)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2378)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 81)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2385)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 82)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2386)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 83)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2387)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 207)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 90)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 207)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2394)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 208)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 91)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 208)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2395)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 209)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 92)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 209)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2396)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 288)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 99)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 288)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2403)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 289)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 100)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 289)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2404)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 290)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 101)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 290)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2405)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 54)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 72)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 54)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2376)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 55)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 73)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 55)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2377)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 56)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 74)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 56)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2378)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 135)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 81)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 135)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2385)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 136)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 82)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 136)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2386)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 137)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 83)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 137)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2387)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 216)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 90)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 216)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2394)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 217)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 91)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 217)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2395)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 218)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 92)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 218)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2396)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 297)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 99)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 297)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2403)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 298)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 100)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 298)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2404)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 299)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 101)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 299)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2405)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 9)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 3)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 9)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2307)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 10)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 4)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 10)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2308)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 11)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 5)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 11)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2309)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 90)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 12)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 90)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2316)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 91)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 13)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 91)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2317)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 92)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 14)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 92)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2318)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 171)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 21)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 171)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2325)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 172)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 22)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 172)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2326)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 173)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 23)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 173)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2327)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 30)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2334)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 31)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2335)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 32)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2336)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 3)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2307)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 4)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2308)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 5)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2309)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 12)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2316)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 13)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2317)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 14)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2318)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 21)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2325)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 22)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2326)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 23)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2327)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 30)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2334)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 31)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2335)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 32)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2336)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 27)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 3)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 27)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2307)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 28)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 4)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 28)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2308)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 29)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 5)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 29)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2309)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 108)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 12)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 108)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2316)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 109)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 13)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 109)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2317)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 110)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 14)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 110)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2318)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 21)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2325)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 22)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2326)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 23)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2327)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 270)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 30)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 270)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2334)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 271)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 31)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 271)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2335)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 272)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 32)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 272)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2336)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 36)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 3)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 36)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2307)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 37)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 4)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 37)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2308)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 38)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 5)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 38)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2309)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 117)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 12)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 117)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2316)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 118)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 13)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 118)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2317)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 119)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 14)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 119)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2318)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 198)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 21)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 198)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2325)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 199)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 22)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 199)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2326)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 200)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 23)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 200)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2327)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 279)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 30)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 279)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2334)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 280)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 31)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 280)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2335)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 281)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 32)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 281)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2336)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 45)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 3)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 45)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2307)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 46)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 4)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 46)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2308)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 47)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 5)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 47)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2309)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 12)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2316)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 13)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2317)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 14)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2318)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 207)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 21)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 207)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2325)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 208)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 22)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 208)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2326)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 209)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 23)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 209)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2327)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 288)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 30)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 288)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2334)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 289)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 31)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 289)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2335)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 290)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 32)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 290)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2336)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 54)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 3)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 54)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2307)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 55)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 4)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 55)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2308)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 56)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 5)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 56)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2309)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 135)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 12)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 135)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2316)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 136)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 13)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 136)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2317)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 137)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 14)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 137)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2318)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 216)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 21)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 216)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2325)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 217)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 22)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 217)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2326)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 218)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 23)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 218)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2327)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 297)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 30)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 297)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2334)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 298)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 31)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 298)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2335)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 299)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 32)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 299)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2336)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 3)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2307)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 4)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2308)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 5)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2309)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 144)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 12)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 144)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2316)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 145)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 13)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 145)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2317)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 146)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 14)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 146)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2318)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 225)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 21)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 225)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2325)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 226)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 22)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 226)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2326)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 227)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 23)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 227)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2327)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 306)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 30)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 306)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2334)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 307)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 31)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 307)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2335)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 308)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 32)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 308)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2336)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 9)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 75)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 9)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2379)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 10)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 76)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 10)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2380)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 11)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 77)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 11)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2381)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 90)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 84)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 90)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2388)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 91)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 85)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 91)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2389)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 92)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 86)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 92)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2390)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 171)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 93)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 171)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2397)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 172)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 94)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 172)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2398)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 173)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 95)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 173)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2399)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 102)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2406)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 103)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2407)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 104)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2408)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 75)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2379)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 76)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2380)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 77)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2381)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 84)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2388)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 85)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2389)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 86)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2390)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 93)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2397)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 94)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2398)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 95)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2399)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 102)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2406)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 103)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2407)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 104)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2408)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 27)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 75)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 27)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2379)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 28)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 76)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 28)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2380)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 29)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 77)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 29)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2381)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 108)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 84)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 108)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2388)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 109)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 85)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 109)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2389)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 110)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 86)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 110)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2390)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 93)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2397)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 94)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2398)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 95)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2399)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 270)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 102)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 270)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2406)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 271)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 103)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 271)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2407)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 272)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 104)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 272)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2408)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 36)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 75)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 36)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2379)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 37)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 76)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 37)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2380)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 38)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 77)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 38)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2381)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 117)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 84)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 117)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2388)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 118)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 85)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 118)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2389)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 119)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 86)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 119)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2390)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 198)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 93)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 198)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2397)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 199)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 94)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 199)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2398)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 200)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 95)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 200)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2399)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 279)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 102)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 279)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2406)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 280)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 103)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 280)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2407)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 281)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 104)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 281)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2408)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 45)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 75)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 45)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2379)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 46)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 76)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 46)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2380)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 47)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 77)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 47)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2381)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 84)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2388)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 85)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2389)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 86)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2390)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 207)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 93)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 207)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2397)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 208)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 94)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 208)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2398)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 209)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 95)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 209)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2399)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 288)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 102)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 288)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2406)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 289)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 103)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 289)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2407)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 290)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 104)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 290)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2408)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 54)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 75)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 54)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2379)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 55)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 76)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 55)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2380)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 56)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 77)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 56)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2381)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 135)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 84)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 135)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2388)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 136)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 85)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 136)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2389)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 137)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 86)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 137)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2390)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 216)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 93)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 216)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2397)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 217)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 94)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 217)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2398)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 218)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 95)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 218)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2399)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 297)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 102)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 297)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2406)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 298)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 103)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 298)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2407)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 299)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 104)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 299)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2408)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 75)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2379)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 76)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2380)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 77)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2381)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 144)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 84)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 144)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2388)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 145)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 85)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 145)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2389)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 146)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 86)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 146)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2390)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 225)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 93)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 225)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2397)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 226)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 94)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 226)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2398)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 227)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 95)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 227)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2399)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 306)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 102)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 306)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2406)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 307)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 103)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 307)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2407)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 308)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 104)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 308)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2408)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 6)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2310)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 7)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2311)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 8)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2312)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 15)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2319)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 16)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2320)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 17)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2321)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 24)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2328)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 25)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2329)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 26)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2330)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 33)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2337)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 34)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2338)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 35)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2339)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 27)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 6)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 27)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2310)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 28)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 7)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 28)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2311)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 29)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 8)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 29)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2312)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 108)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 15)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 108)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2319)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 109)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 16)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 109)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2320)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 110)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 17)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 110)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2321)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 24)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2328)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 25)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2329)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 26)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2330)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 270)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 33)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 270)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2337)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 271)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 34)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 271)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2338)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 272)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 35)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 272)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2339)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 36)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 6)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 36)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2310)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 37)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 7)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 37)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2311)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 38)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 8)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 38)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2312)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 117)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 15)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 117)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2319)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 118)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 16)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 118)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2320)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 119)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 17)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 119)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2321)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 198)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 24)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 198)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2328)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 199)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 25)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 199)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2329)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 200)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 26)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 200)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2330)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 279)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 33)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 279)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2337)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 280)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 34)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 280)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2338)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 281)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 35)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 281)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2339)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 45)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 6)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 45)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2310)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 46)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 7)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 46)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2311)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 47)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 8)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 47)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2312)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 15)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2319)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 16)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2320)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 17)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2321)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 207)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 24)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 207)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2328)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 208)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 25)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 208)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2329)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 209)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 26)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 209)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2330)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 288)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 33)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 288)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2337)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 289)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 34)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 289)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2338)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 290)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 35)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 290)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2339)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 54)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 6)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 54)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2310)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 55)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 7)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 55)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2311)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 56)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 8)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 56)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2312)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 135)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 15)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 135)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2319)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 136)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 16)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 136)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2320)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 137)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 17)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 137)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2321)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 216)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 24)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 216)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2328)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 217)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 25)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 217)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2329)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 218)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 26)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 218)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2330)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 297)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 33)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 297)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2337)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 298)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 34)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 298)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2338)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 299)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 35)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 299)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2339)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 6)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2310)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 7)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2311)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 8)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2312)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 144)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 15)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 144)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2319)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 145)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 16)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 145)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2320)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 146)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 17)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 146)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2321)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 225)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 24)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 225)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2328)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 226)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 25)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 226)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2329)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 227)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 26)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 227)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2330)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 306)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 33)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 306)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2337)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 307)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 34)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 307)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2338)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 308)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 35)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 308)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2339)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 72)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 6)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 72)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2310)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 73)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 7)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 73)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2311)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 74)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 8)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 74)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2312)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 153)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 15)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 153)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2319)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 154)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 16)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 154)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2320)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 155)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 17)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 155)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2321)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 234)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 24)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 234)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2328)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 235)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 25)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 235)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2329)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 236)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 26)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 236)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2330)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 315)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 33)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 315)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2337)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 316)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 34)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 316)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2338)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 317)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 35)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 317)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2339)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 78)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2382)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 79)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2383)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 80)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2384)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 87)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2391)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 88)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2392)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 89)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2393)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 96)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2400)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 97)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2401)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 98)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2402)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 105)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2409)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 106)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2410)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 107)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2411)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 27)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 78)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 27)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2382)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 28)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 79)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 28)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2383)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 29)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 80)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 29)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2384)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 108)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 87)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 108)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2391)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 109)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 88)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 109)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2392)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 110)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 89)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 110)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2393)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 96)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2400)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 97)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2401)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 98)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2402)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 270)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 105)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 270)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2409)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 271)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 106)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 271)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2410)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 272)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 107)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 272)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2411)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 36)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 78)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 36)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2382)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 37)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 79)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 37)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2383)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 38)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 80)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 38)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2384)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 117)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 87)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 117)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2391)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 118)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 88)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 118)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2392)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 119)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 89)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 119)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2393)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 198)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 96)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 198)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2400)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 199)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 97)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 199)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2401)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 200)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 98)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 200)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2402)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 279)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 105)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 279)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2409)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 280)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 106)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 280)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2410)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 281)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 107)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 281)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2411)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 45)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 78)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 45)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2382)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 46)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 79)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 46)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2383)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 47)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 80)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 47)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2384)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 87)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2391)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 88)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2392)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 89)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2393)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 207)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 96)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 207)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2400)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 208)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 97)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 208)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2401)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 209)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 98)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 209)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2402)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 288)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 105)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 288)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2409)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 289)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 106)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 289)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2410)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 290)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 107)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 290)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2411)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 54)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 78)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 54)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2382)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 55)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 79)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 55)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2383)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 56)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 80)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 56)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2384)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 135)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 87)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 135)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2391)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 136)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 88)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 136)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2392)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 137)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 89)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 137)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2393)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 216)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 96)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 216)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2400)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 217)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 97)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 217)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2401)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 218)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 98)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 218)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2402)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 297)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 105)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 297)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2409)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 298)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 106)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 298)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2410)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 299)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 107)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 299)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2411)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 78)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2382)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 79)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2383)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 80)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2384)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 144)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 87)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 144)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2391)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 145)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 88)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 145)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2392)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 146)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 89)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 146)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2393)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 225)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 96)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 225)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2400)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 226)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 97)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 226)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2401)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 227)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 98)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 227)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2402)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 306)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 105)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 306)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2409)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 307)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 106)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 307)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2410)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 308)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 107)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 308)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2411)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 72)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 78)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 72)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2382)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 73)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 79)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 73)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2383)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 74)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 80)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 74)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2384)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 153)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 87)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 153)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2391)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 154)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 88)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 154)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2392)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 155)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 89)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 155)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2393)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 234)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 96)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 234)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2400)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 235)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 97)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 235)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2401)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 236)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 98)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 236)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2402)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 315)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 105)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 315)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2409)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 316)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 106)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 316)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2410)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 317)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 107)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 317)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2411)]))
}
}
}
- for (i1.inner: int32, 0, 4) {
- compute[(((((floordiv(blockIdx.x, 7)*1568) + (floordiv(threadIdx.x, 7)*196)) + (i1.inner*49)) + (floormod(threadIdx.x, 7)*7)) + floormod(blockIdx.x, 7))] = max((conv2d_nchw_1[i1.inner] + bias[(((floordiv(blockIdx.x, 7)*32) + (floordiv(threadIdx.x, 7)*4)) + i1.inner)]), 0f32)
+ for (i1.inner: int32, 0, 2) {
+ for (i2.inner: int32, 0, 7) {
+ let cse_var_3: int32 = ((i1.inner*7) + i2.inner)
+ {
+ compute[(((((blockIdx.x*3136) + (floordiv(threadIdx.x, 7)*98)) + (i1.inner*49)) + (i2.inner*7)) + floormod(threadIdx.x, 7))] = max((conv2d_nchw_1[cse_var_3] + bias[(((blockIdx.x*64) + (floordiv(threadIdx.x, 7)*2)) + i1.inner)]), 0f32)
+ compute[((((((blockIdx.x*3136) + (floordiv(threadIdx.x, 7)*98)) + (i1.inner*49)) + (i2.inner*7)) + floormod(threadIdx.x, 7)) + 1568)] = max((conv2d_nchw_1[(cse_var_3 + 14)] + bias[((((blockIdx.x*64) + (floordiv(threadIdx.x, 7)*2)) + i1.inner) + 32)]), 0f32)
+ }
+ }
}
}
}
@@ -341,7 +1430,7 @@ We build the binary and check its correctness and performance.
.. code-block:: none
- Execution time of this operator: 0.469 ms
+ Execution time of this operator: 0.340 ms
@@ -386,35 +1475,35 @@ They can be used for debugging and learning the behavior of the auto-scheduler.
conv2d_nchw_nn_o_o_o_i, conv2d_nchw_nn_o_o_i = s[conv2d_nchw].split(conv2d_nchw_nn_o_o_i, factor=1)
conv2d_nchw_nn_o_o_o_o, conv2d_nchw_nn_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_nn_o_o_o_i, factor=1)
conv2d_nchw_ff_o_i, conv2d_nchw_ff_i = s[conv2d_nchw].split(conv2d_nchw_ff, factor=1)
- conv2d_nchw_ff_o_o_i, conv2d_nchw_ff_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_i, factor=4)
- conv2d_nchw_ff_o_o_o_i, conv2d_nchw_ff_o_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_o_i, factor=8)
- conv2d_nchw_ff_o_o_o_o, conv2d_nchw_ff_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_o_o_i, factor=1)
+ conv2d_nchw_ff_o_o_i, conv2d_nchw_ff_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_i, factor=2)
+ conv2d_nchw_ff_o_o_o_i, conv2d_nchw_ff_o_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_o_i, factor=16)
+ conv2d_nchw_ff_o_o_o_o, conv2d_nchw_ff_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_o_o_i, factor=2)
conv2d_nchw_yy_o_i, conv2d_nchw_yy_i = s[conv2d_nchw].split(conv2d_nchw_yy, factor=1)
- conv2d_nchw_yy_o_o_i, conv2d_nchw_yy_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_i, factor=1)
- conv2d_nchw_yy_o_o_o_i, conv2d_nchw_yy_o_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_o_i, factor=7)
+ conv2d_nchw_yy_o_o_i, conv2d_nchw_yy_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_i, factor=7)
+ conv2d_nchw_yy_o_o_o_i, conv2d_nchw_yy_o_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_o_i, factor=1)
conv2d_nchw_yy_o_o_o_o, conv2d_nchw_yy_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_o_o_i, factor=1)
conv2d_nchw_xx_o_i, conv2d_nchw_xx_i = s[conv2d_nchw].split(conv2d_nchw_xx, factor=1)
conv2d_nchw_xx_o_o_i, conv2d_nchw_xx_o_i = s[conv2d_nchw].split(conv2d_nchw_xx_o_i, factor=1)
- conv2d_nchw_xx_o_o_o_i, conv2d_nchw_xx_o_o_i = s[conv2d_nchw].split(conv2d_nchw_xx_o_o_i, factor=1)
+ conv2d_nchw_xx_o_o_o_i, conv2d_nchw_xx_o_o_i = s[conv2d_nchw].split(conv2d_nchw_xx_o_o_i, factor=7)
conv2d_nchw_xx_o_o_o_o, conv2d_nchw_xx_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_xx_o_o_o_i, factor=1)
- conv2d_nchw_rc_o_i, conv2d_nchw_rc_i = s[conv2d_nchw].split(conv2d_nchw_rc, factor=1)
- conv2d_nchw_rc_o_o, conv2d_nchw_rc_o_i = s[conv2d_nchw].split(conv2d_nchw_rc_o_i, factor=8)
- conv2d_nchw_ry_o_i, conv2d_nchw_ry_i = s[conv2d_nchw].split(conv2d_nchw_ry, factor=3)
- conv2d_nchw_ry_o_o, conv2d_nchw_ry_o_i = s[conv2d_nchw].split(conv2d_nchw_ry_o_i, factor=1)
- conv2d_nchw_rx_o_i, conv2d_nchw_rx_i = s[conv2d_nchw].split(conv2d_nchw_rx, factor=1)
+ conv2d_nchw_rc_o_i, conv2d_nchw_rc_i = s[conv2d_nchw].split(conv2d_nchw_rc, factor=4)
+ conv2d_nchw_rc_o_o, conv2d_nchw_rc_o_i = s[conv2d_nchw].split(conv2d_nchw_rc_o_i, factor=2)
+ conv2d_nchw_ry_o_i, conv2d_nchw_ry_i = s[conv2d_nchw].split(conv2d_nchw_ry, factor=1)
+ conv2d_nchw_ry_o_o, conv2d_nchw_ry_o_i = s[conv2d_nchw].split(conv2d_nchw_ry_o_i, factor=3)
+ conv2d_nchw_rx_o_i, conv2d_nchw_rx_i = s[conv2d_nchw].split(conv2d_nchw_rx, factor=3)
conv2d_nchw_rx_o_o, conv2d_nchw_rx_o_i = s[conv2d_nchw].split(conv2d_nchw_rx_o_i, factor=1)
s[conv2d_nchw].reorder(conv2d_nchw_nn_o_o_o_o, conv2d_nchw_ff_o_o_o_o, conv2d_nchw_yy_o_o_o_o, conv2d_nchw_xx_o_o_o_o, conv2d_nchw_nn_o_o_o_i, conv2d_nchw_ff_o_o_o_i, conv2d_nchw_yy_o_o_o_i, conv2d_nchw_xx_o_o_o_i, conv2d_nchw_nn_o_o_i, conv2d_nchw_ff_o_o_i, conv2d_nchw_yy_o_o_i, conv2d_nchw_xx_o_o_i, conv2d_nchw_rc_o_o, conv2d_nchw_ry_o_o, conv2d_nchw_rx_o_o, conv2d_nchw_rc_o_i, conv2d_nchw_ry_o_i, conv2d_nchw_rx_o_i, conv2d_nchw_nn_o_i, conv2d_nchw_ff_o_i, conv2d_nchw_yy_o_i, conv2 [...]
compute_i0_o_i, compute_i0_i = s[compute].split(compute_i0, factor=1)
compute_i0_o_o_i, compute_i0_o_i = s[compute].split(compute_i0_o_i, factor=1)
compute_i0_o_o_o, compute_i0_o_o_i = s[compute].split(compute_i0_o_o_i, factor=1)
- compute_i1_o_i, compute_i1_i = s[compute].split(compute_i1, factor=4)
- compute_i1_o_o_i, compute_i1_o_i = s[compute].split(compute_i1_o_i, factor=8)
- compute_i1_o_o_o, compute_i1_o_o_i = s[compute].split(compute_i1_o_o_i, factor=1)
- compute_i2_o_i, compute_i2_i = s[compute].split(compute_i2, factor=1)
- compute_i2_o_o_i, compute_i2_o_i = s[compute].split(compute_i2_o_i, factor=7)
+ compute_i1_o_i, compute_i1_i = s[compute].split(compute_i1, factor=2)
+ compute_i1_o_o_i, compute_i1_o_i = s[compute].split(compute_i1_o_i, factor=16)
+ compute_i1_o_o_o, compute_i1_o_o_i = s[compute].split(compute_i1_o_o_i, factor=2)
+ compute_i2_o_i, compute_i2_i = s[compute].split(compute_i2, factor=7)
+ compute_i2_o_o_i, compute_i2_o_i = s[compute].split(compute_i2_o_i, factor=1)
compute_i2_o_o_o, compute_i2_o_o_i = s[compute].split(compute_i2_o_o_i, factor=1)
compute_i3_o_i, compute_i3_i = s[compute].split(compute_i3, factor=1)
- compute_i3_o_o_i, compute_i3_o_i = s[compute].split(compute_i3_o_i, factor=1)
+ compute_i3_o_o_i, compute_i3_o_i = s[compute].split(compute_i3_o_i, factor=7)
compute_i3_o_o_o, compute_i3_o_o_i = s[compute].split(compute_i3_o_o_i, factor=1)
s[compute].reorder(compute_i0_o_o_o, compute_i1_o_o_o, compute_i2_o_o_o, compute_i3_o_o_o, compute_i0_o_o_i, compute_i1_o_o_i, compute_i2_o_o_i, compute_i3_o_o_i, compute_i0_o_i, compute_i1_o_i, compute_i2_o_i, compute_i3_o_i, compute_i0_i, compute_i1_i, compute_i2_i, compute_i3_i)
s[conv2d_nchw].compute_at(s[compute], compute_i3_o_i)
@@ -434,14 +1523,14 @@ They can be used for debugging and learning the behavior of the auto-scheduler.
kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused = s[kernel_shared].fuse(kernel_shared_ax0, kernel_shared_ax1, kernel_shared_ax2, kernel_shared_ax3)
kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_i = s[kernel_shared].split(kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused, factor=1)
s[kernel_shared].vectorize(kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_i)
- kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_o, kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i = s[kernel_shared].split(kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, factor=56)
+ kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_o, kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i = s[kernel_shared].split(kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, factor=112)
s[kernel_shared].bind(kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i, te.thread_axis("threadIdx.x"))
pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused = s[pad_temp_shared].fuse(pad_temp_shared_ax0, pad_temp_shared_ax1, pad_temp_shared_ax2, pad_temp_shared_ax3)
pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_i = s[pad_temp_shared].split(pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused, factor=1)
s[pad_temp_shared].vectorize(pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_i)
- pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_o, pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i = s[pad_temp_shared].split(pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, factor=56)
+ pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_o, pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i = s[pad_temp_shared].split(pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, factor=112)
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", 64)
+ 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:
@@ -459,56 +1548,1109 @@ They can be used for debugging and learning the behavior of the auto-scheduler.
#define int64_t long long
#define uint64_t unsigned long long
#endif
- extern "C" __global__ void __launch_bounds__(56) default_function_kernel0(float* __restrict__ data, float* __restrict__ kernel, float* __restrict__ compute, float* __restrict__ bias) {
- float conv2d_nchw[4];
- __shared__ float pad_temp_shared[72];
- __shared__ float kernel_shared[768];
+ extern "C" __global__ void __launch_bounds__(112) default_function_kernel0(float* __restrict__ data, float* __restrict__ kernel, float* __restrict__ compute, float* __restrict__ bias) {
+ float conv2d_nchw[28];
+ __shared__ float pad_temp_shared[648];
+ __shared__ float kernel_shared[4608];
conv2d_nchw[0] = 0.000000e+00f;
+ conv2d_nchw[14] = 0.000000e+00f;
conv2d_nchw[1] = 0.000000e+00f;
+ conv2d_nchw[15] = 0.000000e+00f;
conv2d_nchw[2] = 0.000000e+00f;
+ conv2d_nchw[16] = 0.000000e+00f;
conv2d_nchw[3] = 0.000000e+00f;
+ conv2d_nchw[17] = 0.000000e+00f;
+ conv2d_nchw[4] = 0.000000e+00f;
+ conv2d_nchw[18] = 0.000000e+00f;
+ conv2d_nchw[5] = 0.000000e+00f;
+ conv2d_nchw[19] = 0.000000e+00f;
+ conv2d_nchw[6] = 0.000000e+00f;
+ conv2d_nchw[20] = 0.000000e+00f;
+ conv2d_nchw[7] = 0.000000e+00f;
+ conv2d_nchw[21] = 0.000000e+00f;
+ conv2d_nchw[8] = 0.000000e+00f;
+ conv2d_nchw[22] = 0.000000e+00f;
+ conv2d_nchw[9] = 0.000000e+00f;
+ conv2d_nchw[23] = 0.000000e+00f;
+ conv2d_nchw[10] = 0.000000e+00f;
+ conv2d_nchw[24] = 0.000000e+00f;
+ conv2d_nchw[11] = 0.000000e+00f;
+ conv2d_nchw[25] = 0.000000e+00f;
+ conv2d_nchw[12] = 0.000000e+00f;
+ conv2d_nchw[26] = 0.000000e+00f;
+ conv2d_nchw[13] = 0.000000e+00f;
+ conv2d_nchw[27] = 0.000000e+00f;
for (int rc_outer_outer = 0; rc_outer_outer < 64; ++rc_outer_outer) {
- for (int rx_outer_outer = 0; rx_outer_outer < 3; ++rx_outer_outer) {
- __syncthreads();
- pad_temp_shared[((int)threadIdx.x)] = (((((1 <= (((int)threadIdx.x) % 9)) && ((((int)threadIdx.x) % 9) < 8)) && (1 <= (rx_outer_outer + (((int)blockIdx.x) % 7)))) && ((rx_outer_outer + (((int)blockIdx.x) % 7)) < 8)) ? data[((((((rc_outer_outer * 392) + ((((int)threadIdx.x) / 9) * 49)) + ((((int)threadIdx.x) % 9) * 7)) + rx_outer_outer) + (((int)blockIdx.x) % 7)) - 8)] : 0.000000e+00f);
- if (((int)threadIdx.x) < 16) {
- pad_temp_shared[(((int)threadIdx.x) + 56)] = (((((1 <= ((((int)threadIdx.x) + 2) % 9)) && (((((int)threadIdx.x) + 2) % 9) < 8)) && (1 <= (rx_outer_outer + (((int)blockIdx.x) % 7)))) && ((rx_outer_outer + (((int)blockIdx.x) % 7)) < 8)) ? data[((((((rc_outer_outer * 392) + (((((int)threadIdx.x) + 56) / 9) * 49)) + (((((int)threadIdx.x) + 2) % 9) * 7)) + rx_outer_outer) + (((int)blockIdx.x) % 7)) - 8)] : 0.000000e+00f);
- }
- kernel_shared[((int)threadIdx.x)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + ((((int)threadIdx.x) / 24) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) % 24) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 56)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 56) / 24) * 4608)) + (rc_outer_outer * 72)) + (((((int)threadIdx.x) + 8) % 24) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 112)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 112) / 24) * 4608)) + (rc_outer_outer * 72)) + (((((int)threadIdx.x) + 16) % 24) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 168)] = kernel[(((((((((int)blockIdx.x) / 7) * 147456) + ((((int)threadIdx.x) / 24) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) % 24) * 3)) + rx_outer_outer) + 32256)];
- kernel_shared[(((int)threadIdx.x) + 224)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 224) / 24) * 4608)) + (rc_outer_outer * 72)) + (((((int)threadIdx.x) + 8) % 24) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 280)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 280) / 24) * 4608)) + (rc_outer_outer * 72)) + (((((int)threadIdx.x) + 16) % 24) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 336)] = kernel[(((((((((int)blockIdx.x) / 7) * 147456) + ((((int)threadIdx.x) / 24) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) % 24) * 3)) + rx_outer_outer) + 64512)];
- kernel_shared[(((int)threadIdx.x) + 392)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 392) / 24) * 4608)) + (rc_outer_outer * 72)) + (((((int)threadIdx.x) + 8) % 24) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 448)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 448) / 24) * 4608)) + (rc_outer_outer * 72)) + (((((int)threadIdx.x) + 16) % 24) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 504)] = kernel[(((((((((int)blockIdx.x) / 7) * 147456) + ((((int)threadIdx.x) / 24) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) % 24) * 3)) + rx_outer_outer) + 96768)];
- kernel_shared[(((int)threadIdx.x) + 560)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 560) / 24) * 4608)) + (rc_outer_outer * 72)) + (((((int)threadIdx.x) + 8) % 24) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 616)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 616) / 24) * 4608)) + (rc_outer_outer * 72)) + (((((int)threadIdx.x) + 16) % 24) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 672)] = kernel[(((((((((int)blockIdx.x) / 7) * 147456) + ((((int)threadIdx.x) / 24) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) % 24) * 3)) + rx_outer_outer) + 129024)];
- if (((int)threadIdx.x) < 40) {
- kernel_shared[(((int)threadIdx.x) + 728)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 728) / 24) * 4608)) + (rc_outer_outer * 72)) + (((((int)threadIdx.x) + 8) % 24) * 3)) + rx_outer_outer)];
- }
- __syncthreads();
- for (int rc_outer_inner = 0; rc_outer_inner < 8; ++rc_outer_inner) {
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((rc_outer_inner * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[(((((int)threadIdx.x) / 7) * 96) + (rc_outer_inner * 3))]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[((((((int)threadIdx.x) / 7) * 96) + (rc_outer_inner * 3)) + 1)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[((((((int)threadIdx.x) / 7) * 96) + (rc_outer_inner * 3)) + 2)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((rc_outer_inner * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[((((((int)threadIdx.x) / 7) * 96) + (rc_outer_inner * 3)) + 24)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[((((((int)threadIdx.x) / 7) * 96) + (rc_outer_inner * 3)) + 25)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[((((((int)threadIdx.x) / 7) * 96) + (rc_outer_inner * 3)) + 26)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((rc_outer_inner * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[((((((int)threadIdx.x) / 7) * 96) + (rc_outer_inner * 3)) + 48)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[((((((int)threadIdx.x) / 7) * 96) + (rc_outer_inner * 3)) + 49)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[((((((int)threadIdx.x) / 7) * 96) + (rc_outer_inner * 3)) + 50)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((rc_outer_inner * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[((((((int)threadIdx.x) / 7) * 96) + (rc_outer_inner * 3)) + 72)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[((((((int)threadIdx.x) / 7) * 96) + (rc_outer_inner * 3)) + 73)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[((((((int)threadIdx.x) / 7) * 96) + (rc_outer_inner * 3)) + 74)]));
- }
+ __syncthreads();
+ pad_temp_shared[((int)threadIdx.x)] = (((((9 <= (((int)threadIdx.x) % 81)) && ((((int)threadIdx.x) % 81) < 72)) && (1 <= (((int)threadIdx.x) % 9))) && ((((int)threadIdx.x) % 9) < 8)) ? data[(((((rc_outer_outer * 392) + ((((int)threadIdx.x) / 81) * 49)) + (((((int)threadIdx.x) % 81) / 9) * 7)) + (((int)threadIdx.x) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 112)] = (((((9 <= ((((int)threadIdx.x) + 31) % 81)) && (((((int)threadIdx.x) + 31) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 4) % 9))) && (((((int)threadIdx.x) + 4) % 9) < 8)) ? data[(((((rc_outer_outer * 392) + (((((int)threadIdx.x) + 112) / 81) * 49)) + ((((((int)threadIdx.x) + 31) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 4) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 224)] = (((((9 <= ((((int)threadIdx.x) + 62) % 81)) && (((((int)threadIdx.x) + 62) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 8) % 9))) && (((((int)threadIdx.x) + 8) % 9) < 8)) ? data[(((((rc_outer_outer * 392) + (((((int)threadIdx.x) + 224) / 81) * 49)) + ((((((int)threadIdx.x) + 62) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 8) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 336)] = (((((9 <= ((((int)threadIdx.x) + 12) % 81)) && (((((int)threadIdx.x) + 12) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 3) % 9))) && (((((int)threadIdx.x) + 3) % 9) < 8)) ? data[(((((rc_outer_outer * 392) + (((((int)threadIdx.x) + 336) / 81) * 49)) + ((((((int)threadIdx.x) + 12) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 3) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 448)] = (((((9 <= ((((int)threadIdx.x) + 43) % 81)) && (((((int)threadIdx.x) + 43) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 7) % 9))) && (((((int)threadIdx.x) + 7) % 9) < 8)) ? data[(((((rc_outer_outer * 392) + (((((int)threadIdx.x) + 448) / 81) * 49)) + ((((((int)threadIdx.x) + 43) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 7) % 9)) - 8)] : 0.000000e+00f);
+ if (((int)threadIdx.x) < 88) {
+ pad_temp_shared[(((int)threadIdx.x) + 560)] = (((((9 <= ((((int)threadIdx.x) + 74) % 81)) && (((((int)threadIdx.x) + 74) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 2) % 9))) && (((((int)threadIdx.x) + 2) % 9) < 8)) ? data[(((((rc_outer_outer * 392) + (((((int)threadIdx.x) + 560) / 81) * 49)) + ((((((int)threadIdx.x) + 74) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 2) % 9)) - 8)] : 0.000000e+00f);
+ }
+ kernel_shared[((int)threadIdx.x)] = kernel[((((((int)blockIdx.x) * 294912) + ((((int)threadIdx.x) / 72) * 4608)) + (rc_outer_outer * 72)) + (((int)threadIdx.x) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 112)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 112) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 40) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 224)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 224) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 8) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 336)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 336) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 48) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 448)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 448) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 16) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 560)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 560) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 56) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 672)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 672) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 24) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 784)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 784) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 64) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 896)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 896) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 32) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 1008)] = kernel[(((((((int)blockIdx.x) * 294912) + ((((int)threadIdx.x) / 72) * 4608)) + (rc_outer_outer * 72)) + (((int)threadIdx.x) % 72)) + 64512)];
+ kernel_shared[(((int)threadIdx.x) + 1120)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 1120) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 40) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 1232)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 1232) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 8) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 1344)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 1344) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 48) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 1456)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 1456) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 16) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 1568)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 1568) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 56) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 1680)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 1680) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 24) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 1792)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 1792) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 64) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 1904)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 1904) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 32) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 2016)] = kernel[(((((((int)blockIdx.x) * 294912) + ((((int)threadIdx.x) / 72) * 4608)) + (rc_outer_outer * 72)) + (((int)threadIdx.x) % 72)) + 129024)];
+ kernel_shared[(((int)threadIdx.x) + 2128)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 2128) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 40) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 2240)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 2240) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 8) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 2352)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 2352) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 48) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 2464)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 2464) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 16) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 2576)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 2576) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 56) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 2688)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 2688) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 24) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 2800)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 2800) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 64) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 2912)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 2912) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 32) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 3024)] = kernel[(((((((int)blockIdx.x) * 294912) + ((((int)threadIdx.x) / 72) * 4608)) + (rc_outer_outer * 72)) + (((int)threadIdx.x) % 72)) + 193536)];
+ kernel_shared[(((int)threadIdx.x) + 3136)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 3136) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 40) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 3248)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 3248) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 8) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 3360)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 3360) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 48) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 3472)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 3472) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 16) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 3584)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 3584) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 56) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 3696)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 3696) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 24) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 3808)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 3808) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 64) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 3920)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 3920) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 32) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 4032)] = kernel[(((((((int)blockIdx.x) * 294912) + ((((int)threadIdx.x) / 72) * 4608)) + (rc_outer_outer * 72)) + (((int)threadIdx.x) % 72)) + 258048)];
+ kernel_shared[(((int)threadIdx.x) + 4144)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 4144) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 40) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 4256)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 4256) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 8) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 4368)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 4368) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 48) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 4480)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 4480) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 16) % 72))];
+ if (((int)threadIdx.x) < 16) {
+ kernel_shared[(((int)threadIdx.x) + 4592)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 4592) / 72) * 4608)) + (rc_outer_outer * 72)) + (((int)threadIdx.x) + 56))];
+ }
+ __syncthreads();
+ for (int rc_outer_inner = 0; rc_outer_inner < 2; ++rc_outer_inner) {
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((rc_outer_inner * 324) + (((int)threadIdx.x) % 7))] * kernel_shared[(((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36))]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[((rc_outer_inner * 324) + (((int)threadIdx.x) % 7))] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2304)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 1)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2305)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2306)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 81)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 9)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 81)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2313)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 82)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 10)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 82)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2314)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 83)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 11)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 83)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2315)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 162)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 18)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 162)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2322)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 163)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 19)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 163)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2323)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 164)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 20)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 164)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2324)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 243)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 27)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 243)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2331)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 244)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 28)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 244)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2332)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 245)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 29)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 245)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2333)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[(((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36))]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2304)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 1)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2305)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2306)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 9)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2313)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 10)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2314)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 11)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2315)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 18)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2322)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 19)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2323)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 20)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2324)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 27)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2331)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 28)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2332)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 29)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2333)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[(((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36))]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2304)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 1)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2305)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2306)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 9)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2313)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 10)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2314)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 11)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2315)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 18)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2322)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 19)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2323)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 20)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2324)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 27)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2331)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 28)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2332)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 29)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2333)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[(((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36))]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2304)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 28)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 1)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 28)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2305)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 29)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 29)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2306)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 9)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2313)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 109)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 10)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 109)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2314)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 110)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 11)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 110)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2315)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 18)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2322)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 19)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2323)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 20)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2324)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 27)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2331)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 271)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 28)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 271)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2332)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 272)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 29)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 272)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2333)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[(((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36))]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2304)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 37)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 1)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 37)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2305)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 38)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 38)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2306)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 9)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2313)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 118)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 10)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 118)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2314)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 119)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 11)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 119)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2315)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 18)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2322)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 199)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 19)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 199)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2323)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 200)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 20)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 200)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2324)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 27)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2331)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 280)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 28)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 280)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2332)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 281)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 29)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 281)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2333)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[(((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36))]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2304)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 46)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 1)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 46)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2305)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 47)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 47)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2306)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 9)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2313)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 10)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2314)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 11)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2315)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 18)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2322)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 208)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 19)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 208)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2323)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 209)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 20)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 209)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2324)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 27)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2331)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 289)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 28)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 289)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2332)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 290)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 29)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 290)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2333)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[(((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36))]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2304)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 1)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2305)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 56)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 56)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2306)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 9)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2313)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 136)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 10)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 136)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2314)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 137)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 11)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 137)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2315)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 18)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2322)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 217)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 19)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 217)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2323)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 218)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 20)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 218)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2324)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 27)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2331)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 298)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 28)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 298)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2332)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 299)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 29)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 299)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2333)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((rc_outer_inner * 324) + (((int)threadIdx.x) % 7))] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 72)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[((rc_outer_inner * 324) + (((int)threadIdx.x) % 7))] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2376)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 73)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2377)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 74)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2378)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 81)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 81)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 81)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2385)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 82)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 82)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 82)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2386)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 83)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 83)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 83)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2387)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 162)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 90)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 162)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2394)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 163)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 91)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 163)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2395)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 164)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 92)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 164)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2396)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 243)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 99)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 243)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2403)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 244)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 100)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 244)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2404)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 245)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 101)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 245)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2405)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 72)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2376)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 73)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2377)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 74)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2378)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 81)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2385)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 82)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2386)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 83)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2387)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 90)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2394)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 91)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2395)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 92)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2396)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 99)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2403)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 100)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2404)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 101)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2405)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 72)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2376)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 73)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2377)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 74)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2378)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 81)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2385)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 82)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2386)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 83)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2387)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 90)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2394)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 91)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2395)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 92)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2396)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 99)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2403)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 100)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2404)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 101)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2405)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 72)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2376)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 28)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 73)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 28)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2377)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 29)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 74)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 29)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2378)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 81)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2385)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 109)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 82)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 109)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2386)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 110)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 83)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 110)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2387)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 90)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2394)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 91)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2395)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 92)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2396)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 99)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2403)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 271)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 100)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 271)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2404)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 272)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 101)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 272)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2405)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 72)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2376)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 37)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 73)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 37)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2377)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 38)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 74)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 38)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2378)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 81)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2385)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 118)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 82)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 118)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2386)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 119)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 83)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 119)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2387)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 90)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2394)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 199)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 91)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 199)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2395)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 200)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 92)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 200)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2396)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 99)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2403)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 280)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 100)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 280)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2404)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 281)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 101)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 281)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2405)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 72)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2376)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 46)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 73)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 46)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2377)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 47)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 74)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 47)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2378)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 81)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2385)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 82)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2386)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 83)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2387)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 90)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2394)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 208)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 91)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 208)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2395)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 209)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 92)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 209)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2396)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 99)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2403)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 289)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 100)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 289)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2404)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 290)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 101)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 290)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2405)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 72)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2376)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 73)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2377)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 56)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 74)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 56)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2378)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 81)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2385)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 136)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 82)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 136)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2386)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 137)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 83)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 137)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2387)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 90)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2394)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 217)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 91)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 217)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2395)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 218)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 92)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 218)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2396)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 99)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2403)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 298)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 100)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 298)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2404)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 299)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 101)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 299)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2405)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 3)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2307)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 4)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2308)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 5)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2309)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 12)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2316)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 13)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2317)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 14)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2318)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 21)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2325)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 22)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2326)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 23)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2327)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 30)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2334)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 31)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2335)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 32)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2336)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 3)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2307)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 4)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2308)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 5)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2309)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 12)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2316)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 13)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2317)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 14)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2318)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 21)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2325)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 22)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2326)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 23)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2327)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 30)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2334)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 31)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2335)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 32)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2336)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 3)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2307)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 28)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 4)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 28)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2308)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 29)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 5)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 29)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2309)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 12)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2316)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 109)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 13)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 109)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2317)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 110)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 14)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 110)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2318)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 21)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2325)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 22)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2326)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 23)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2327)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 30)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2334)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 271)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 31)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 271)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2335)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 272)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 32)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 272)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2336)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 3)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2307)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 37)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 4)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 37)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2308)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 38)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 5)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 38)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2309)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 12)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2316)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 118)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 13)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 118)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2317)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 119)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 14)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 119)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2318)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 21)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2325)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 199)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 22)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 199)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2326)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 200)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 23)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 200)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2327)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 30)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2334)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 280)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 31)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 280)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2335)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 281)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 32)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 281)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2336)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 3)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2307)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 46)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 4)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 46)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2308)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 47)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 5)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 47)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2309)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 12)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2316)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 13)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2317)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 14)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2318)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 21)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2325)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 208)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 22)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 208)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2326)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 209)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 23)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 209)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2327)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 30)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2334)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 289)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 31)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 289)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2335)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 290)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 32)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 290)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2336)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 3)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2307)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 4)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2308)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 56)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 5)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 56)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2309)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 12)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2316)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 136)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 13)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 136)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2317)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 137)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 14)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 137)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2318)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 21)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2325)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 217)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 22)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 217)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2326)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 218)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 23)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 218)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2327)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 30)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2334)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 298)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 31)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 298)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2335)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 299)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 32)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 299)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2336)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 3)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2307)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 4)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2308)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 5)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2309)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 144)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 12)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 144)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2316)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 145)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 13)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 145)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2317)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 146)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 14)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 146)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2318)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 225)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 21)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 225)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2325)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 226)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 22)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 226)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2326)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 227)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 23)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 227)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2327)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 306)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 30)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 306)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2334)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 307)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 31)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 307)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2335)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 308)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 32)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 308)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2336)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 75)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2379)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 76)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2380)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 77)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2381)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 84)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2388)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 85)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2389)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 86)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2390)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 93)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2397)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 94)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2398)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 95)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2399)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 102)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2406)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 103)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2407)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 104)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2408)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 75)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2379)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 76)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2380)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 77)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2381)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 84)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2388)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 85)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2389)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 86)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2390)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 93)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2397)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 94)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2398)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 95)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2399)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 102)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2406)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 103)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2407)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 104)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2408)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 75)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2379)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 28)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 76)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 28)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2380)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 29)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 77)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 29)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2381)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 84)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2388)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 109)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 85)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 109)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2389)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 110)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 86)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 110)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2390)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 93)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2397)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 94)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2398)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 95)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2399)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 102)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2406)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 271)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 103)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 271)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2407)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 272)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 104)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 272)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2408)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 75)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2379)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 37)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 76)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 37)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2380)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 38)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 77)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 38)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2381)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 84)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2388)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 118)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 85)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 118)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2389)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 119)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 86)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 119)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2390)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 93)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2397)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 199)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 94)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 199)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2398)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 200)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 95)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 200)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2399)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 102)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2406)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 280)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 103)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 280)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2407)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 281)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 104)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 281)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2408)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 75)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2379)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 46)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 76)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 46)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2380)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 47)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 77)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 47)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2381)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 84)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2388)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 85)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2389)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 86)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2390)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 93)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2397)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 208)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 94)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 208)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2398)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 209)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 95)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 209)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2399)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 102)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2406)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 289)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 103)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 289)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2407)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 290)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 104)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 290)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2408)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 75)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2379)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 76)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2380)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 56)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 77)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 56)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2381)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 84)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2388)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 136)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 85)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 136)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2389)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 137)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 86)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 137)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2390)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 93)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2397)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 217)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 94)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 217)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2398)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 218)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 95)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 218)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2399)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 102)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2406)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 298)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 103)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 298)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2407)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 299)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 104)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 299)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2408)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 75)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2379)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 76)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2380)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 77)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2381)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 144)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 84)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 144)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2388)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 145)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 85)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 145)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2389)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 146)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 86)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 146)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2390)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 225)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 93)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 225)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2397)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 226)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 94)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 226)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2398)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 227)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 95)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 227)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2399)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 306)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 102)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 306)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2406)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 307)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 103)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 307)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2407)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 308)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 104)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 308)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2408)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 6)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2310)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 7)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2311)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 8)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2312)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 15)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2319)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 16)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2320)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 17)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2321)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 24)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2328)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 25)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2329)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 26)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2330)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 33)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2337)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 34)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2338)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 35)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2339)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 6)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2310)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 28)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 7)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 28)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2311)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 29)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 8)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 29)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2312)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 15)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2319)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 109)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 16)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 109)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2320)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 110)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 17)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 110)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2321)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 24)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2328)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 25)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2329)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 26)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2330)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 33)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2337)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 271)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 34)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 271)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2338)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 272)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 35)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 272)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2339)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 6)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2310)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 37)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 7)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 37)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2311)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 38)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 8)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 38)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2312)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 15)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2319)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 118)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 16)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 118)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2320)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 119)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 17)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 119)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2321)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 24)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2328)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 199)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 25)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 199)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2329)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 200)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 26)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 200)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2330)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 33)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2337)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 280)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 34)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 280)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2338)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 281)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 35)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 281)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2339)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 6)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2310)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 46)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 7)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 46)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2311)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 47)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 8)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 47)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2312)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 15)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2319)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 16)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2320)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 17)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2321)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 24)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2328)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 208)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 25)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 208)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2329)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 209)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 26)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 209)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2330)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 33)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2337)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 289)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 34)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 289)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2338)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 290)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 35)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 290)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2339)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 6)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2310)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 7)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2311)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 56)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 8)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 56)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2312)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 15)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2319)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 136)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 16)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 136)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2320)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 137)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 17)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 137)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2321)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 24)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2328)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 217)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 25)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 217)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2329)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 218)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 26)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 218)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2330)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 33)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2337)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 298)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 34)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 298)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2338)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 299)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 35)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 299)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2339)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 6)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2310)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 7)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2311)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 8)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2312)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 144)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 15)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 144)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2319)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 145)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 16)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 145)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2320)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 146)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 17)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 146)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2321)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 225)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 24)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 225)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2328)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 226)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 25)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 226)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2329)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 227)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 26)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 227)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2330)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 306)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 33)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 306)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2337)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 307)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 34)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 307)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2338)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 308)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 35)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 308)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2339)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 72)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 6)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 72)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2310)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 73)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 7)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 73)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2311)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 74)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 8)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 74)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2312)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 153)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 15)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 153)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2319)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 154)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 16)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 154)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2320)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 155)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 17)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 155)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2321)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 234)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 24)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 234)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2328)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 235)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 25)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 235)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2329)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 236)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 26)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 236)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2330)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 33)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2337)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 316)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 34)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 316)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2338)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 317)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 35)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 317)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2339)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 78)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2382)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 79)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2383)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 80)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2384)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 87)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2391)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 88)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2392)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 89)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2393)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 96)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2400)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 97)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2401)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 98)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2402)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 105)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2409)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 106)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2410)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 107)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2411)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 78)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2382)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 28)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 79)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 28)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2383)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 29)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 80)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 29)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2384)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 87)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2391)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 109)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 88)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 109)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2392)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 110)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 89)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 110)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2393)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 96)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2400)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 97)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2401)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 98)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2402)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 105)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2409)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 271)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 106)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 271)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2410)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 272)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 107)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 272)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2411)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 78)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2382)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 37)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 79)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 37)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2383)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 38)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 80)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 38)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2384)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 87)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2391)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 118)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 88)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 118)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2392)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 119)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 89)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 119)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2393)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 96)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2400)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 199)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 97)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 199)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2401)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 200)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 98)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 200)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2402)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 105)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2409)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 280)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 106)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 280)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2410)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 281)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 107)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 281)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2411)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 78)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2382)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 46)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 79)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 46)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2383)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 47)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 80)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 47)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2384)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 87)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2391)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 88)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2392)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 89)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2393)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 96)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2400)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 208)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 97)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 208)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2401)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 209)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 98)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 209)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2402)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 105)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2409)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 289)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 106)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 289)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2410)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 290)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 107)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 290)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2411)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 78)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2382)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 79)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2383)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 56)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 80)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 56)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2384)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 87)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2391)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 136)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 88)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 136)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2392)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 137)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 89)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 137)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2393)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 96)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2400)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 217)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 97)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 217)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2401)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 218)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 98)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 218)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2402)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 105)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2409)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 298)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 106)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 298)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2410)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 299)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 107)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 299)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2411)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 78)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2382)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 79)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2383)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 80)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2384)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 144)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 87)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 144)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2391)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 145)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 88)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 145)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2392)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 146)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 89)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 146)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2393)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 225)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 96)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 225)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2400)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 226)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 97)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 226)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2401)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 227)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 98)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 227)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2402)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 306)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 105)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 306)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2409)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 307)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 106)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 307)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2410)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 308)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 107)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 308)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2411)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 72)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 78)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 72)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2382)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 73)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 79)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 73)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2383)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 74)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 80)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 74)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2384)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 153)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 87)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 153)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2391)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 154)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 88)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 154)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2392)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 155)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 89)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 155)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2393)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 234)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 96)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 234)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2400)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 235)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 97)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 235)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2401)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 236)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 98)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 236)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2402)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 105)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2409)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 316)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 106)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 316)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2410)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 317)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 107)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 317)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2411)]));
}
}
- for (int i1_inner = 0; i1_inner < 4; ++i1_inner) {
- compute[((((((((int)blockIdx.x) / 7) * 1568) + ((((int)threadIdx.x) / 7) * 196)) + (i1_inner * 49)) + ((((int)threadIdx.x) % 7) * 7)) + (((int)blockIdx.x) % 7))] = max((conv2d_nchw[i1_inner] + bias[((((((int)blockIdx.x) / 7) * 32) + ((((int)threadIdx.x) / 7) * 4)) + i1_inner)]), 0.000000e+00f);
+ for (int i1_inner = 0; i1_inner < 2; ++i1_inner) {
+ for (int i2_inner = 0; i2_inner < 7; ++i2_inner) {
+ compute[(((((((int)blockIdx.x) * 3136) + ((((int)threadIdx.x) / 7) * 98)) + (i1_inner * 49)) + (i2_inner * 7)) + (((int)threadIdx.x) % 7))] = max((conv2d_nchw[((i1_inner * 7) + i2_inner)] + bias[(((((int)blockIdx.x) * 64) + ((((int)threadIdx.x) / 7) * 2)) + i1_inner)]), 0.000000e+00f);
+ compute[((((((((int)blockIdx.x) * 3136) + ((((int)threadIdx.x) / 7) * 98)) + (i1_inner * 49)) + (i2_inner * 7)) + (((int)threadIdx.x) % 7)) + 1568)] = max((conv2d_nchw[(((i1_inner * 7) + i2_inner) + 14)] + bias[((((((int)blockIdx.x) * 64) + ((((int)threadIdx.x) / 7) * 2)) + i1_inner) + 32)]), 0.000000e+00f);
+ }
}
}
@@ -567,7 +2709,7 @@ 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 21.084 seconds)
+ **Total running time of the script:** ( 2 minutes 21.977 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 6b76b74..c503794 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
@@ -614,7 +614,7 @@ 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)
- 9.6230 9.6119 9.6561 9.6010 0.0238
+ 9.6781 9.7014 9.7153 9.6174 0.0433
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 0772cfa..f3ecc8d 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
@@ -633,7 +633,7 @@ 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)
- 763.7144 765.4548 767.6819 758.0064 4.1373
+ 759.9460 761.3004 763.9244 754.6133 3.9200
@@ -658,7 +658,7 @@ Other Tips
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 19.840 seconds)
+ **Total running time of the script:** ( 1 minutes 19.340 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 8622833..a6f7645 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
@@ -362,73 +362,120 @@ 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} {
- for (i0.outer.i1.outer.fused: int32, 0, 128) "parallel" {
- allocate(compute_3: Pointer(global float32), float32, [512]), storage_scope = global {
- for (i.inner.init: int32, 0, 32) {
- let cse_var_1: int32 = (i.inner.init*16)
- {
- compute_4: Buffer(compute_3, float32, [512], [])[cse_var_1] = 0f32
- compute_4[(cse_var_1 + 1)] = 0f32
- compute_4[(cse_var_1 + 2)] = 0f32
- compute_4[(cse_var_1 + 3)] = 0f32
- compute_4[(cse_var_1 + 4)] = 0f32
- compute_4[(cse_var_1 + 5)] = 0f32
- compute_4[(cse_var_1 + 6)] = 0f32
- compute_4[(cse_var_1 + 7)] = 0f32
- compute_4[(cse_var_1 + 8)] = 0f32
- compute_4[(cse_var_1 + 9)] = 0f32
- compute_4[(cse_var_1 + 10)] = 0f32
- compute_4[(cse_var_1 + 11)] = 0f32
- compute_4[(cse_var_1 + 12)] = 0f32
- compute_4[(cse_var_1 + 13)] = 0f32
- compute_4[(cse_var_1 + 14)] = 0f32
- compute_4[(cse_var_1 + 15)] = 0f32
- }
- }
- for (elem_idx: int32, 0, let cse_var_2: int32 = floormod(i0.outer.i1.outer.fused, 32) in (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])) {
- for (i.inner: int32, 0, 32) {
- let cse_var_21: int32 = floormod(i0.outer.i1.outer.fused, 32)
- let cse_var_20: int32 = (i.inner*16)
- let cse_var_19: int32 = (elem_idx*16)
- let cse_var_18: int32 = (cse_var_20 + 10)
- let cse_var_17: int32 = (cse_var_20 + 11)
- let cse_var_16: int32 = (cse_var_20 + 12)
- let cse_var_15: int32 = (cse_var_20 + 13)
- let cse_var_14: int32 = (cse_var_20 + 14)
- let cse_var_13: int32 = (cse_var_20 + 15)
- let cse_var_12: int32 = (cse_var_20 + 2)
- let cse_var_11: int32 = (cse_var_20 + 3)
- let cse_var_10: int32 = (cse_var_20 + 4)
- let cse_var_9: int32 = (cse_var_20 + 5)
- let cse_var_8: int32 = (cse_var_20 + 6)
- let cse_var_7: int32 = (cse_var_20 + 7)
- let cse_var_6: int32 = (cse_var_20 + 8)
- let cse_var_5: int32 = (cse_var_20 + 9)
- let cse_var_4: int32 = (cse_var_20 + 1)
- let cse_var_3: int32 = ((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.inner*256))
+ for (i0.outer.i1.outer.fused: int32, 0, 32) "parallel" {
+ allocate(compute_3: Pointer(global float32), float32, [2048]), storage_scope = global {
+ for (i.outer.inner: int32, 0, 32) {
+ for (nb_j.inner: int32, 0, 2) {
+ let cse_var_2: int32 = ((i.outer.inner*64) + (nb_j.inner*16))
+ let cse_var_1: int32 = ((floormod(i0.outer.i1.outer.fused, 16)*2) + nb_j.inner)
{
- compute_4[cse_var_20] = (compute_4[cse_var_20] + (placeholder_1[((placeholder_3[cse_var_21]*16) + cse_var_19)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_21] + elem_idx)])], 0f32)))
- compute_4[cse_var_4] = (compute_4[cse_var_4] + (placeholder_1[(((placeholder_3[cse_var_21]*16) + cse_var_19) + 1)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_21] + elem_idx)])], 0f32)))
- compute_4[cse_var_12] = (compute_4[cse_var_12] + (placeholder_1[(((placeholder_3[cse_var_21]*16) + cse_var_19) + 2)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_21] + elem_idx)])], 0f32)))
- compute_4[cse_var_11] = (compute_4[cse_var_11] + (placeholder_1[(((placeholder_3[cse_var_21]*16) + cse_var_19) + 3)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_21] + elem_idx)])], 0f32)))
- compute_4[cse_var_10] = (compute_4[cse_var_10] + (placeholder_1[(((placeholder_3[cse_var_21]*16) + cse_var_19) + 4)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_21] + elem_idx)])], 0f32)))
- compute_4[cse_var_9] = (compute_4[cse_var_9] + (placeholder_1[(((placeholder_3[cse_var_21]*16) + cse_var_19) + 5)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_21] + elem_idx)])], 0f32)))
- compute_4[cse_var_8] = (compute_4[cse_var_8] + (placeholder_1[(((placeholder_3[cse_var_21]*16) + cse_var_19) + 6)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_21] + elem_idx)])], 0f32)))
- compute_4[cse_var_7] = (compute_4[cse_var_7] + (placeholder_1[(((placeholder_3[cse_var_21]*16) + cse_var_19) + 7)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_21] + elem_idx)])], 0f32)))
- compute_4[cse_var_6] = (compute_4[cse_var_6] + (placeholder_1[(((placeholder_3[cse_var_21]*16) + cse_var_19) + 8)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_21] + elem_idx)])], 0f32)))
- compute_4[cse_var_5] = (compute_4[cse_var_5] + (placeholder_1[(((placeholder_3[cse_var_21]*16) + cse_var_19) + 9)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_21] + elem_idx)])], 0f32)))
- compute_4[cse_var_18] = (compute_4[cse_var_18] + (placeholder_1[(((placeholder_3[cse_var_21]*16) + cse_var_19) + 10)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_21] + elem_idx)])], 0f32)))
- compute_4[cse_var_17] = (compute_4[cse_var_17] + (placeholder_1[(((placeholder_3[cse_var_21]*16) + cse_var_19) + 11)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_21] + elem_idx)])], 0f32)))
- compute_4[cse_var_16] = (compute_4[cse_var_16] + (placeholder_1[(((placeholder_3[cse_var_21]*16) + cse_var_19) + 12)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_21] + elem_idx)])], 0f32)))
- compute_4[cse_var_15] = (compute_4[cse_var_15] + (placeholder_1[(((placeholder_3[cse_var_21]*16) + cse_var_19) + 13)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_21] + elem_idx)])], 0f32)))
- compute_4[cse_var_14] = (compute_4[cse_var_14] + (placeholder_1[(((placeholder_3[cse_var_21]*16) + cse_var_19) + 14)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_21] + elem_idx)])], 0f32)))
- compute_4[cse_var_13] = (compute_4[cse_var_13] + (placeholder_1[(((placeholder_3[cse_var_21]*16) + cse_var_19) + 15)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_21] + elem_idx)])], 0f32)))
+ compute_4: Buffer(compute_3, float32, [2048], [])[cse_var_2] = 0f32
+ compute_4[(cse_var_2 + 1)] = 0f32
+ compute_4[(cse_var_2 + 2)] = 0f32
+ compute_4[(cse_var_2 + 3)] = 0f32
+ compute_4[(cse_var_2 + 4)] = 0f32
+ compute_4[(cse_var_2 + 5)] = 0f32
+ compute_4[(cse_var_2 + 6)] = 0f32
+ compute_4[(cse_var_2 + 7)] = 0f32
+ compute_4[(cse_var_2 + 8)] = 0f32
+ compute_4[(cse_var_2 + 9)] = 0f32
+ compute_4[(cse_var_2 + 10)] = 0f32
+ compute_4[(cse_var_2 + 11)] = 0f32
+ compute_4[(cse_var_2 + 12)] = 0f32
+ compute_4[(cse_var_2 + 13)] = 0f32
+ compute_4[(cse_var_2 + 14)] = 0f32
+ compute_4[(cse_var_2 + 15)] = 0f32
+ compute_4[(cse_var_2 + 32)] = 0f32
+ compute_4[(cse_var_2 + 33)] = 0f32
+ compute_4[(cse_var_2 + 34)] = 0f32
+ compute_4[(cse_var_2 + 35)] = 0f32
+ compute_4[(cse_var_2 + 36)] = 0f32
+ compute_4[(cse_var_2 + 37)] = 0f32
+ compute_4[(cse_var_2 + 38)] = 0f32
+ compute_4[(cse_var_2 + 39)] = 0f32
+ compute_4[(cse_var_2 + 40)] = 0f32
+ compute_4[(cse_var_2 + 41)] = 0f32
+ compute_4[(cse_var_2 + 42)] = 0f32
+ compute_4[(cse_var_2 + 43)] = 0f32
+ compute_4[(cse_var_2 + 44)] = 0f32
+ compute_4[(cse_var_2 + 45)] = 0f32
+ compute_4[(cse_var_2 + 46)] = 0f32
+ compute_4[(cse_var_2 + 47)] = 0f32
+ for (elem_idx: int32, 0, (placeholder_3[(cse_var_1 + 1)] - placeholder_3[cse_var_1])) {
+ let cse_var_35: int32 = (cse_var_2 + 1)
+ let cse_var_34: int32 = (cse_var_2 + 10)
+ let cse_var_33: int32 = (cse_var_2 + 11)
+ let cse_var_32: int32 = (cse_var_2 + 12)
+ let cse_var_31: int32 = (cse_var_2 + 13)
+ let cse_var_30: int32 = (cse_var_2 + 14)
+ let cse_var_29: int32 = (cse_var_2 + 15)
+ let cse_var_28: int32 = (cse_var_2 + 2)
+ let cse_var_27: int32 = (cse_var_2 + 3)
+ let cse_var_26: int32 = (cse_var_2 + 32)
+ let cse_var_25: int32 = (cse_var_2 + 33)
+ let cse_var_24: int32 = (cse_var_2 + 34)
+ let cse_var_23: int32 = (cse_var_2 + 35)
+ let cse_var_22: int32 = (cse_var_2 + 36)
+ let cse_var_21: int32 = (cse_var_2 + 37)
+ let cse_var_20: int32 = (cse_var_2 + 39)
+ let cse_var_19: int32 = (elem_idx*16)
+ let cse_var_18: int32 = (cse_var_2 + 9)
+ let cse_var_17: int32 = (cse_var_2 + 8)
+ let cse_var_16: int32 = (cse_var_2 + 7)
+ let cse_var_15: int32 = (cse_var_2 + 6)
+ let cse_var_14: int32 = (cse_var_2 + 5)
+ let cse_var_13: int32 = (cse_var_2 + 47)
+ let cse_var_12: int32 = (cse_var_2 + 38)
+ let cse_var_11: int32 = (cse_var_2 + 45)
+ let cse_var_10: int32 = (cse_var_2 + 44)
+ let cse_var_9: int32 = (cse_var_2 + 43)
+ let cse_var_8: int32 = (cse_var_2 + 42)
+ let cse_var_7: int32 = (cse_var_2 + 41)
+ let cse_var_6: int32 = (cse_var_2 + 40)
+ let cse_var_5: int32 = (cse_var_2 + 4)
+ let cse_var_4: int32 = (cse_var_2 + 46)
+ let cse_var_3: int32 = ((floordiv(i0.outer.i1.outer.fused, 16)*16384) + (i.outer.inner*512))
+ {
+ compute_4[cse_var_2] = (compute_4[cse_var_2] + (placeholder_1[((placeholder_3[cse_var_1]*16) + cse_var_19)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)])], 0f32)))
+ compute_4[cse_var_35] = (compute_4[cse_var_35] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 1)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)])], 0f32)))
+ compute_4[cse_var_28] = (compute_4[cse_var_28] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 2)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)])], 0f32)))
+ compute_4[cse_var_27] = (compute_4[cse_var_27] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 3)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)])], 0f32)))
+ compute_4[cse_var_5] = (compute_4[cse_var_5] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 4)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)])], 0f32)))
+ compute_4[cse_var_14] = (compute_4[cse_var_14] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 5)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)])], 0f32)))
+ compute_4[cse_var_15] = (compute_4[cse_var_15] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 6)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)])], 0f32)))
+ compute_4[cse_var_16] = (compute_4[cse_var_16] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 7)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)])], 0f32)))
+ compute_4[cse_var_17] = (compute_4[cse_var_17] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 8)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)])], 0f32)))
+ compute_4[cse_var_18] = (compute_4[cse_var_18] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 9)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)])], 0f32)))
+ compute_4[cse_var_34] = (compute_4[cse_var_34] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 10)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)])], 0f32)))
+ compute_4[cse_var_33] = (compute_4[cse_var_33] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 11)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)])], 0f32)))
+ compute_4[cse_var_32] = (compute_4[cse_var_32] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 12)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)])], 0f32)))
+ compute_4[cse_var_31] = (compute_4[cse_var_31] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 13)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)])], 0f32)))
+ compute_4[cse_var_30] = (compute_4[cse_var_30] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 14)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)])], 0f32)))
+ compute_4[cse_var_29] = (compute_4[cse_var_29] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 15)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)])], 0f32)))
+ compute_4[cse_var_26] = (compute_4[cse_var_26] + (placeholder_1[((placeholder_3[cse_var_1]*16) + cse_var_19)]*max(placeholder[((cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)]) + 256)], 0f32)))
+ compute_4[cse_var_25] = (compute_4[cse_var_25] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 1)]*max(placeholder[((cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)]) + 256)], 0f32)))
+ compute_4[cse_var_24] = (compute_4[cse_var_24] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 2)]*max(placeholder[((cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)]) + 256)], 0f32)))
+ compute_4[cse_var_23] = (compute_4[cse_var_23] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 3)]*max(placeholder[((cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)]) + 256)], 0f32)))
+ compute_4[cse_var_22] = (compute_4[cse_var_22] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 4)]*max(placeholder[((cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)]) + 256)], 0f32)))
+ compute_4[cse_var_21] = (compute_4[cse_var_21] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 5)]*max(placeholder[((cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)]) + 256)], 0f32)))
+ compute_4[cse_var_12] = (compute_4[cse_var_12] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 6)]*max(placeholder[((cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)]) + 256)], 0f32)))
+ compute_4[cse_var_20] = (compute_4[cse_var_20] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 7)]*max(placeholder[((cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)]) + 256)], 0f32)))
+ compute_4[cse_var_6] = (compute_4[cse_var_6] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 8)]*max(placeholder[((cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)]) + 256)], 0f32)))
+ compute_4[cse_var_7] = (compute_4[cse_var_7] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 9)]*max(placeholder[((cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)]) + 256)], 0f32)))
+ compute_4[cse_var_8] = (compute_4[cse_var_8] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 10)]*max(placeholder[((cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)]) + 256)], 0f32)))
+ compute_4[cse_var_9] = (compute_4[cse_var_9] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 11)]*max(placeholder[((cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)]) + 256)], 0f32)))
+ compute_4[cse_var_10] = (compute_4[cse_var_10] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 12)]*max(placeholder[((cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)]) + 256)], 0f32)))
+ compute_4[cse_var_11] = (compute_4[cse_var_11] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 13)]*max(placeholder[((cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)]) + 256)], 0f32)))
+ compute_4[cse_var_4] = (compute_4[cse_var_4] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 14)]*max(placeholder[((cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)]) + 256)], 0f32)))
+ compute_4[cse_var_13] = (compute_4[cse_var_13] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 15)]*max(placeholder[((cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)]) + 256)], 0f32)))
+ }
+ }
}
}
}
- for (i0.inner: int32, 0, 32) {
- let cse_var_22: int32 = (((floordiv(i0.outer.i1.outer.fused, 32)*16384) + (i0.inner*512)) + (floormod(i0.outer.i1.outer.fused, 32)*16))
- compute[ramp(cse_var_22, 1, 16)] = max((compute_4[ramp((i0.inner*16), 1, 16)] + placeholder_4[ramp(cse_var_22, 1, 16)]), broadcast(0f32, 16))
+ for (i0.inner: int32, 0, 64) {
+ let cse_var_36: int32 = (((floordiv(i0.outer.i1.outer.fused, 16)*32768) + (i0.inner*512)) + (floormod(i0.outer.i1.outer.fused, 16)*32))
+ compute[ramp(cse_var_36, 1, 32)] = max((compute_4[ramp((i0.inner*32), 1, 32)] + placeholder_4[ramp(cse_var_36, 1, 32)]), broadcast(0f32, 32))
}
}
}
@@ -482,7 +529,7 @@ We build the binary and check its correctness and performance.
.. code-block:: none
- Execution time of this operator: 1.733 ms
+ Execution time of this operator: 3.501 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 fe6ff30..7d7f5ef 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
=================
-**00:44.405** total execution time for **how_to_tune_with_autotvm** files:
+**00:44.528** total execution time for **how_to_tune_with_autotvm** files:
-- **00:43.515**: :ref:`sphx_glr_how_to_tune_with_autotvm_tune_conv2d_cuda.py` (``tune_conv2d_cuda.py``)
-- **00:00.232**: :ref:`sphx_glr_how_to_tune_with_autotvm_tune_relay_x86.py` (``tune_relay_x86.py``)
-- **00:00.223**: :ref:`sphx_glr_how_to_tune_with_autotvm_tune_relay_mobile_gpu.py` (``tune_relay_mobile_gpu.py``)
-- **00:00.219**: :ref:`sphx_glr_how_to_tune_with_autotvm_tune_relay_arm.py` (``tune_relay_arm.py``)
-- **00:00.216**: :ref:`sphx_glr_how_to_tune_with_autotvm_tune_relay_cuda.py` (``tune_relay_cuda.py``)
+- **00:43.676**: :ref:`sphx_glr_how_to_tune_with_autotvm_tune_conv2d_cuda.py` (``tune_conv2d_cuda.py``)
+- **00:00.222**: :ref:`sphx_glr_how_to_tune_with_autotvm_tune_relay_x86.py` (``tune_relay_x86.py``)
+- **00:00.215**: :ref:`sphx_glr_how_to_tune_with_autotvm_tune_relay_mobile_gpu.py` (``tune_relay_mobile_gpu.py``)
+- **00:00.212**: :ref:`sphx_glr_how_to_tune_with_autotvm_tune_relay_arm.py` (``tune_relay_arm.py``)
+- **00:00.202**: :ref:`sphx_glr_how_to_tune_with_autotvm_tune_relay_cuda.py` (``tune_relay_cuda.py``)
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 ef96d82..e8db18e 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
@@ -859,8 +859,8 @@ for this template
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 854, 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: 112.57/112.57 result: MeasureResult(costs=(0.002056472038961039,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.8501737117767334, timestamp=1648696103.9259007) [('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: 0.00/112.57 result: Traceback (most recent call last):
+ No: 6 GFLOPS: 97.59/97.59 result: MeasureResult(costs=(0.002372302375,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.609654426574707, timestamp=1648705752.068103) [('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: 0.00/97.59 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 571, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 523, in _build_func_common
@@ -983,7 +983,7 @@ for this template
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 854, 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: 0.00/112.57 result: Traceback (most recent call last):
+ No: 8 GFLOPS: 0.00/97.59 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 571, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 523, in _build_func_common
@@ -1106,7 +1106,7 @@ for this template
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 854, 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: 0.00/112.57 result: Traceback (most recent call last):
+ No: 9 GFLOPS: 0.00/97.59 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 571, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 523, in _build_func_common
@@ -1229,7 +1229,7 @@ for this template
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 854, 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: 0.00/112.57 result: Traceback (most recent call last):
+ No: 10 GFLOPS: 0.00/97.59 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
@@ -1247,7 +1247,7 @@ 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: 0.00/112.57 result: Traceback (most recent call last):
+ No: 11 GFLOPS: 0.00/97.59 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 571, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 523, in _build_func_common
@@ -1370,7 +1370,7 @@ for this template
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 854, 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: 0.00/112.57 result: Traceback (most recent call last):
+ No: 12 GFLOPS: 0.00/97.59 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 571, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 523, in _build_func_common
@@ -1493,7 +1493,7 @@ for this template
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 854, 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: 0.00/112.57 result: Traceback (most recent call last):
+ No: 13 GFLOPS: 0.00/97.59 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 571, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 523, in _build_func_common
@@ -1616,7 +1616,7 @@ for this template
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 854, 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: 0.00/112.57 result: Traceback (most recent call last):
+ No: 14 GFLOPS: 0.00/97.59 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 571, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 523, in _build_func_common
@@ -1739,7 +1739,7 @@ for this template
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 854, 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: 0.00/112.57 result: Traceback (most recent call last):
+ No: 15 GFLOPS: 0.00/97.59 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 571, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 523, in _build_func_common
@@ -1862,7 +1862,7 @@ for this template
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 854, 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: 0.00/112.57 result: Traceback (most recent call last):
+ No: 16 GFLOPS: 0.00/97.59 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 571, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 523, in _build_func_common
@@ -1985,7 +1985,7 @@ for this template
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 854, 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: 0.00/112.57 result: Traceback (most recent call last):
+ No: 17 GFLOPS: 0.00/97.59 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 571, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 523, in _build_func_common
@@ -2108,7 +2108,7 @@ for this template
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 854, 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: 0.00/112.57 result: Traceback (most recent call last):
+ No: 18 GFLOPS: 0.00/97.59 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 571, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 523, in _build_func_common
@@ -2231,7 +2231,7 @@ for this template
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 854, 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: 0.00/112.57 result: Traceback (most recent call last):
+ No: 19 GFLOPS: 0.00/97.59 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 721, in __call__
yield remote, remote.load_module(os.path.split(build_result.filename)[1])
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 685, in run_through_rpc
@@ -2319,7 +2319,7 @@ for this template
15: _PyEval_EvalFrameDefault
14: 0x0000000000537c30
13: _PyObject_FastCallKeywords
- 12: 0x00007ff01a5f6fa2
+ 12: 0x00007faf2622dfa2
11: _ctypes_callproc
10: ffi_call
9: ffi_call_unix64
@@ -2384,7 +2384,7 @@ 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: 144.81/144.81 result: MeasureResult(costs=(0.0015986197799999998,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.4213595390319824, timestamp=1648696130.3614411) [('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: 144.71/144.71 result: MeasureResult(costs=(0.00159980322,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.4247119426727295, timestamp=1648705778.465009) [('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
@@ -2437,7 +2437,7 @@ 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
- Time cost of this operator: 0.002018
+ Time cost of this operator: 0.002029
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 6a656ba..6158f1a 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
@@ -292,10 +292,10 @@ Timing the untuned program
########## Build without Autotuning ##########
Node Name Ops Time(us) Time(%) Shape Inputs Outputs
--------- --- -------- ------- ----- ------ -------
- tvmgen_default_fused_nn_contrib_conv2d_NCHWc tvmgen_default_fused_nn_contrib_conv2d_NCHWc 310.8 98.675 (1, 2, 10, 10, 3) 2 1
- tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 3.228 1.025 (1, 6, 10, 10) 1 1
- tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.944 0.3 (1, 1, 10, 10, 3) 1 1
- Total_time - 314.972 - - - -
+ tvmgen_default_fused_nn_contrib_conv2d_NCHWc tvmgen_default_fused_nn_contrib_conv2d_NCHWc 307.3 98.698 (1, 2, 10, 10, 3) 2 1
+ tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 3.141 1.009 (1, 6, 10, 10) 1 1
+ tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.912 0.293 (1, 1, 10, 10, 3) 1 1
+ Total_time - 311.352 - - - -
@@ -357,10 +357,10 @@ Timing the tuned program
########## Build with Autotuning ##########
Node Name Ops Time(us) Time(%) Shape Inputs Outputs
--------- --- -------- ------- ----- ------ -------
- tvmgen_default_fused_nn_contrib_conv2d_NCHWc tvmgen_default_fused_nn_contrib_conv2d_NCHWc 197.1 98.678 (1, 6, 10, 10, 1) 2 1
- tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 1.74 0.871 (1, 6, 10, 10) 1 1
- tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.9 0.451 (1, 3, 10, 10, 1) 1 1
- Total_time - 199.74 - - - -
+ tvmgen_default_fused_nn_contrib_conv2d_NCHWc tvmgen_default_fused_nn_contrib_conv2d_NCHWc 214.3 98.668 (1, 1, 10, 10, 6) 2 1
+ tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 1.944 0.895 (1, 6, 10, 10) 1 1
+ tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.948 0.437 (1, 3, 10, 10, 1) 1 1
+ Total_time - 217.192 - - - -
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 bc14940..9a87f52 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
@@ -5,10 +5,10 @@
Computation times
=================
-**00:44.623** total execution time for **how_to_work_with_microtvm** files:
+**00:44.518** total execution time for **how_to_work_with_microtvm** files:
-- **00:40.445**: :ref:`sphx_glr_how_to_work_with_microtvm_micro_autotune.py` (``micro_autotune.py``)
-- **00:03.560**: :ref:`sphx_glr_how_to_work_with_microtvm_micro_tflite.py` (``micro_tflite.py``)
-- **00:00.218**: :ref:`sphx_glr_how_to_work_with_microtvm_micro_tvmc.py` (``micro_tvmc.py``)
-- **00:00.200**: :ref:`sphx_glr_how_to_work_with_microtvm_micro_ethosu.py` (``micro_ethosu.py``)
-- **00:00.200**: :ref:`sphx_glr_how_to_work_with_microtvm_micro_reference_vm.py` (``micro_reference_vm.py``)
+- **00:40.235**: :ref:`sphx_glr_how_to_work_with_microtvm_micro_autotune.py` (``micro_autotune.py``)
+- **00:03.671**: :ref:`sphx_glr_how_to_work_with_microtvm_micro_tflite.py` (``micro_tflite.py``)
+- **00:00.211**: :ref:`sphx_glr_how_to_work_with_microtvm_micro_tvmc.py` (``micro_tvmc.py``)
+- **00:00.202**: :ref:`sphx_glr_how_to_work_with_microtvm_micro_ethosu.py` (``micro_ethosu.py``)
+- **00:00.199**: :ref:`sphx_glr_how_to_work_with_microtvm_micro_reference_vm.py` (``micro_reference_vm.py``)
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 0b2f426..f52c8e3 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
@@ -5,8 +5,8 @@
Computation times
=================
-**00:04.438** total execution time for **how_to_work_with_relay** files:
+**00:04.538** total execution time for **how_to_work_with_relay** files:
-- **00:02.526**: :ref:`sphx_glr_how_to_work_with_relay_using_external_lib.py` (``using_external_lib.py``)
-- **00:01.691**: :ref:`sphx_glr_how_to_work_with_relay_build_gcn.py` (``build_gcn.py``)
-- **00:00.221**: :ref:`sphx_glr_how_to_work_with_relay_using_relay_viz.py` (``using_relay_viz.py``)
+- **00:02.532**: :ref:`sphx_glr_how_to_work_with_relay_using_external_lib.py` (``using_external_lib.py``)
+- **00:01.809**: :ref:`sphx_glr_how_to_work_with_relay_build_gcn.py` (``build_gcn.py``)
+- **00:00.197**: :ref:`sphx_glr_how_to_work_with_relay_using_relay_viz.py` (``using_relay_viz.py``)
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 098190a..63438f9 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
@@ -5,13 +5,13 @@
Computation times
=================
-**00:06.049** total execution time for **how_to_work_with_schedules** files:
+**00:06.173** total execution time for **how_to_work_with_schedules** files:
-- **00:02.233**: :ref:`sphx_glr_how_to_work_with_schedules_intrin_math.py` (``intrin_math.py``)
-- **00:01.223**: :ref:`sphx_glr_how_to_work_with_schedules_tensorize.py` (``tensorize.py``)
-- **00:00.772**: :ref:`sphx_glr_how_to_work_with_schedules_scan.py` (``scan.py``)
-- **00:00.764**: :ref:`sphx_glr_how_to_work_with_schedules_reduction.py` (``reduction.py``)
-- **00:00.332**: :ref:`sphx_glr_how_to_work_with_schedules_extern_op.py` (``extern_op.py``)
-- **00:00.249**: :ref:`sphx_glr_how_to_work_with_schedules_schedule_primitives.py` (``schedule_primitives.py``)
-- **00:00.242**: :ref:`sphx_glr_how_to_work_with_schedules_tedd.py` (``tedd.py``)
-- **00:00.233**: :ref:`sphx_glr_how_to_work_with_schedules_tuple_inputs.py` (``tuple_inputs.py``)
+- **00:02.272**: :ref:`sphx_glr_how_to_work_with_schedules_intrin_math.py` (``intrin_math.py``)
+- **00:01.333**: :ref:`sphx_glr_how_to_work_with_schedules_tensorize.py` (``tensorize.py``)
+- **00:00.771**: :ref:`sphx_glr_how_to_work_with_schedules_scan.py` (``scan.py``)
+- **00:00.763**: :ref:`sphx_glr_how_to_work_with_schedules_reduction.py` (``reduction.py``)
+- **00:00.324**: :ref:`sphx_glr_how_to_work_with_schedules_extern_op.py` (``extern_op.py``)
+- **00:00.239**: :ref:`sphx_glr_how_to_work_with_schedules_schedule_primitives.py` (``schedule_primitives.py``)
+- **00:00.238**: :ref:`sphx_glr_how_to_work_with_schedules_tuple_inputs.py` (``tuple_inputs.py``)
+- **00:00.231**: :ref:`sphx_glr_how_to_work_with_schedules_tedd.py` (``tedd.py``)
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 3ca4bce..255ea17 100644
--- a/docs/_sources/how_to/work_with_schedules/tensorize.rst.txt
+++ b/docs/_sources/how_to/work_with_schedules/tensorize.rst.txt
@@ -314,8 +314,8 @@ The importing needs to happen before the tensorized GEMV being executed.
B: Buffer(B_2: Pointer(float32), float32, [32768], []),
C: Buffer(C_2: Pointer(float32), float32, [524288], [])}
buffer_map = {A_1: A, B_1: B, C_1: C} {
- attr [IterVar(i: int32, (nullptr), "DataPar", "")] "pragma_import_llvm" = "; ModuleID = '/tmp/tmp768o3bb6/input0.cc'
- source_filename = "/tmp/tmp768o3bb6/input0.cc"
+ attr [IterVar(i: int32, (nullptr), "DataPar", "")] "pragma_import_llvm" = "; ModuleID = '/tmp/tmpy9m74646/input0.cc'
+ source_filename = "/tmp/tmpy9m74646/input0.cc"
target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
target triple = "x86_64-pc-linux-gnu"
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 4d99f40..e0a1850 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
@@ -5,7 +5,7 @@
Computation times
=================
-**00:20.065** total execution time for **topic_vta_tutorials_autotvm** files:
+**00:19.863** total execution time for **topic_vta_tutorials_autotvm** files:
-- **00:19.853**: :ref:`sphx_glr_topic_vta_tutorials_autotvm_tune_relay_vta.py` (``tune_relay_vta.py``)
-- **00:00.213**: :ref:`sphx_glr_topic_vta_tutorials_autotvm_tune_alu_vta.py` (``tune_alu_vta.py``)
+- **00:19.657**: :ref:`sphx_glr_topic_vta_tutorials_autotvm_tune_relay_vta.py` (``tune_relay_vta.py``)
+- **00:00.206**: :ref:`sphx_glr_topic_vta_tutorials_autotvm_tune_alu_vta.py` (``tune_alu_vta.py``)
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 fdbff84..07e4149 100644
--- a/docs/_sources/topic/vta/tutorials/frontend/deploy_classification.rst.txt
+++ b/docs/_sources/topic/vta/tutorials/frontend/deploy_classification.rst.txt
@@ -265,7 +265,7 @@ 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 21.11s!
+ resnet18_v1 inference graph built in 21.17s!
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 74ec53f..8e90585 100644
--- a/docs/_sources/topic/vta/tutorials/frontend/deploy_detection.rst.txt
+++ b/docs/_sources/topic/vta/tutorials/frontend/deploy_detection.rst.txt
@@ -301,7 +301,7 @@ The compilation steps are:
/workspace/python/tvm/relay/build_module.py:439: DeprecationWarning: Please use input parameter mod (tvm.IRModule) instead of deprecated parameter mod (tvm.relay.function.Function)
DeprecationWarning,
- yolov3-tiny inference graph built in 14.52s!
+ yolov3-tiny inference graph built in 14.75s!
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 4ea6f90..3f1eb83 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
@@ -5,7 +5,7 @@
Computation times
=================
-**01:28.066** total execution time for **topic_vta_tutorials_frontend** files:
+**01:28.688** total execution time for **topic_vta_tutorials_frontend** files:
-- **00:46.656**: :ref:`sphx_glr_topic_vta_tutorials_frontend_deploy_detection.py` (``deploy_detection.py``)
-- **00:41.410**: :ref:`sphx_glr_topic_vta_tutorials_frontend_deploy_classification.py` (``deploy_classification.py``)
+- **00:46.999**: :ref:`sphx_glr_topic_vta_tutorials_frontend_deploy_detection.py` (``deploy_detection.py``)
+- **00:41.689**: :ref:`sphx_glr_topic_vta_tutorials_frontend_deploy_classification.py` (``deploy_classification.py``)
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 a1f00ed..13a1e6b 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
@@ -5,7 +5,7 @@
Computation times
=================
-**00:03.554** total execution time for **topic_vta_tutorials_optimize** files:
+**00:03.641** total execution time for **topic_vta_tutorials_optimize** files:
-- **00:02.979**: :ref:`sphx_glr_topic_vta_tutorials_optimize_convolution_opt.py` (``convolution_opt.py``)
-- **00:00.575**: :ref:`sphx_glr_topic_vta_tutorials_optimize_matrix_multiply_opt.py` (``matrix_multiply_opt.py``)
+- **00:03.049**: :ref:`sphx_glr_topic_vta_tutorials_optimize_convolution_opt.py` (``convolution_opt.py``)
+- **00:00.592**: :ref:`sphx_glr_topic_vta_tutorials_optimize_matrix_multiply_opt.py` (``matrix_multiply_opt.py``)
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 fb0c862..e2f12ec 100644
--- a/docs/_sources/topic/vta/tutorials/sg_execution_times.rst.txt
+++ b/docs/_sources/topic/vta/tutorials/sg_execution_times.rst.txt
@@ -5,7 +5,7 @@
Computation times
=================
-**00:01.062** total execution time for **topic_vta_tutorials** files:
+**00:01.101** total execution time for **topic_vta_tutorials** files:
-- **00:00.539**: :ref:`sphx_glr_topic_vta_tutorials_matrix_multiply.py` (``matrix_multiply.py``)
-- **00:00.523**: :ref:`sphx_glr_topic_vta_tutorials_vta_get_started.py` (``vta_get_started.py``)
+- **00:00.554**: :ref:`sphx_glr_topic_vta_tutorials_matrix_multiply.py` (``matrix_multiply.py``)
+- **00:00.547**: :ref:`sphx_glr_topic_vta_tutorials_vta_get_started.py` (``vta_get_started.py``)
diff --git a/docs/_sources/tutorial/auto_scheduler_matmul_x86.rst.txt b/docs/_sources/tutorial/auto_scheduler_matmul_x86.rst.txt
index 70fc767..6ce3030 100644
--- a/docs/_sources/tutorial/auto_scheduler_matmul_x86.rst.txt
+++ b/docs/_sources/tutorial/auto_scheduler_matmul_x86.rst.txt
@@ -305,7 +305,7 @@ We build the binary and check its correctness and performance.
.. code-block:: none
- Execution time of this operator: 91.454 ms
+ Execution time of this operator: 93.368 ms
@@ -401,7 +401,7 @@ resume the status and do more 5 trials.
Resume search:
/usr/local/lib/python3.7/dist-packages/xgboost/training.py:17: UserWarning: Old style callback is deprecated. See: https://xgboost.readthedocs.io/en/latest/python/callbacks.html
warnings.warn(f'Old style callback is deprecated. See: {link}', UserWarning)
-
+ *E
@@ -414,6 +414,11 @@ Expression (TE) language that demonstrates how TVM can optimize computational
operations.
+.. rst-class:: sphx-glr-timing
+
+ **Total running time of the script:** ( 1 minutes 17.751 seconds)
+
+
.. _sphx_glr_download_tutorial_auto_scheduler_matmul_x86.py:
diff --git a/docs/_sources/tutorial/autotvm_relay_x86.rst.txt b/docs/_sources/tutorial/autotvm_relay_x86.rst.txt
index a5586f4..5b9bdba 100644
--- a/docs/_sources/tutorial/autotvm_relay_x86.rst.txt
+++ b/docs/_sources/tutorial/autotvm_relay_x86.rst.txt
@@ -268,7 +268,7 @@ standard deviation.
.. code-block:: none
- {'mean': 492.0994797199995, 'median': 492.4136981499999, 'std': 0.8176268748468473}
+ {'mean': 492.5393293900015, 'median': 493.21396980003556, 'std': 1.366467218342447}
@@ -482,31 +482,31 @@ the tuning data to.
.. code-block:: none
-
[Task 1/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 1/25] Current/Best: 14.68/ 17.16 GFLOPS | Progress: (4/10) | 6.09 s
[Task 1/25] Current/Best: 16.91/ 17.52 GFLOPS | Progress: (8/10) | 9.09 s
[Task 1/25] Current/Best: 24.16/ 24.16 GFLOPS | Progress: (10/10) | 9.77 s Done.
-
[Task 2/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 2/25] Current/Best: 14.21/ 14.21 GFLOPS | Progress: (4/10) | 2.78 s
[Task 2/25] Current/Best: 16.79/ 16.79 GFLOPS | Progress: (8/10) | 4.26 s
[Task 2/25] Current/Best: 16.29/ 20.27 GFLOPS | Progress: (10/10) | 4.79 s Done.
-
[Task 3/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 3/25] Current/Best: 7.34/ 16.96 GFLOPS | Progress: (4/10) | 3.91 s
[Task 3/25] Current/Best: 10.48/ 17.97 GFLOPS | Progress: (8/10) | 6.04 s
[Task 3/25] Current/Best: 23.97/ 23.97 GFLOPS | Progress: (10/10) | 6.85 s Done.
-
[Task 4/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 4/25] Current/Best: 10.01/ 10.02 GFLOPS | Progress: (4/10) | 10.04 s
[Task 4/25] Current/Best: 6.56/ 17.54 GFLOPS | Progress: (8/10) | 11.67 s
[Task 4/25] Current/Best: 13.14/ 17.54 GFLOPS | Progress: (10/10) | 12.50 s Done.
-
[Task 5/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 5/25] Current/Best: 17.78/ 17.78 GFLOPS | Progress: (4/10) | 2.57 s
[Task 5/25] Current/Best: 5.34/ 20.12 GFLOPS | Progress: (8/10) | 3.97 s
[Task 5/25] Current/Best: 18.13/ 20.12 GFLOPS | Progress: (10/10) | 4.71 s Done.
-
[Task 6/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 6/25] Current/Best: 5.41/ 17.66 GFLOPS | Progress: (4/10) | 3.68 s
[Task 6/25] Current/Best: 21.14/ 21.15 GFLOPS | Progress: (8/10) | 5.73 s
[Task 6/25] Current/Best: 10.14/ 21.15 GFLOPS | Progress: (10/10) | 6.63 s Done.
-
[Task 7/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 7/25] Current/Best: 16.61/ 16.61 GFLOPS | Progress: (4/10) | 3.52 s
[Task 7/25] Current/Best: 14.40/ 23.63 GFLOPS | Progress: (8/10) | 5.51 s
[Task 7/25] Current/Best: 3.09/ 23.63 GFLOPS | Progress: (10/10) | 7.05 s Done.
-
[Task 8/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 8/25] Current/Best: 10.49/ 15.25 GFLOPS | Progress: (4/10) | 7.04 s
[Task 8/25] Current/Best: 13.08/ 15.25 GFLOPS | Progress: (8/10) | 9.95 s
[Task 8/25] Current/Best: 15.00/ 15.25 GFLOPS | Progress: (10/10) | 11.22 s Done.
-
[Task 9/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 9/25] Current/Best: 12.54/ 13.33 GFLOPS | Progress: (4/10) | 6.17 s
[Task 9/25] Current/Best: 5.16/ 13.33 GFLOPS | Progress: (8/10) | 8.44 s
[Task 9/25] Current/Best: 15.26/ 15.26 GFLOPS | Progress: (10/10) | 11.28 s Done.
-
[Task 10/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 10/25] Current/Best: 15.72/ 22.52 GFLOPS | Progress: (4/10) | 2.98 s
[Task 10/25] Current/Best: 9.85/ 22.52 GFLOPS | Progress: (8/10) | 4.48 s
[Task 10/25] Current/Best: 16.49/ 22.52 GFLOPS | Progress: (10/10) | 5.37 s Done.
-
[Task 11/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 11/25] Current/Best: 12.13/ 21.01 GFLOPS | Progress: (4/10) | 2.67 s
[Task 11/25] Current/Best: 12.39/ 21.01 GFLOPS | Progress: (8/10) | 5.07 s
[Task 11/25] Current/Best: 17.16/ 21.01 GFLOPS | Progress: (10/10) | 6.01 s Done.
-
[Task 12/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 12/25] Current/Best: 9.45/ 14.54 GFLOPS | Progress: (4/10) | 5.72 s
[Task 12/25] Current/Best: 10.77/ 19.21 GFLOPS | Progress: (8/10) | 8.96 s
[Task 12/25] Current/Best: 11.66/ 19.21 GFLOPS | Progress: (10/10) | 12.47 s Done.
-
[Task 13/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 13/25] Current/Best: 22.48/ 22.48 GFLOPS | Progress: (4/10) | 3.51 s
[Task 13/25] Current/Best: 13.66/ 22.48 GFLOPS | Progress: (8/10) | 6.44 s
[Task 13/25] Current/Best: 18.25/ 22.48 GFLOPS | Progress: (10/10) | 7.51 s Done.
-
[Task 14/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 14/25] Current/Best: 16.17/ 18.68 GFLOPS | Progress: (4/10) | 2.96 s
[Task 14/25] Current/Best: 5.49/ 18.68 GFLOPS | Progress: (8/10) | 5.58 s
[Task 14/25] Current/Best: 13.81/ 18.68 GFLOPS | Progress: (10/10) | 7.45 s Done.
-
[Task 15/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 15/25] Current/Best: 12.15/ 14.94 GFLOPS | Progress: (4/10) | 5.44 s
[Task 15/25] Current/Best: 10.82/ 19.09 GFLOPS | Progress: (8/10) | 7.01 s
[Task 15/25] Current/Best: 8.69/ 19.09 GFLOPS | Progress: (10/10) | 12.72 s
[Task 16/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 16/25] Current/Best: 22.29/ 22.29 GFLOPS | Progress: (4/10) | 2.56 s
[Task 16/25] Current/Best: 9.47/ 22.29 GFLOPS | Progress: (8/10) | 4.45 s
[Task 16/25] Current/Best: 13.41/ 22.29 GFLOPS | Progress: (10/10) | 5.70 s Done.
-
[Task 17/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 17/25] Current/Best: 6.15/ 22.61 GFLOPS | Progress: (4/10) | 2.92 s
[Task 17/25] Current/Best: 1.56/ 22.61 GFLOPS | Progress: (8/10) | 5.99 s
[Task 17/25] Current/Best: 9.21/ 22.61 GFLOPS | Progress: (10/10) | 6.90 s Done.
-
[Task 18/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 18/25] Current/Best: 18.22/ 18.60 GFLOPS | Progress: (4/10) | 2.57 s
[Task 18/25] Current/Best: 18.38/ 20.35 GFLOPS | Progress: (8/10) | 5.75 s
[Task 18/25] Current/Best: 9.34/ 20.35 GFLOPS | Progress: (10/10) | 11.06 s Done.
-
[Task 19/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 19/25] Current/Best: 2.70/ 20.18 GFLOPS | Progress: (4/10) | 5.35 s
[Task 19/25] Current/Best: 6.07/ 23.32 GFLOPS | Progress: (8/10) | 8.30 s
[Task 19/25] Current/Best: 20.95/ 23.32 GFLOPS | Progress: (10/10) | 9.80 s Done.
-
[Task 20/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 20/25] Current/Best: 5.25/ 17.59 GFLOPS | Progress: (4/10) | 2.92 s
[Task 20/25] Current/Best: 7.44/ 17.59 GFLOPS | Progress: (8/10) | 8.21 s
[Task 20/25] Current/Best: 9.39/ 17.59 GFLOPS | Progress: (10/10) | 14.76 s Done.
-
[Task 21/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 21/25] Current/Best: 13.48/ 14.97 GFLOPS | Progress: (4/10) | 3.02 s Done.
-
[Task 21/25] Current/Best: 13.76/ 16.82 GFLOPS | Progress: (8/10) | 4.79 s
[Task 21/25] Current/Best: 5.32/ 16.82 GFLOPS | Progress: (10/10) | 5.59 s Done.
-
[Task 22/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 22/25] Current/Best: 14.38/ 14.55 GFLOPS | Progress: (4/10) | 3.02 s
[Task 22/25] Current/Best: 4.32/ 17.90 GFLOPS | Progress: (8/10) | 4.95 s
[Task 22/25] Current/Best: 12.50/ 17.90 GFLOPS | Progress: (10/10) | 5.87 s Done.
-
[Task 23/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 23/25] Current/Best: 7.96/ 20.26 GFLOPS | Progress: (4/10) | 4.13 s
[Task 23/25] Current/Best: 19.26/ 20.26 GFLOPS | Progress: (8/10) | 7.90 s
[Task 23/25] Current/Best: 12.12/ 20.26 GFLOPS | Progress: (10/10) | 9.48 s Done.
-
[Task 24/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 24/25] Current/Best: 3.16/ 4.15 GFLOPS | Progress: (4/10) | 83.92 s
[Task 24/25] Current/Best: 6.04/ 7.55 GFLOPS | Progress: (8/10) | 88.51 s
[Task 24/25] Current/Best: 4.16/ 7.55 GFLOPS | Progress: (10/10) | 7221.37 s
[Task 25/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 25/25] Current/Best: 1.50/ 8.25 GFLOPS | Progress: (4/10) | 18.54 s Done.
-
[Task 25/25] Current/Best: 9.96/ 9.96 GFLOPS | Progress: (8/10) | 35.05 s
[Task 25/25] Current/Best: 3.02/ 9.96 GFLOPS | Progress: (10/10) | 62.84 s
+
[Task 1/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 1/25] Current/Best: 18.20/ 18.20 GFLOPS | Progress: (4/10) | 5.39 s
[Task 1/25] Current/Best: 13.14/ 18.20 GFLOPS | Progress: (8/10) | 9.13 s
[Task 1/25] Current/Best: 15.24/ 18.20 GFLOPS | Progress: (10/10) | 10.05 s Done.
+
[Task 2/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 2/25] Current/Best: 17.52/ 18.96 GFLOPS | Progress: (4/10) | 2.56 s
[Task 2/25] Current/Best: 5.94/ 18.96 GFLOPS | Progress: (8/10) | 3.93 s
[Task 2/25] Current/Best: 17.63/ 18.96 GFLOPS | Progress: (10/10) | 4.76 s Done.
+
[Task 3/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 3/25] Current/Best: 9.24/ 24.08 GFLOPS | Progress: (4/10) | 2.61 s
[Task 3/25] Current/Best: 19.00/ 24.08 GFLOPS | Progress: (8/10) | 4.43 s
[Task 3/25] Current/Best: 13.73/ 24.08 GFLOPS | Progress: (10/10) | 5.45 s Done.
+
[Task 4/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 4/25] Current/Best: 9.92/ 13.23 GFLOPS | Progress: (4/10) | 5.52 s
[Task 4/25] Current/Best: 11.44/ 13.23 GFLOPS | Progress: (8/10) | 7.65 s
[Task 4/25] Current/Best: 10.09/ 22.03 GFLOPS | Progress: (10/10) | 12.52 s Done.
+
[Task 5/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 5/25] Current/Best: 12.58/ 23.46 GFLOPS | Progress: (4/10) | 2.20 s
[Task 5/25] Current/Best: 12.11/ 23.46 GFLOPS | Progress: (8/10) | 3.62 s
[Task 5/25] Current/Best: 13.08/ 23.46 GFLOPS | Progress: (10/10) | 4.39 s Done.
+
[Task 6/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 6/25] Current/Best: 3.55/ 19.79 GFLOPS | Progress: (4/10) | 3.45 s
[Task 6/25] Current/Best: 5.53/ 19.79 GFLOPS | Progress: (8/10) | 5.78 s
[Task 6/25] Current/Best: 14.42/ 19.79 GFLOPS | Progress: (10/10) | 7.87 s Done.
+
[Task 7/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 7/25] Current/Best: 7.44/ 17.31 GFLOPS | Progress: (4/10) | 3.69 s
[Task 7/25] Current/Best: 9.67/ 21.26 GFLOPS | Progress: (8/10) | 5.31 s
[Task 7/25] Current/Best: 23.44/ 23.44 GFLOPS | Progress: (10/10) | 6.07 s Done.
+
[Task 8/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 8/25] Current/Best: 15.70/ 15.70 GFLOPS | Progress: (4/10) | 4.25 s
[Task 8/25] Current/Best: 11.51/ 15.70 GFLOPS | Progress: (8/10) | 6.18 s
[Task 8/25] Current/Best: 14.24/ 15.70 GFLOPS | Progress: (10/10) | 7.18 s Done.
+
[Task 9/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 9/25] Current/Best: 6.33/ 20.24 GFLOPS | Progress: (4/10) | 9.11 s
[Task 9/25] Current/Best: 7.09/ 20.24 GFLOPS | Progress: (8/10) | 10.75 s
[Task 9/25] Current/Best: 14.67/ 20.24 GFLOPS | Progress: (10/10) | 11.77 s Done.
+
[Task 10/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 10/25] Current/Best: 13.78/ 15.24 GFLOPS | Progress: (4/10) | 2.36 s
[Task 10/25] Current/Best: 9.31/ 16.68 GFLOPS | Progress: (8/10) | 3.93 s
[Task 10/25] Current/Best: 14.04/ 16.68 GFLOPS | Progress: (10/10) | 4.71 s Done.
+
[Task 11/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 11/25] Current/Best: 21.22/ 21.22 GFLOPS | Progress: (4/10) | 3.08 s
[Task 11/25] Current/Best: 5.76/ 21.22 GFLOPS | Progress: (8/10) | 5.71 s
[Task 11/25] Current/Best: 16.20/ 21.22 GFLOPS | Progress: (10/10) | 6.85 s Done.
+
[Task 12/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 12/25] Current/Best: 12.83/ 19.73 GFLOPS | Progress: (4/10) | 2.60 s
[Task 12/25] Current/Best: 16.05/ 20.96 GFLOPS | Progress: (8/10) | 5.53 s
[Task 12/25] Current/Best: 10.19/ 20.96 GFLOPS | Progress: (10/10) | 6.67 s Done.
+
[Task 13/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 13/25] Current/Best: 18.86/ 18.86 GFLOPS | Progress: (4/10) | 2.80 s
[Task 13/25] Current/Best: 16.96/ 23.16 GFLOPS | Progress: (8/10) | 5.58 s
[Task 13/25] Current/Best: 1.56/ 23.16 GFLOPS | Progress: (10/10) | 8.23 s Done.
+
[Task 14/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 14/25] Current/Best: 12.39/ 16.48 GFLOPS | Progress: (4/10) | 3.65 s
[Task 14/25] Current/Best: 23.13/ 23.13 GFLOPS | Progress: (8/10) | 6.80 s
[Task 14/25] Current/Best: 13.59/ 23.13 GFLOPS | Progress: (10/10) | 7.78 s
[Task 15/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 15/25] Current/Best: 23.91/ 23.91 GFLOPS | Progress: (4/10) | 6.71 s
[Task 15/25] Current/Best: 19.01/ 23.91 GFLOPS | Progress: (8/10) | 8.74 s
[Task 15/25] Current/Best: 6.08/ 23.91 GFLOPS | Progress: (10/10) | 9.76 s Done.
+
[Task 16/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 16/25] Current/Best: 14.74/ 21.92 GFLOPS | Progress: (4/10) | 2.56 s
[Task 16/25] Current/Best: 6.00/ 21.92 GFLOPS | Progress: (8/10) | 4.04 s
[Task 16/25] Current/Best: 13.51/ 21.92 GFLOPS | Progress: (10/10) | 6.69 s Done.
+
[Task 17/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 17/25] Current/Best: 15.77/ 23.21 GFLOPS | Progress: (4/10) | 3.08 s
[Task 17/25] Current/Best: 23.92/ 23.92 GFLOPS | Progress: (8/10) | 4.94 s
[Task 17/25] Current/Best: 16.04/ 23.92 GFLOPS | Progress: (10/10) | 5.95 s Done.
+
[Task 18/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 18/25] Current/Best: 2.91/ 17.35 GFLOPS | Progress: (4/10) | 4.54 s
[Task 18/25] Current/Best: 6.14/ 17.35 GFLOPS | Progress: (8/10) | 7.29 s
[Task 18/25] Current/Best: 20.79/ 20.79 GFLOPS | Progress: (10/10) | 8.54 s Done.
+
[Task 19/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 19/25] Current/Best: 10.73/ 22.03 GFLOPS | Progress: (4/10) | 3.84 s
[Task 19/25] Current/Best: 8.06/ 22.81 GFLOPS | Progress: (8/10) | 6.28 s
[Task 19/25] Current/Best: 19.35/ 22.81 GFLOPS | Progress: (10/10) | 9.96 s Done.
+
[Task 20/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 20/25] Current/Best: 11.99/ 18.06 GFLOPS | Progress: (4/10) | 3.72 s
[Task 20/25] Current/Best: 15.20/ 18.06 GFLOPS | Progress: (8/10) | 5.59 s Done.
+
[Task 20/25] Current/Best: 18.81/ 18.81 GFLOPS | Progress: (10/10) | 6.48 s Done.
+
[Task 21/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 21/25] Current/Best: 6.30/ 13.65 GFLOPS | Progress: (4/10) | 2.90 s
[Task 21/25] Current/Best: 9.87/ 13.65 GFLOPS | Progress: (8/10) | 5.26 s
[Task 21/25] Current/Best: 14.21/ 14.21 GFLOPS | Progress: (10/10) | 5.86 s
[Task 22/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 22/25] Current/Best: 8.81/ 12.39 GFLOPS | Progress: (4/10) | 3.37 s
[Task 22/25] Current/Best: 19.36/ 21.35 GFLOPS | Progress: (8/10) | 5.45 s
[Task 22/25] Current/Best: 4.55/ 21.35 GFLOPS | Progress: (10/10) | 6.40 s Done.
+
[Task 23/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 23/25] Current/Best: 13.10/ 19.96 GFLOPS | Progress: (4/10) | 3.27 s
[Task 23/25] Current/Best: 10.81/ 23.59 GFLOPS | Progress: (8/10) | 5.69 s
[Task 23/25] Current/Best: 13.11/ 23.59 GFLOPS | Progress: (10/10) | 6.70 s Done.
+
[Task 24/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s
[Task 24/25] Current/Best: 2.24/ 3.10 GFLOPS | Progress: (4/10) | 350.03 s
[Task 24/25] Current/Best: 4.02/ 9.04 GFLOPS | Progress: (8/10) | 353.88 s
[Task 24/25] Current/Best: 6.81/ 9.04 GFLOPS | Progress: (10/10) | 354.81 s
[Task 25/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/10) | 0.00 s Done.
+ Done.
+
[Task 25/25] Current/Best: 2.87/ 8.30 GFLOPS | Progress: (4/10) | 16.81 s
[Task 25/25] Current/Best: 3.04/ 9.81 GFLOPS | Progress: (8/10) | 20.16 s
[Task 25/25] Current/Best: 10.36/ 10.36 GFLOPS | Progress: (10/10) | 24.91 s
The output from this tuning process will look something like this:
@@ -594,8 +594,8 @@ Verify that the optimized model runs and produces the same results:
.. code-block:: none
- class='n02123045 tabby, tabby cat' with probability=0.621103
- class='n02123159 tiger cat' with probability=0.356379
+ class='n02123045 tabby, tabby cat' with probability=0.621104
+ class='n02123159 tiger cat' with probability=0.356378
class='n02124075 Egyptian cat' with probability=0.019712
class='n02129604 tiger, Panthera tigris' with probability=0.001215
class='n04040759 radiator' with probability=0.000262
@@ -648,8 +648,8 @@ improvement in comparing the optimized model to the unoptimized model.
.. code-block:: none
- optimized: {'mean': 431.59794143003637, 'median': 429.7326266500022, 'std': 3.369179216079745}
- unoptimized: {'mean': 492.0994797199995, 'median': 492.4136981499999, 'std': 0.8176268748468473}
+ optimized: {'mean': 420.703913010002, 'median': 420.26930499996524, 'std': 1.2707582899836265}
+ unoptimized: {'mean': 492.5393293900015, 'median': 493.21396980003556, 'std': 1.366467218342447}
@@ -669,7 +669,7 @@ profiling/benchmarking.
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 127 minutes 49.205 seconds)
+ **Total running time of the script:** ( 12 minutes 22.078 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 043b9f9..4236614 100644
--- a/docs/_sources/tutorial/cross_compilation_and_rpc.rst.txt
+++ b/docs/_sources/tutorial/cross_compilation_and_rpc.rst.txt
@@ -235,7 +235,7 @@ device and returns the measured cost. Network overhead is excluded.
.. code-block:: none
- 1.328e-07 secs/op
+ 1.256e-07 secs/op
diff --git a/docs/_sources/tutorial/intro_topi.rst.txt b/docs/_sources/tutorial/intro_topi.rst.txt
index 70237c3..51c0277 100644
--- a/docs/_sources/tutorial/intro_topi.rst.txt
+++ b/docs/_sources/tutorial/intro_topi.rst.txt
@@ -230,7 +230,7 @@ As you can see, scheduled stages of computation have been accumulated and we can
.. code-block:: none
- [stage(a, placeholder(a, 0x224e2e70)), stage(b, placeholder(b, 0x20f8f850)), 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(mi [...]
+ [stage(a, placeholder(a, 0x76137d0)), stage(b, placeholder(b, 0x22155c60)), 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 3ef8d49..bb1b31c 100644
--- a/docs/_sources/tutorial/sg_execution_times.rst.txt
+++ b/docs/_sources/tutorial/sg_execution_times.rst.txt
@@ -5,17 +5,17 @@
Computation times
=================
-**130:22.300** total execution time for **tutorial** files:
+**15:38.539** total execution time for **tutorial** files:
-- **127:49.205**: :ref:`sphx_glr_tutorial_autotvm_relay_x86.py` (``autotvm_relay_x86.py``)
-- **00:57.918**: :ref:`sphx_glr_tutorial_tensor_expr_get_started.py` (``tensor_expr_get_started.py``)
-- **00:50.172**: :ref:`sphx_glr_tutorial_auto_scheduler_matmul_x86.py` (``auto_scheduler_matmul_x86.py``)
-- **00:25.193**: :ref:`sphx_glr_tutorial_relay_quick_start.py` (``relay_quick_start.py``)
-- **00:17.487**: :ref:`sphx_glr_tutorial_autotvm_matmul_x86.py` (``autotvm_matmul_x86.py``)
-- **00:01.283**: :ref:`sphx_glr_tutorial_tensor_ir_blitz_course.py` (``tensor_ir_blitz_course.py``)
-- **00:00.674**: :ref:`sphx_glr_tutorial_intro_topi.py` (``intro_topi.py``)
-- **00:00.218**: :ref:`sphx_glr_tutorial_cross_compilation_and_rpc.py` (``cross_compilation_and_rpc.py``)
-- **00:00.040**: :ref:`sphx_glr_tutorial_tvmc_python.py` (``tvmc_python.py``)
+- **12:22.078**: :ref:`sphx_glr_tutorial_autotvm_relay_x86.py` (``autotvm_relay_x86.py``)
+- **01:17.751**: :ref:`sphx_glr_tutorial_auto_scheduler_matmul_x86.py` (``auto_scheduler_matmul_x86.py``)
+- **00:59.421**: :ref:`sphx_glr_tutorial_tensor_expr_get_started.py` (``tensor_expr_get_started.py``)
+- **00:31.974**: :ref:`sphx_glr_tutorial_autotvm_matmul_x86.py` (``autotvm_matmul_x86.py``)
+- **00:25.730**: :ref:`sphx_glr_tutorial_relay_quick_start.py` (``relay_quick_start.py``)
+- **00:00.698**: :ref:`sphx_glr_tutorial_intro_topi.py` (``intro_topi.py``)
+- **00:00.553**: :ref:`sphx_glr_tutorial_tensor_ir_blitz_course.py` (``tensor_ir_blitz_course.py``)
+- **00:00.186**: :ref:`sphx_glr_tutorial_cross_compilation_and_rpc.py` (``cross_compilation_and_rpc.py``)
+- **00:00.039**: :ref:`sphx_glr_tutorial_introduction.py` (``introduction.py``)
- **00:00.039**: :ref:`sphx_glr_tutorial_tvmc_command_line_driver.py` (``tvmc_command_line_driver.py``)
- **00:00.036**: :ref:`sphx_glr_tutorial_install.py` (``install.py``)
-- **00:00.035**: :ref:`sphx_glr_tutorial_introduction.py` (``introduction.py``)
+- **00:00.034**: :ref:`sphx_glr_tutorial_tvmc_python.py` (``tvmc_python.py``)
diff --git a/docs/_sources/tutorial/tensor_expr_get_started.rst.txt b/docs/_sources/tutorial/tensor_expr_get_started.rst.txt
index 80e5436..e18477a 100644
--- a/docs/_sources/tutorial/tensor_expr_get_started.rst.txt
+++ b/docs/_sources/tutorial/tensor_expr_get_started.rst.txt
@@ -387,7 +387,7 @@ factor to be the number of threads on your CPU.
.. code-block:: none
- vector: 0.000024
+ vector: 0.000025
@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, [(stride: int32*n: int32)], [], type="auto"),
@@ -436,10 +436,10 @@ We can now compare the different schedules
.. code-block:: none
Operator Timing Performance
- numpy 8.148249999067047e-06 1.0
- naive 5.698e-06 0.6992912589393313
- parallel 5.804900000000001e-06 0.7124106404031106
- vector 2.37747e-05 2.9177676191479325
+ numpy 7.829910000509699e-06 1.0
+ naive 5.8454e-06 0.7465475337033869
+ parallel 6.0315e-06 0.7703153675594446
+ vector 2.45599e-05 3.1366771774389792
@@ -828,7 +828,7 @@ matrix multiplication.
.. code-block:: none
- Numpy running time: 0.017625
+ Numpy running time: 0.018044
@@ -884,7 +884,7 @@ optimizations.
.. code-block:: none
- none: 3.212491
+ none: 3.300723
@@ -982,7 +982,7 @@ schedule.
.. code-block:: none
- blocking: 0.293904
+ blocking: 0.306370
@@ -1073,7 +1073,7 @@ already cache friendly from our previous optimizations.
.. code-block:: none
- vectorization: 0.330108
+ vectorization: 0.335301
@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], []),
@@ -1144,7 +1144,7 @@ more cache friendly.
.. code-block:: none
- loop permutation: 0.113881
+ loop permutation: 0.115655
@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], []),
@@ -1240,7 +1240,7 @@ optimized schedule.
.. code-block:: none
- array packing: 0.108160
+ array packing: 0.108664
@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], []),
@@ -1330,7 +1330,7 @@ to `C` when all the block results are ready.
.. code-block:: none
- block caching: 0.108630
+ block caching: 0.110934
@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], []),
@@ -1413,7 +1413,7 @@ of thread-level parallelization.
.. code-block:: none
- parallelization: 0.141288
+ parallelization: 0.144610
@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], []),
@@ -1491,13 +1491,13 @@ working, we can compare the results.
.. code-block:: none
Operator Timing Performance
- none 3.2124908298 1.0
- blocking 0.2939038752 0.09148784876634108
- vectorization 0.33010801170000004 0.10275765105313983
- loop permutation 0.1138805889 0.03544931174389994
- array packing 0.1081595603 0.0336684417264885
- block caching 0.10863010529999999 0.03381491529635369
- parallelization 0.14128778339999998 0.04398075850968145
+ none 3.3007231828999997 1.0
+ blocking 0.3063697227 0.09281896897237685
+ vectorization 0.3353008133 0.1015840452895557
+ loop permutation 0.1156545108 0.03503914275488758
+ array packing 0.10866350259999999 0.03292111957856725
+ block caching 0.110934426 0.033609127410234245
+ parallelization 0.1446097572 0.04381153740767396
diff --git a/docs/commit_hash b/docs/commit_hash
index 1483a02..da56b48 100644
--- a/docs/commit_hash
+++ b/docs/commit_hash
@@ -1 +1 @@
-72c761c807beae48948bd78bb7e6478dae829e0b
+42985c372681bc1915a6be1be917aa45f8f94a55
diff --git a/docs/how_to/compile_models/from_mxnet.html b/docs/how_to/compile_models/from_mxnet.html
index 02af063..424e3e6 100644
--- a/docs/how_to/compile_models/from_mxnet.html
+++ b/docs/how_to/compile_models/from_mxnet.html
@@ -400,7 +400,7 @@
</div>
<img alt="../../_images/sphx_glr_from_mxnet_001.png" class="sphx-glr-single-img" src="../../_images/sphx_glr_from_mxnet_001.png" />
<p class="sphx-glr-script-out">Out:</p>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Downloading /workspace/.mxnet/models/resnet18_v1-a0666292.zipcad3e097-2ad2-47c3-8ea4-462b69f1426b from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/resnet18_v1-a0666292.zip...
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Downloading /workspace/.mxnet/models/resnet18_v1-a0666292.zip8c4ab368-acda-4b07-b69a-0498f0305aba 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_paddle.html b/docs/how_to/compile_models/from_paddle.html
index 37ab4d6..07dadbe 100644
--- a/docs/how_to/compile_models/from_paddle.html
+++ b/docs/how_to/compile_models/from_paddle.html
@@ -448,7 +448,7 @@ A quick solution is</p>
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>TVM prediction top-1 id: 282, class name: 282: 'tiger cat',
</pre></div>
</div>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 5.435 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 5.921 seconds)</p>
<div class="sphx-glr-footer class sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-compile-models-from-paddle-py">
<div class="sphx-glr-download docutils container">
<p><a class="reference download internal" download="" href="../../_downloads/16269b77359771348d507395692524cf/from_paddle.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_paddle.py</span></code></a></p>
diff --git a/docs/how_to/compile_models/from_pytorch.html b/docs/how_to/compile_models/from_pytorch.html
index cc48056..1ee0737 100644
--- a/docs/how_to/compile_models/from_pytorch.html
+++ b/docs/how_to/compile_models/from_pytorch.html
@@ -386,31 +386,10 @@ 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]
- 6%|6 | 2.81M/44.7M [00:00<00:01, 29.4MB/s]
- 13%|#2 | 5.62M/44.7M [00:00<00:03, 11.5MB/s]
- 16%|#6 | 7.23M/44.7M [00:00<00:03, 12.5MB/s]
- 20%|#9 | 8.81M/44.7M [00:00<00:02, 13.2MB/s]
- 26%|##5 | 11.5M/44.7M [00:00<00:02, 17.1MB/s]
- 30%|### | 13.4M/44.7M [00:01<00:02, 13.5MB/s]
- 34%|###3 | 15.0M/44.7M [00:01<00:02, 13.8MB/s]
- 37%|###6 | 16.5M/44.7M [00:01<00:02, 12.8MB/s]
- 40%|###9 | 17.9M/44.7M [00:01<00:02, 11.7MB/s]
- 43%|####2 | 19.1M/44.7M [00:01<00:02, 11.3MB/s]
- 45%|####5 | 20.2M/44.7M [00:01<00:02, 9.06MB/s]
- 48%|####7 | 21.3M/44.7M [00:01<00:02, 9.26MB/s]
- 50%|####9 | 22.3M/44.7M [00:01<00:02, 9.45MB/s]
- 54%|#####4 | 24.3M/44.7M [00:02<00:01, 12.0MB/s]
- 59%|#####8 | 26.2M/44.7M [00:02<00:01, 13.9MB/s]
- 63%|######2 | 28.1M/44.7M [00:02<00:01, 15.5MB/s]
- 66%|######6 | 29.7M/44.7M [00:02<00:01, 14.9MB/s]
- 70%|######9 | 31.1M/44.7M [00:02<00:01, 9.33MB/s]
- 73%|#######3 | 32.7M/44.7M [00:02<00:01, 10.5MB/s]
- 78%|#######7 | 34.7M/44.7M [00:03<00:00, 11.1MB/s]
- 82%|########2 | 36.7M/44.7M [00:03<00:00, 13.1MB/s]
- 85%|########5 | 38.2M/44.7M [00:03<00:00, 13.4MB/s]
- 90%|######### | 40.3M/44.7M [00:03<00:00, 15.5MB/s]
- 95%|#########4| 42.4M/44.7M [00:03<00:00, 17.2MB/s]
-100%|##########| 44.7M/44.7M [00:03<00:00, 13.4MB/s]
+ 21%|##1 | 9.52M/44.7M [00:00<00:00, 99.8MB/s]
+ 57%|#####7 | 25.5M/44.7M [00:00<00:00, 140MB/s]
+ 93%|#########3| 41.8M/44.7M [00:00<00:00, 153MB/s]
+100%|##########| 44.7M/44.7M [00:00<00:00, 147MB/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 db00bed..ca73121 100644
--- a/docs/how_to/compile_models/from_tensorflow.html
+++ b/docs/how_to/compile_models/from_tensorflow.html
@@ -610,7 +610,7 @@ 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 11.242 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 8.286 seconds)</p>
<div class="sphx-glr-footer class sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-compile-models-from-tensorflow-py">
<div class="sphx-glr-download 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 caec00e..5ce55fa 100644
--- a/docs/how_to/compile_models/sg_execution_times.html
+++ b/docs/how_to/compile_models/sg_execution_times.html
@@ -300,17 +300,17 @@
<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>04:55.760</strong> total execution time for <strong>how_to_compile_models</strong> files:</p>
+<p><strong>04:51.558</strong> total execution time for <strong>how_to_compile_models</strong> files:</p>
<ul class="simple">
-<li><p><strong>01:11.242</strong>: <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></li>
-<li><p><strong>01:05.435</strong>: <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></li>
-<li><p><strong>00:54.383</strong>: <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></li>
-<li><p><strong>00:24.480</strong>: <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></li>
-<li><p><strong>00:23.067</strong>: <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></li>
-<li><p><strong>00:20.618</strong>: <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></li>
-<li><p><strong>00:20.223</strong>: <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></li>
-<li><p><strong>00:13.886</strong>: <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></li>
-<li><p><strong>00:02.426</strong>: <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></li>
+<li><p><strong>01:08.286</strong>: <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></li>
+<li><p><strong>01:05.921</strong>: <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></li>
+<li><p><strong>00:55.651</strong>: <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></li>
+<li><p><strong>00:25.350</strong>: <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></li>
+<li><p><strong>00:20.574</strong>: <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></li>
+<li><p><strong>00:20.378</strong>: <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></li>
+<li><p><strong>00:18.481</strong>: <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></li>
+<li><p><strong>00:13.972</strong>: <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></li>
+<li><p><strong>00:02.945</strong>: <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></li>
</ul>
</div>
diff --git a/docs/how_to/deploy_models/deploy_model_on_android.html b/docs/how_to/deploy_models/deploy_model_on_android.html
index 2cf7105..9b7bc16 100644
--- a/docs/how_to/deploy_models/deploy_model_on_android.html
+++ b/docs/how_to/deploy_models/deploy_model_on_android.html
@@ -622,7 +622,7 @@ to the remote android device.</p>
Evaluate inference time cost...
Execution time summary:
mean (ms) median (ms) max (ms) min (ms) std (ms)
- 15.4084 15.4311 15.5087 15.2838 0.0684
+ 16.0157 15.9729 16.5046 15.8410 0.1828
</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 9834d04..208ca31 100644
--- a/docs/how_to/deploy_models/deploy_object_detection_pytorch.html
+++ b/docs/how_to/deploy_models/deploy_object_detection_pytorch.html
@@ -409,96 +409,64 @@ 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]
- 0%| | 824k/170M [00:00<00:21, 8.14MB/s]
- 1%|1 | 2.38M/170M [00:00<00:13, 12.9MB/s]
- 3%|2 | 5.06M/170M [00:00<00:09, 18.5MB/s]
- 4%|4 | 7.24M/170M [00:00<00:09, 18.1MB/s]
- 5%|5 | 8.96M/170M [00:00<00:11, 14.8MB/s]
- 6%|6 | 10.4M/170M [00:00<00:14, 11.8MB/s]
- 7%|6 | 11.9M/170M [00:00<00:13, 12.4MB/s]
- 8%|7 | 13.6M/170M [00:01<00:12, 13.6MB/s]
- 9%|8 | 15.1M/170M [00:01<00:11, 14.1MB/s]
- 10%|9 | 16.6M/170M [00:01<00:11, 14.2MB/s]
- 11%|# | 18.0M/170M [00:01<00:11, 13.7MB/s]
- 12%|#2 | 20.5M/170M [00:01<00:09, 16.8MB/s]
- 13%|#3 | 22.5M/170M [00:01<00:08, 18.0MB/s]
- 14%|#4 | 24.2M/170M [00:01<00:08, 17.0MB/s]
- 15%|#5 | 25.9M/170M [00:01<00:09, 15.1MB/s]
- 16%|#6 | 27.4M/170M [00:01<00:11, 13.6MB/s]
- 17%|#7 | 29.1M/170M [00:02<00:11, 12.9MB/s]
- 18%|#8 | 31.0M/170M [00:02<00:10, 14.5MB/s]
- 20%|#9 | 33.5M/170M [00:02<00:08, 17.4MB/s]
- 21%|## | 35.3M/170M [00:02<00:08, 17.4MB/s]
- 22%|##1 | 37.0M/170M [00:02<00:09, 15.3MB/s]
- 23%|##2 | 38.5M/170M [00:02<00:09, 15.1MB/s]
- 24%|##3 | 40.7M/170M [00:02<00:08, 16.9MB/s]
- 25%|##4 | 42.4M/170M [00:02<00:08, 15.6MB/s]
- 26%|##6 | 44.2M/170M [00:03<00:07, 16.6MB/s]
- 27%|##7 | 45.9M/170M [00:03<00:08, 15.8MB/s]
- 28%|##7 | 47.4M/170M [00:03<00:08, 15.8MB/s]
- 29%|##8 | 48.9M/170M [00:03<00:11, 11.0MB/s]
- 30%|##9 | 50.4M/170M [00:03<00:10, 11.8MB/s]
- 31%|###1 | 53.3M/170M [00:03<00:07, 15.6MB/s]
- 33%|###2 | 55.4M/170M [00:03<00:06, 17.4MB/s]
- 34%|###4 | 57.9M/170M [00:03<00:06, 19.2MB/s]
- 35%|###5 | 59.9M/170M [00:04<00:07, 16.5MB/s]
- 36%|###6 | 61.6M/170M [00:04<00:08, 13.7MB/s]
- 37%|###7 | 63.3M/170M [00:04<00:07, 14.6MB/s]
- 39%|###8 | 65.5M/170M [00:04<00:08, 13.2MB/s]
- 40%|###9 | 67.7M/170M [00:04<00:07, 14.9MB/s]
- 42%|####1 | 70.5M/170M [00:04<00:05, 17.5MB/s]
- 43%|####2 | 72.9M/170M [00:04<00:05, 19.0MB/s]
- 44%|####4 | 75.1M/170M [00:05<00:04, 20.1MB/s]
- 45%|####5 | 77.1M/170M [00:05<00:04, 20.1MB/s]
- 47%|####6 | 79.2M/170M [00:05<00:04, 20.1MB/s]
- 48%|####7 | 81.1M/170M [00:05<00:05, 17.6MB/s]
- 49%|####8 | 82.9M/170M [00:05<00:05, 16.8MB/s]
- 50%|####9 | 84.6M/170M [00:05<00:06, 13.5MB/s]
- 51%|##### | 86.0M/170M [00:05<00:06, 12.8MB/s]
- 51%|#####1 | 87.3M/170M [00:06<00:07, 12.3MB/s]
- 52%|#####2 | 88.7M/170M [00:06<00:06, 12.7MB/s]
- 53%|#####3 | 90.7M/170M [00:06<00:05, 14.2MB/s]
- 55%|#####4 | 93.2M/170M [00:06<00:04, 17.3MB/s]
- 56%|#####6 | 95.9M/170M [00:06<00:03, 20.2MB/s]
- 58%|#####7 | 97.9M/170M [00:06<00:04, 18.0MB/s]
- 59%|#####8 | 99.7M/170M [00:06<00:04, 15.5MB/s]
- 60%|#####9 | 101M/170M [00:06<00:04, 15.2MB/s]
- 61%|###### | 103M/170M [00:06<00:04, 16.1MB/s]
- 62%|######1 | 105M/170M [00:07<00:04, 16.5MB/s]
- 63%|######3 | 107M/170M [00:07<00:03, 17.4MB/s]
- 64%|######4 | 109M/170M [00:07<00:03, 18.8MB/s]
- 66%|######6 | 113M/170M [00:07<00:02, 23.2MB/s]
- 68%|######7 | 115M/170M [00:07<00:02, 22.7MB/s]
- 69%|######9 | 118M/170M [00:07<00:02, 24.9MB/s]
- 71%|####### | 120M/170M [00:07<00:02, 21.8MB/s]
- 72%|#######2 | 122M/170M [00:07<00:02, 21.6MB/s]
- 73%|#######3 | 125M/170M [00:07<00:02, 21.1MB/s]
- 75%|#######4 | 127M/170M [00:08<00:02, 19.7MB/s]
- 76%|#######5 | 129M/170M [00:08<00:02, 20.4MB/s]
- 77%|#######7 | 131M/170M [00:08<00:02, 16.9MB/s]
- 78%|#######8 | 133M/170M [00:08<00:02, 14.9MB/s]
- 79%|#######9 | 134M/170M [00:08<00:02, 14.6MB/s]
- 80%|#######9 | 136M/170M [00:08<00:02, 13.5MB/s]
- 81%|######## | 137M/170M [00:09<00:03, 10.9MB/s]
- 81%|########1 | 138M/170M [00:09<00:03, 10.9MB/s]
- 82%|########2 | 139M/170M [00:09<00:03, 9.60MB/s]
- 83%|########2 | 141M/170M [00:09<00:02, 10.5MB/s]
- 84%|########3 | 142M/170M [00:09<00:02, 11.6MB/s]
- 85%|########4 | 144M/170M [00:09<00:01, 13.5MB/s]
- 86%|########5 | 146M/170M [00:09<00:01, 14.4MB/s]
- 87%|########6 | 147M/170M [00:09<00:01, 13.6MB/s]
- 88%|########7 | 149M/170M [00:09<00:01, 14.1MB/s]
- 89%|########8 | 151M/170M [00:10<00:01, 15.7MB/s]
- 90%|########9 | 152M/170M [00:10<00:01, 13.6MB/s]
- 91%|#########1| 155M/170M [00:10<00:00, 17.3MB/s]
- 93%|#########2| 157M/170M [00:10<00:00, 18.9MB/s]
- 94%|#########3| 159M/170M [00:10<00:00, 19.6MB/s]
- 95%|#########5| 161M/170M [00:10<00:00, 19.4MB/s]
- 96%|#########6| 163M/170M [00:10<00:00, 18.9MB/s]
- 97%|#########7| 165M/170M [00:10<00:00, 17.8MB/s]
- 98%|#########8| 167M/170M [00:11<00:00, 16.8MB/s]
- 99%|#########9| 168M/170M [00:11<00:00, 13.8MB/s]
-100%|##########| 170M/170M [00:11<00:00, 15.7MB/s]
+ 3%|2 | 4.56M/170M [00:00<00:03, 46.9MB/s]
+ 5%|5 | 9.04M/170M [00:00<00:04, 33.7MB/s]
+ 7%|7 | 12.5M/170M [00:00<00:06, 24.7MB/s]
+ 10%|# | 17.4M/170M [00:00<00:04, 32.1MB/s]
+ 13%|#2 | 22.0M/170M [00:00<00:04, 35.7MB/s]
+ 15%|#5 | 25.8M/170M [00:00<00:04, 35.4MB/s]
+ 18%|#7 | 30.0M/170M [00:00<00:03, 37.8MB/s]
+ 20%|#9 | 33.8M/170M [00:01<00:03, 36.8MB/s]
+ 22%|##2 | 37.4M/170M [00:01<00:04, 30.3MB/s]
+ 24%|##3 | 40.5M/170M [00:01<00:04, 29.0MB/s]
+ 26%|##5 | 43.5M/170M [00:01<00:04, 28.9MB/s]
+ 27%|##7 | 46.3M/170M [00:01<00:05, 23.4MB/s]
+ 30%|### | 51.0M/170M [00:01<00:04, 29.2MB/s]
+ 33%|###2 | 55.5M/170M [00:01<00:03, 33.4MB/s]
+ 35%|###4 | 58.9M/170M [00:02<00:04, 27.7MB/s]
+ 37%|###6 | 62.3M/170M [00:02<00:03, 29.4MB/s]
+ 39%|###8 | 65.4M/170M [00:02<00:03, 29.9MB/s]
+ 40%|#### | 68.5M/170M [00:02<00:03, 30.2MB/s]
+ 42%|####2 | 71.5M/170M [00:02<00:03, 26.9MB/s]
+ 44%|####3 | 74.2M/170M [00:02<00:04, 24.1MB/s]
+ 45%|####5 | 76.7M/170M [00:02<00:04, 19.5MB/s]
+ 47%|####7 | 80.6M/170M [00:02<00:03, 24.1MB/s]
+ 49%|####8 | 83.2M/170M [00:03<00:03, 22.9MB/s]
+ 51%|##### | 86.0M/170M [00:03<00:03, 24.0MB/s]
+ 53%|#####2 | 89.8M/170M [00:03<00:03, 27.9MB/s]
+ 55%|#####4 | 92.6M/170M [00:03<00:02, 28.3MB/s]
+ 56%|#####6 | 95.5M/170M [00:03<00:03, 22.1MB/s]
+ 58%|#####7 | 97.9M/170M [00:03<00:03, 19.9MB/s]
+ 59%|#####9 | 100M/170M [00:03<00:03, 21.2MB/s]
+ 61%|###### | 103M/170M [00:03<00:03, 22.5MB/s]
+ 62%|######2 | 106M/170M [00:04<00:02, 24.0MB/s]
+ 64%|######3 | 108M/170M [00:04<00:02, 23.9MB/s]
+ 66%|######6 | 112M/170M [00:04<00:02, 28.3MB/s]
+ 68%|######7 | 115M/170M [00:04<00:02, 26.3MB/s]
+ 69%|######9 | 118M/170M [00:04<00:02, 24.3MB/s]
+ 71%|####### | 120M/170M [00:04<00:02, 24.4MB/s]
+ 72%|#######2 | 123M/170M [00:05<00:03, 13.9MB/s]
+ 73%|#######3 | 124M/170M [00:05<00:03, 14.3MB/s]
+ 75%|#######4 | 127M/170M [00:05<00:02, 15.6MB/s]
+ 76%|#######6 | 130M/170M [00:05<00:02, 16.6MB/s]
+ 78%|#######7 | 132M/170M [00:05<00:02, 15.6MB/s]
+ 79%|#######9 | 135M/170M [00:05<00:01, 18.6MB/s]
+ 80%|######## | 137M/170M [00:05<00:02, 16.4MB/s]
+ 81%|########1 | 138M/170M [00:05<00:01, 17.0MB/s]
+ 83%|########2 | 140M/170M [00:06<00:02, 15.6MB/s]
+ 84%|########3 | 142M/170M [00:06<00:01, 17.2MB/s]
+ 85%|########5 | 145M/170M [00:06<00:01, 18.8MB/s]
+ 86%|########6 | 146M/170M [00:06<00:01, 14.3MB/s]
+ 88%|########7 | 149M/170M [00:06<00:01, 17.5MB/s]
+ 89%|########9 | 152M/170M [00:06<00:00, 19.4MB/s]
+ 91%|######### | 154M/170M [00:06<00:00, 19.6MB/s]
+ 92%|#########1| 156M/170M [00:06<00:00, 20.0MB/s]
+ 94%|#########3| 160M/170M [00:07<00:00, 24.7MB/s]
+ 95%|#########5| 162M/170M [00:07<00:00, 21.2MB/s]
+ 97%|#########6| 164M/170M [00:07<00:00, 20.6MB/s]
+ 98%|#########8| 167M/170M [00:07<00:00, 23.2MB/s]
+100%|#########9| 170M/170M [00:07<00:00, 21.3MB/s]
+100%|##########| 170M/170M [00:07<00:00, 23.4MB/s]
/usr/local/lib/python3.7/dist-packages/torch/functional.py:445: UserWarning: torch.meshgrid: in an upcoming release, it will be required to pass the indexing argument. (Triggered internally at ../aten/src/ATen/native/TensorShape.cpp:2157.)
return _VF.meshgrid(tensors, **kwargs) # type: ignore[attr-defined]
/usr/local/lib/python3.7/dist-packages/torch/nn/functional.py:3702: 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).
@@ -519,13 +487,6 @@ be unstable.</p>
for s, s_orig in zip(new_size, original_size)
/usr/local/lib/python3.7/dist-packages/torchvision/models/detection/roi_heads.py:372: 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).
return torch.tensor(M + 2 * padding).to(torch.float32) / torch.tensor(M).to(torch.float32)
-/usr/local/lib/python3.7/dist-packages/torch/jit/_trace.py:991: TracerWarning: Output nr 2. of the traced function does not match the corresponding output of the Python function. Detailed error:
-Tensor-likes are not close!
-
-Mismatched elements: 1 / 2 (50.0%)
-Greatest absolute difference: 0.9288840740919113 at index 1 (up to 1e-05 allowed)
-Greatest relative difference: 13.16154513835738 at index 1 (up to 1e-05 allowed)
- _module_class,
</pre></div>
</div>
</div>
@@ -600,7 +561,7 @@ torchvision rcnn models.</p>
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Get 9 valid boxes
</pre></div>
</div>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 3 minutes 6.085 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 3 minutes 7.676 seconds)</p>
<div class="sphx-glr-footer class sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-deploy-models-deploy-object-detection-pytorch-py">
<div class="sphx-glr-download 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 1dc7b40..54227ff 100644
--- a/docs/how_to/deploy_models/deploy_prequantized.html
+++ b/docs/how_to/deploy_models/deploy_prequantized.html
@@ -450,7 +450,12 @@ 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]
-100%|##########| 13.6M/13.6M [00:00<00:00, 144MB/s]
+ 9%|8 | 1.16M/13.6M [00:00<00:01, 11.9MB/s]
+ 22%|##2 | 3.01M/13.6M [00:00<00:00, 16.2MB/s]
+ 45%|####5 | 6.13M/13.6M [00:00<00:00, 23.2MB/s]
+ 61%|######1 | 8.34M/13.6M [00:00<00:00, 19.7MB/s]
+ 90%|########9 | 12.1M/13.6M [00:00<00:00, 26.1MB/s]
+100%|##########| 13.6M/13.6M [00:00<00:00, 23.6MB/s]
</pre></div>
</div>
</div>
@@ -543,7 +548,7 @@ output values are identical out of 1000 outputs from mobilenet v2.</p>
<p class="sphx-glr-script-out">Out:</p>
<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)
- 89.3021 89.9408 90.7087 88.0098 0.8485
+ 90.1617 90.1035 93.2273 89.9587 0.3468
</pre></div>
</div>
<div class="admonition note">
@@ -582,7 +587,7 @@ 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 2.571 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 3.920 seconds)</p>
<div class="sphx-glr-footer class sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-deploy-models-deploy-prequantized-py">
<div class="sphx-glr-download 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 0d28610..1256a88 100644
--- a/docs/how_to/deploy_models/deploy_prequantized_tflite.html
+++ b/docs/how_to/deploy_models/deploy_prequantized_tflite.html
@@ -540,7 +540,7 @@ TFLite Top-5 labels: [387 102 386 341 349]
<p class="sphx-glr-script-out">Out:</p>
<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)
- 118.1848 118.1044 126.6462 116.9886 0.9616
+ 119.4513 119.4375 122.0499 118.4966 0.4518
</pre></div>
</div>
<div class="admonition note">
@@ -568,7 +568,7 @@ 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> ( 1 minutes 52.560 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 58.957 seconds)</p>
<div class="sphx-glr-footer class sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-deploy-models-deploy-prequantized-tflite-py">
<div class="sphx-glr-download 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 a9938be..18346ff 100644
--- a/docs/how_to/deploy_models/deploy_quantized.html
+++ b/docs/how_to/deploy_models/deploy_quantized.html
@@ -480,7 +480,7 @@ 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 15.134 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 11.281 seconds)</p>
<div class="sphx-glr-footer class sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-deploy-models-deploy-quantized-py">
<div class="sphx-glr-download 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 44790b4..4567983 100644
--- a/docs/how_to/deploy_models/deploy_ssd_gluoncv.html
+++ b/docs/how_to/deploy_models/deploy_ssd_gluoncv.html
@@ -415,24 +415,23 @@ to your device.</p>
Downloading /workspace/.mxnet/models/ssd_512_resnet50_v1_voc-9c8b225a.zip from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/ssd_512_resnet50_v1_voc-9c8b225a.zip...
0%| | 0/132723 [00:00<?, ?KB/s]
- 4%|3 | 4830/132723 [00:00<00:02, 48286.23KB/s]
- 9%|9 | 12337/132723 [00:00<00:01, 64027.72KB/s]
- 15%|#4 | 19833/132723 [00:00<00:01, 69014.41KB/s]
- 21%|## | 27444/132723 [00:00<00:01, 71813.69KB/s]
- 27%|##6 | 35206/132723 [00:00<00:01, 73904.50KB/s]
- 32%|###2 | 42597/132723 [00:00<00:01, 68105.15KB/s]
- 37%|###7 | 49540/132723 [00:00<00:01, 68513.57KB/s]
- 43%|####3 | 57099/132723 [00:00<00:01, 70679.52KB/s]
- 49%|####8 | 64719/132723 [00:00<00:00, 72357.98KB/s]
- 54%|#####4 | 72138/132723 [00:01<00:00, 72911.50KB/s]
- 60%|###### | 79790/132723 [00:01<00:00, 74000.67KB/s]
- 66%|######5 | 87391/132723 [00:01<00:00, 74604.83KB/s]
- 72%|#######1 | 95004/132723 [00:01<00:00, 75062.31KB/s]
- 77%|#######7 | 102552/132723 [00:01<00:00, 75182.93KB/s]
- 83%|########3 | 110227/132723 [00:01<00:00, 75652.81KB/s]
- 89%|########8 | 117841/132723 [00:01<00:00, 75794.22KB/s]
- 95%|#########4| 125424/132723 [00:01<00:00, 75772.01KB/s]
-100%|##########| 132723/132723 [00:01<00:00, 72865.56KB/s]
+ 4%|3 | 5041/132723 [00:00<00:02, 50394.55KB/s]
+ 10%|9 | 12799/132723 [00:00<00:01, 66382.15KB/s]
+ 16%|#5 | 20593/132723 [00:00<00:01, 71654.23KB/s]
+ 21%|##1 | 28437/132723 [00:00<00:01, 74330.43KB/s]
+ 27%|##7 | 36266/132723 [00:00<00:01, 75722.39KB/s]
+ 33%|###3 | 44150/132723 [00:00<00:01, 76780.15KB/s]
+ 39%|###9 | 52193/132723 [00:00<00:01, 77971.38KB/s]
+ 45%|####5 | 60362/132723 [00:00<00:00, 79152.79KB/s]
+ 52%|#####1 | 68530/132723 [00:00<00:00, 79939.76KB/s]
+ 58%|#####7 | 76637/132723 [00:01<00:00, 80287.36KB/s]
+ 64%|######3 | 84666/132723 [00:01<00:00, 77459.33KB/s]
+ 70%|######9 | 92788/132723 [00:01<00:00, 78571.90KB/s]
+ 76%|#######5 | 100869/132723 [00:01<00:00, 79233.85KB/s]
+ 82%|########2 | 108983/132723 [00:01<00:00, 79801.12KB/s]
+ 88%|########8 | 117095/132723 [00:01<00:00, 80192.06KB/s]
+ 94%|#########4| 125195/132723 [00:01<00:00, 80429.95KB/s]
+100%|##########| 132723/132723 [00:01<00:00, 77817.54KB/s]
</pre></div>
</div>
<p>Create TVM runtime and do inference
@@ -472,7 +471,7 @@ Downloading /workspace/.mxnet/models/ssd_512_resnet50_v1_voc-9c8b225a.zip from h
</pre></div>
</div>
<img alt="../../_images/sphx_glr_deploy_ssd_gluoncv_001.png" class="sphx-glr-single-img" src="../../_images/sphx_glr_deploy_ssd_gluoncv_001.png" />
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 2 minutes 21.793 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 2 minutes 20.271 seconds)</p>
<div class="sphx-glr-footer class sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-deploy-models-deploy-ssd-gluoncv-py">
<div class="sphx-glr-download 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 49e0ab9..812dd35 100644
--- a/docs/how_to/deploy_models/sg_execution_times.html
+++ b/docs/how_to/deploy_models/sg_execution_times.html
@@ -300,15 +300,15 @@
<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>10:25.692</strong> total execution time for <strong>how_to_deploy_models</strong> files:</p>
+<p><strong>10:30.903</strong> total execution time for <strong>how_to_deploy_models</strong> files:</p>
<ul class="simple">
-<li><p><strong>03:06.085</strong>: <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></li>
-<li><p><strong>02:21.793</strong>: <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></li>
-<li><p><strong>01:52.560</strong>: <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></li>
-<li><p><strong>01:15.134</strong>: <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></li>
-<li><p><strong>01:02.571</strong>: <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></li>
-<li><p><strong>00:26.628</strong>: <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></li>
-<li><p><strong>00:20.715</strong>: <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></li>
+<li><p><strong>03:07.676</strong>: <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></li>
+<li><p><strong>02:20.271</strong>: <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></li>
+<li><p><strong>01:58.957</strong>: <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></li>
+<li><p><strong>01:11.281</strong>: <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></li>
+<li><p><strong>01:03.920</strong>: <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></li>
+<li><p><strong>00:26.856</strong>: <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></li>
+<li><p><strong>00:21.739</strong>: <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></li>
<li><p><strong>00:00.204</strong>: <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></li>
</ul>
</div>
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 779b65a..6830821 100644
--- a/docs/how_to/extend_tvm/bring_your_own_datatypes.html
+++ b/docs/how_to/extend_tvm/bring_your_own_datatypes.html
@@ -588,7 +588,7 @@ In this alpha state of the Bring Your Own Datatypes framework, we have not imple
</pre></div>
</div>
<p class="sphx-glr-script-out">Out:</p>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Downloading /workspace/.mxnet/models/mobilenet0.25-9f83e440.zip0bee6f03-5151-42e6-bd49-fb366a42751b 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.zipaa47f14e-527d-4ca7-8541-3bd59544493d from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/mobilenet0.25-9f83e440.zip...
</pre></div>
</div>
<p>It’s easy to execute MobileNet with native TVM:</p>
diff --git a/docs/how_to/extend_tvm/sg_execution_times.html b/docs/how_to/extend_tvm/sg_execution_times.html
index 21a421e..1b0548f 100644
--- a/docs/how_to/extend_tvm/sg_execution_times.html
+++ b/docs/how_to/extend_tvm/sg_execution_times.html
@@ -300,12 +300,12 @@
<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>00:38.327</strong> total execution time for <strong>how_to_extend_tvm</strong> files:</p>
+<p><strong>00:38.274</strong> total execution time for <strong>how_to_extend_tvm</strong> files:</p>
<ul class="simple">
-<li><p><strong>00:34.744</strong>: <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></li>
-<li><p><strong>00:02.288</strong>: <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></li>
-<li><p><strong>00:01.083</strong>: <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></li>
-<li><p><strong>00:00.211</strong>: <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></li>
+<li><p><strong>00:34.679</strong>: <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></li>
+<li><p><strong>00:02.282</strong>: <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></li>
+<li><p><strong>00:01.094</strong>: <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></li>
+<li><p><strong>00:00.219</strong>: <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></li>
</ul>
</div>
diff --git a/docs/how_to/extend_tvm/use_pass_instrument.html b/docs/how_to/extend_tvm/use_pass_instrument.html
index 65933e6..be57f8f 100644
--- a/docs/how_to/extend_tvm/use_pass_instrument.html
+++ b/docs/how_to/extend_tvm/use_pass_instrument.html
@@ -486,10 +486,10 @@ profile the execution time of each passes.</p>
</div>
<p class="sphx-glr-script-out">Out:</p>
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Printing results of timing profile...
-InferType: 6122us [6122us] (45.27%; 45.27%)
-FoldScaleAxis: 7401us [2us] (54.73%; 54.73%)
- FoldConstant: 7399us [1546us] (54.71%; 99.97%)
- InferType: 5852us [5852us] (43.28%; 79.10%)
+InferType: 5982us [5982us] (45.45%; 45.45%)
+FoldScaleAxis: 7180us [2us] (54.55%; 54.55%)
+ FoldConstant: 7178us [1474us] (54.54%; 99.97%)
+ InferType: 5703us [5703us] (43.33%; 79.46%)
</pre></div>
</div>
</div>
@@ -512,10 +512,10 @@ Refer to following sections and <a class="reference internal" href="../../refere
</div>
<p class="sphx-glr-script-out">Out:</p>
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Printing results of timing profile...
-InferType: 5938us [5938us] (44.77%; 44.77%)
-FoldScaleAxis: 7325us [2us] (55.23%; 55.23%)
- FoldConstant: 7323us [1522us] (55.21%; 99.98%)
- InferType: 5800us [5800us] (43.74%; 79.21%)
+InferType: 5767us [5767us] (44.83%; 44.83%)
+FoldScaleAxis: 7097us [2us] (55.17%; 55.17%)
+ FoldConstant: 7095us [1478us] (55.16%; 99.97%)
+ InferType: 5617us [5617us] (43.67%; 79.17%)
</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 0296cc2..d0f8e99 100644
--- a/docs/how_to/optimize_operators/opt_conv_cuda.html
+++ b/docs/how_to/optimize_operators/opt_conv_cuda.html
@@ -534,7 +534,7 @@ latency of convolution.</p>
</pre></div>
</div>
<p class="sphx-glr-script-out">Out:</p>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Convolution: 45.131203 ms
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Convolution: 35.525673 ms
</pre></div>
</div>
<div class="sphx-glr-footer class 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 66041a8..cafdc69 100644
--- a/docs/how_to/optimize_operators/opt_conv_tensorcore.html
+++ b/docs/how_to/optimize_operators/opt_conv_tensorcore.html
@@ -876,7 +876,7 @@ be able to run on our build server</p>
</pre></div>
</div>
<p class="sphx-glr-script-out">Out:</p>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>conv2d with tensor core: 10.130518 ms
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>conv2d with tensor core: 11.048099 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 85a84b6..aa03793 100644
--- a/docs/how_to/optimize_operators/opt_gemm.html
+++ b/docs/how_to/optimize_operators/opt_gemm.html
@@ -431,8 +431,8 @@ Then we write a baseline implementation, the simplest way to write a matrix mult
</pre></div>
</div>
<p class="sphx-glr-script-out">Out:</p>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Numpy running time: 0.017912
-Baseline: 3.381993
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Numpy running time: 0.018033
+Baseline: 3.296873
</pre></div>
</div>
<p>In TVM, we can always inspect lower level IR to debug or optimize our schedule.
@@ -493,7 +493,7 @@ fill 32 * 32 * sizeof(float) which is 4KB in the cache whose total size is 32KB
</pre></div>
</div>
<p class="sphx-glr-script-out">Out:</p>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt1: 0.308665
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt1: 0.297950
</pre></div>
</div>
<p>Here is the generated IR after blocking.</p>
@@ -561,7 +561,7 @@ vastly.</p>
</pre></div>
</div>
<p class="sphx-glr-script-out">Out:</p>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt2: 0.328278
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt2: 0.327325
</pre></div>
</div>
<p>Here is the generated IR after vectorization.</p>
@@ -623,7 +623,7 @@ the access pattern for A matrix is more cache friendly.</p>
</pre></div>
</div>
<p class="sphx-glr-script-out">Out:</p>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt3: 0.118575
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt3: 0.116769
</pre></div>
</div>
<p>Here is the generated IR after loop permutation.</p>
@@ -707,7 +707,7 @@ flattening.</p>
</pre></div>
</div>
<p class="sphx-glr-script-out">Out:</p>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt4: 0.111692
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt4: 0.111140
</pre></div>
</div>
<p>Here is the generated IR after array packing.</p>
@@ -794,7 +794,7 @@ write to C when all the block results are ready.</p>
</pre></div>
</div>
<p class="sphx-glr-script-out">Out:</p>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt5: 0.111826
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt5: 0.111151
</pre></div>
</div>
<p>Here is the generated IR after blocking.</p>
@@ -885,7 +885,7 @@ write to C when all the block results are ready.</p>
</pre></div>
</div>
<p class="sphx-glr-script-out">Out:</p>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt6: 0.144389
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt6: 0.144714
</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 0fdc25d..93711fa 100644
--- a/docs/how_to/optimize_operators/sg_execution_times.html
+++ b/docs/how_to/optimize_operators/sg_execution_times.html
@@ -300,11 +300,11 @@
<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>00:35.018</strong> total execution time for <strong>how_to_optimize_operators</strong> files:</p>
+<p><strong>00:34.816</strong> total execution time for <strong>how_to_optimize_operators</strong> files:</p>
<ul class="simple">
-<li><p><strong>00:32.353</strong>: <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></li>
-<li><p><strong>00:01.463</strong>: <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></li>
-<li><p><strong>00:01.202</strong>: <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></li>
+<li><p><strong>00:31.900</strong>: <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></li>
+<li><p><strong>00:01.573</strong>: <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></li>
+<li><p><strong>00:01.343</strong>: <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></li>
</ul>
</div>
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 208643c..9f4e23b 100644
--- a/docs/how_to/tune_with_autoscheduler/sg_execution_times.html
+++ b/docs/how_to/tune_with_autoscheduler/sg_execution_times.html
@@ -300,14 +300,14 @@
<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>04:54.312</strong> total execution time for <strong>how_to_tune_with_autoscheduler</strong> files:</p>
+<p><strong>04:56.572</strong> total execution time for <strong>how_to_tune_with_autoscheduler</strong> files:</p>
<ul class="simple">
-<li><p><strong>02:21.084</strong>: <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></li>
-<li><p><strong>01:19.840</strong>: <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></li>
-<li><p><strong>00:40.131</strong>: <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></li>
-<li><p><strong>00:16.340</strong>: <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></li>
-<li><p><strong>00:08.640</strong>: <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></li>
-<li><p><strong>00:08.278</strong>: <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></li>
+<li><p><strong>02:21.977</strong>: <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></li>
+<li><p><strong>01:19.340</strong>: <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></li>
+<li><p><strong>00:40.052</strong>: <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></li>
+<li><p><strong>00:18.821</strong>: <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></li>
+<li><p><strong>00:08.276</strong>: <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></li>
+<li><p><strong>00:08.106</strong>: <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></li>
</ul>
</div>
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 78f46ea..c662d4b 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
@@ -469,74 +469,1163 @@ cooperative fetching, unrolling and operator fusion.</p>
bias: Buffer(bias_2: Pointer(float32), float32, [512], []),
compute: Buffer(compute_2: Pointer(float32), float32, [25088], [])}
buffer_map = {data_1: data, kernel_1: kernel, bias_1: bias, compute_1: compute} {
- attr [IterVar(blockIdx.x: int32, (nullptr), "ThreadIndex", "blockIdx.x")] "thread_extent" = 112;
- allocate(conv2d_nchw: Pointer(local float32), float32, [4]), storage_scope = local;
- allocate(pad_temp.shared: Pointer(shared float32), float32, [72]), storage_scope = shared;
- allocate(kernel.shared: Pointer(shared float32), float32, [768]), storage_scope = shared;
- attr [IterVar(threadIdx.x: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 56 {
- conv2d_nchw_1: Buffer(conv2d_nchw, float32, [4], [], scope="local", align=16)[0] = 0f32
+ attr [IterVar(blockIdx.x: int32, (nullptr), "ThreadIndex", "blockIdx.x")] "thread_extent" = 8;
+ allocate(conv2d_nchw: Pointer(local float32), float32, [28]), storage_scope = local;
+ allocate(pad_temp.shared: Pointer(shared float32), float32, [648]), storage_scope = shared;
+ allocate(kernel.shared: Pointer(shared float32), float32, [4608]), storage_scope = shared;
+ attr [IterVar(threadIdx.x: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112 {
+ conv2d_nchw_1: Buffer(conv2d_nchw, float32, [196], [], scope="local", align=32)[0] = 0f32
+ conv2d_nchw_1[14] = 0f32
conv2d_nchw_1[1] = 0f32
+ conv2d_nchw_1[15] = 0f32
conv2d_nchw_1[2] = 0f32
+ conv2d_nchw_1[16] = 0f32
conv2d_nchw_1[3] = 0f32
+ conv2d_nchw_1[17] = 0f32
+ conv2d_nchw_1[4] = 0f32
+ conv2d_nchw_1[18] = 0f32
+ conv2d_nchw_1[5] = 0f32
+ conv2d_nchw_1[19] = 0f32
+ conv2d_nchw_1[6] = 0f32
+ conv2d_nchw_1[20] = 0f32
+ conv2d_nchw_1[7] = 0f32
+ conv2d_nchw_1[21] = 0f32
+ conv2d_nchw_1[8] = 0f32
+ conv2d_nchw_1[22] = 0f32
+ conv2d_nchw_1[9] = 0f32
+ conv2d_nchw_1[23] = 0f32
+ conv2d_nchw_1[10] = 0f32
+ conv2d_nchw_1[24] = 0f32
+ conv2d_nchw_1[11] = 0f32
+ conv2d_nchw_1[25] = 0f32
+ conv2d_nchw_1[12] = 0f32
+ conv2d_nchw_1[26] = 0f32
+ conv2d_nchw_1[13] = 0f32
+ conv2d_nchw_1[27] = 0f32
for (rc.outer.outer: int32, 0, 64) {
- for (rx.outer.outer: int32, 0, 3) {
- let cse_var_1: int32 = (rc.outer.outer*72)
- {
- attr [IterVar(threadIdx.x_1: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 56;
- pad_temp.shared_1: Buffer(pad_temp.shared, float32, [72], [], scope="shared")[threadIdx.x_1] = @tir.if_then_else(((((1 <= floormod(threadIdx.x_1, 9)) && (floormod(threadIdx.x_1, 9) < 8)) && (1 <= (rx.outer.outer + floormod(blockIdx.x, 7)))) && ((rx.outer.outer + floormod(blockIdx.x, 7)) < 8)), data[((((((rc.outer.outer*392) + (floordiv(threadIdx.x_1, 9)*49)) + (floormod(threadIdx.x_1, 9)*7)) + rx.outer.outer) + floormod(blockIdx.x, [...]
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 56;
- if @tir.likely((threadIdx.x_1 < 16), dtype=bool) {
- pad_temp.shared_1[(threadIdx.x_1 + 56)] = @tir.if_then_else(((((1 <= floormod((threadIdx.x_1 + 2), 9)) && (floormod((threadIdx.x_1 + 2), 9) < 8)) && (1 <= (rx.outer.outer + floormod(blockIdx.x, 7)))) && ((rx.outer.outer + floormod(blockIdx.x, 7)) < 8)), data[((((((rc.outer.outer*392) + (floordiv((threadIdx.x_1 + 56), 9)*49)) + (floormod((threadIdx.x_1 + 2), 9)*7)) + rx.outer.outer) + floormod(blockIdx.x, 7)) - 8)], 0f32, dtype=float32)
- }
- attr [IterVar(threadIdx.x_2: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 56;
- kernel.shared_1: Buffer(kernel.shared, float32, [768], [], scope="shared")[threadIdx.x_2] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv(threadIdx.x_2, 24)*4608)) + cse_var_1) + (floormod(threadIdx.x_2, 24)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 56;
- kernel.shared_1[(threadIdx.x_2 + 56)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((floordiv(threadIdx.x_2, 8) + 7), 3)*4608)) + cse_var_1) + (floormod((threadIdx.x_2 + 8), 24)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 56;
- kernel.shared_1[(threadIdx.x_2 + 112)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((floordiv(threadIdx.x_2, 8) + 14), 3)*4608)) + cse_var_1) + (floormod((threadIdx.x_2 + 16), 24)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 56;
- kernel.shared_1[(threadIdx.x_2 + 168)] = kernel[((((((floordiv(blockIdx.x, 7)*147456) + (floordiv(floordiv(threadIdx.x_2, 8), 3)*4608)) + cse_var_1) + (floormod(threadIdx.x_2, 24)*3)) + rx.outer.outer) + 32256)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 56;
- kernel.shared_1[(threadIdx.x_2 + 224)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((floordiv(threadIdx.x_2, 8) + 28), 3)*4608)) + cse_var_1) + (floormod((threadIdx.x_2 + 8), 24)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 56;
- kernel.shared_1[(threadIdx.x_2 + 280)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((floordiv(threadIdx.x_2, 8) + 35), 3)*4608)) + cse_var_1) + (floormod((threadIdx.x_2 + 16), 24)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 56;
- kernel.shared_1[(threadIdx.x_2 + 336)] = kernel[((((((floordiv(blockIdx.x, 7)*147456) + (floordiv(floordiv(threadIdx.x_2, 8), 3)*4608)) + cse_var_1) + (floormod(threadIdx.x_2, 24)*3)) + rx.outer.outer) + 64512)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 56;
- kernel.shared_1[(threadIdx.x_2 + 392)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((floordiv(threadIdx.x_2, 8) + 49), 3)*4608)) + cse_var_1) + (floormod((threadIdx.x_2 + 8), 24)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 56;
- kernel.shared_1[(threadIdx.x_2 + 448)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((floordiv(threadIdx.x_2, 8) + 56), 3)*4608)) + cse_var_1) + (floormod((threadIdx.x_2 + 16), 24)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 56;
- kernel.shared_1[(threadIdx.x_2 + 504)] = kernel[((((((floordiv(blockIdx.x, 7)*147456) + (floordiv(floordiv(threadIdx.x_2, 8), 3)*4608)) + cse_var_1) + (floormod(threadIdx.x_2, 24)*3)) + rx.outer.outer) + 96768)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 56;
- kernel.shared_1[(threadIdx.x_2 + 560)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((floordiv(threadIdx.x_2, 8) + 70), 3)*4608)) + cse_var_1) + (floormod((threadIdx.x_2 + 8), 24)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 56;
- kernel.shared_1[(threadIdx.x_2 + 616)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((floordiv(threadIdx.x_2, 8) + 77), 3)*4608)) + cse_var_1) + (floormod((threadIdx.x_2 + 16), 24)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 56;
- kernel.shared_1[(threadIdx.x_2 + 672)] = kernel[((((((floordiv(blockIdx.x, 7)*147456) + (floordiv(floordiv(threadIdx.x_2, 8), 3)*4608)) + cse_var_1) + (floormod(threadIdx.x_2, 24)*3)) + rx.outer.outer) + 129024)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 56;
- if @tir.likely((threadIdx.x_2 < 40), dtype=bool) {
- kernel.shared_1[(threadIdx.x_2 + 728)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((floordiv(threadIdx.x_2, 8) + 91), 3)*4608)) + cse_var_1) + (floormod((threadIdx.x_2 + 8), 24)*3)) + rx.outer.outer)]
- }
- for (rc.outer.inner: int32, 0, 8) {
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((rc.outer.inner*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[((floordiv(threadIdx.x, 7)*96) + (rc.outer.inner*3))]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*96) + (rc.outer.inner*3)) + 1)]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*96) + (rc.outer.inner*3)) + 2)]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((rc.outer.inner*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*96) + (rc.outer.inner*3)) + 24)]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*96) + (rc.outer.inner*3)) + 25)]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*96) + (rc.outer.inner*3)) + 26)]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((rc.outer.inner*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*96) + (rc.outer.inner*3)) + 48)]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*96) + (rc.outer.inner*3)) + 49)]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*96) + (rc.outer.inner*3)) + 50)]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((rc.outer.inner*9) + floormod(threadIdx.x, 7))]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*96) + (rc.outer.inner*3)) + 72)]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*9) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*96) + (rc.outer.inner*3)) + 73)]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*9) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*96) + (rc.outer.inner*3)) + 74)]))
- }
+ let cse_var_2: int32 = (rc.outer.outer*392)
+ let cse_var_1: int32 = (rc.outer.outer*72)
+ {
+ attr [IterVar(threadIdx.x_1: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ pad_temp.shared_1: Buffer(pad_temp.shared, float32, [648], [], scope="shared")[threadIdx.x_1] = @tir.if_then_else(((((9 <= floormod(threadIdx.x_1, 81)) && (floormod(threadIdx.x_1, 81) < 72)) && (1 <= floormod(threadIdx.x_1, 9))) && (floormod(threadIdx.x_1, 9) < 8)), data[((((cse_var_2 + (floordiv(threadIdx.x_1, 81)*49)) + (floordiv(floormod(threadIdx.x_1, 81), 9)*7)) + floormod(threadIdx.x_1, 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ pad_temp.shared_1[(threadIdx.x_1 + 112)] = @tir.if_then_else(((((9 <= floormod((threadIdx.x_1 + 112), 81)) && (floormod((threadIdx.x_1 + 31), 81) < 72)) && (1 <= floormod((threadIdx.x_1 + 4), 9))) && (floormod((threadIdx.x_1 + 4), 9) < 8)), data[((((cse_var_2 + (floordiv((threadIdx.x_1 + 112), 81)*49)) + (floordiv(floormod((threadIdx.x_1 + 112), 81), 9)*7)) + floormod((threadIdx.x_1 + 4), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ pad_temp.shared_1[(threadIdx.x_1 + 224)] = @tir.if_then_else(((((9 <= floormod((threadIdx.x_1 + 224), 81)) && (floormod((threadIdx.x_1 + 62), 81) < 72)) && (1 <= floormod((threadIdx.x_1 + 8), 9))) && (floormod((threadIdx.x_1 + 8), 9) < 8)), data[((((cse_var_2 + (floordiv((threadIdx.x_1 + 224), 81)*49)) + (floordiv(floormod((threadIdx.x_1 + 224), 81), 9)*7)) + floormod((threadIdx.x_1 + 8), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ pad_temp.shared_1[(threadIdx.x_1 + 336)] = @tir.if_then_else(((((9 <= floormod((threadIdx.x_1 + 336), 81)) && (floormod((threadIdx.x_1 + 12), 81) < 72)) && (1 <= floormod((threadIdx.x_1 + 3), 9))) && (floormod((threadIdx.x_1 + 3), 9) < 8)), data[((((cse_var_2 + (floordiv((threadIdx.x_1 + 336), 81)*49)) + (floordiv(floormod((threadIdx.x_1 + 336), 81), 9)*7)) + floormod((threadIdx.x_1 + 3), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ pad_temp.shared_1[(threadIdx.x_1 + 448)] = @tir.if_then_else(((((9 <= floormod((threadIdx.x_1 + 448), 81)) && (floormod((threadIdx.x_1 + 43), 81) < 72)) && (1 <= floormod((threadIdx.x_1 + 7), 9))) && (floormod((threadIdx.x_1 + 7), 9) < 8)), data[((((cse_var_2 + (floordiv((threadIdx.x_1 + 448), 81)*49)) + (floordiv(floormod((threadIdx.x_1 + 448), 81), 9)*7)) + floormod((threadIdx.x_1 + 7), 9)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ if @tir.likely((threadIdx.x_1 < 88), dtype=bool) {
+ pad_temp.shared_1[(threadIdx.x_1 + 560)] = @tir.if_then_else(((((9 <= floormod((threadIdx.x_1 + 560), 81)) && (floormod((threadIdx.x_1 + 74), 81) < 72)) && (1 <= floormod((threadIdx.x_1 + 2), 9))) && (floormod((threadIdx.x_1 + 2), 9) < 8)), data[((((cse_var_2 + (floordiv((threadIdx.x_1 + 560), 81)*49)) + (floordiv(floormod((threadIdx.x_1 + 560), 81), 9)*7)) + floormod((threadIdx.x_1 + 2), 9)) - 8)], 0f32, dtype=float32)
+ }
+ attr [IterVar(threadIdx.x_2: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1: Buffer(kernel.shared, float32, [4608], [], scope="shared")[threadIdx.x_2] = kernel[((((blockIdx.x*294912) + (floordiv(threadIdx.x_2, 72)*4608)) + cse_var_1) + floormod(threadIdx.x_2, 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 112)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 14), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 40), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 224)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 28), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 8), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 336)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 42), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 48), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 448)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 56), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 16), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 560)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 70), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 56), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 672)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 84), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 24), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 784)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 98), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 64), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 896)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 112), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 32), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 1008)] = kernel[(((((blockIdx.x*294912) + (floordiv(floordiv(threadIdx.x_2, 8), 9)*4608)) + cse_var_1) + floormod(threadIdx.x_2, 72)) + 64512)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 1120)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 140), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 40), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 1232)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 154), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 8), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 1344)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 168), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 48), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 1456)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 182), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 16), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 1568)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 196), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 56), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 1680)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 210), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 24), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 1792)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 224), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 64), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 1904)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 238), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 32), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 2016)] = kernel[(((((blockIdx.x*294912) + (floordiv(floordiv(threadIdx.x_2, 8), 9)*4608)) + cse_var_1) + floormod(threadIdx.x_2, 72)) + 129024)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 2128)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 266), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 40), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 2240)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 280), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 8), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 2352)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 294), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 48), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 2464)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 308), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 16), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 2576)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 322), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 56), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 2688)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 336), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 24), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 2800)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 350), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 64), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 2912)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 364), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 32), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 3024)] = kernel[(((((blockIdx.x*294912) + (floordiv(floordiv(threadIdx.x_2, 8), 9)*4608)) + cse_var_1) + floormod(threadIdx.x_2, 72)) + 193536)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 3136)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 392), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 40), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 3248)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 406), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 8), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 3360)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 420), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 48), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 3472)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 434), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 16), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 3584)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 448), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 56), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 3696)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 462), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 24), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 3808)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 476), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 64), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 3920)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 490), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 32), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 4032)] = kernel[(((((blockIdx.x*294912) + (floordiv(floordiv(threadIdx.x_2, 8), 9)*4608)) + cse_var_1) + floormod(threadIdx.x_2, 72)) + 258048)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 4144)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 518), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 40), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 4256)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 532), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 8), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 4368)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 546), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 48), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 4480)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 560), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 16), 72))]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ if @tir.likely((threadIdx.x_2 < 16), dtype=bool) {
+ kernel.shared_1[(threadIdx.x_2 + 4592)] = kernel[((((blockIdx.x*294912) + (floordiv((floordiv(threadIdx.x_2, 8) + 574), 9)*4608)) + cse_var_1) + floormod((threadIdx.x_2 + 56), 72))]
+ }
+ for (rc.outer.inner: int32, 0, 2) {
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((rc.outer.inner*324) + floormod(threadIdx.x, 7))]*kernel.shared_1[((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36))]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[((rc.outer.inner*324) + floormod(threadIdx.x, 7))]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2304)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 1)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2305)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2306)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 81)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 9)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 81)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2313)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 82)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 10)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 82)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2314)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 83)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 11)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 83)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2315)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 162)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 18)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 162)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2322)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 163)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 19)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 163)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2323)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 164)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 20)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 164)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2324)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 243)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 27)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 243)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2331)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 244)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 28)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 244)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2332)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 245)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 29)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 245)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2333)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 9)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36))]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 9)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2304)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 10)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 1)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 10)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2305)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 11)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 11)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2306)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 90)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 9)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 90)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2313)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 91)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 10)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 91)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2314)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 92)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 11)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 92)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2315)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 171)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 18)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 171)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2322)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 172)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 19)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 172)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2323)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 173)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 20)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 173)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2324)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 27)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2331)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 28)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2332)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 29)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2333)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36))]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2304)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 1)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2305)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2306)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 9)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2313)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 10)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2314)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 11)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2315)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 18)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2322)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 19)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2323)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 20)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2324)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 27)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2331)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 28)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2332)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 29)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2333)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 27)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36))]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 27)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2304)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 28)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 1)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 28)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2305)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 29)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 29)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2306)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 108)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 9)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 108)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2313)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 109)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 10)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 109)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2314)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 110)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 11)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 110)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2315)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 18)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2322)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 19)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2323)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 20)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2324)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 270)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 27)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 270)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2331)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 271)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 28)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 271)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2332)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 272)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 29)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 272)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2333)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 36)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36))]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 36)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2304)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 37)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 1)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 37)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2305)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 38)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 38)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2306)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 117)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 9)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 117)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2313)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 118)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 10)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 118)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2314)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 119)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 11)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 119)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2315)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 198)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 18)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 198)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2322)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 199)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 19)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 199)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2323)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 200)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 20)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 200)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2324)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 279)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 27)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 279)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2331)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 280)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 28)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 280)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2332)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 281)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 29)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 281)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2333)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 45)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36))]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 45)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2304)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 46)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 1)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 46)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2305)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 47)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 47)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2306)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 9)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2313)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 10)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2314)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 11)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2315)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 207)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 18)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 207)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2322)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 208)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 19)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 208)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2323)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 209)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 20)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 209)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2324)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 288)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 27)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 288)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2331)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 289)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 28)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 289)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2332)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 290)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 29)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 290)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2333)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 54)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36))]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 54)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2304)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 55)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 1)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 55)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2305)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 56)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 56)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2306)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 135)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 9)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 135)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2313)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 136)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 10)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 136)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2314)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 137)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 11)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 137)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2315)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 216)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 18)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 216)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2322)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 217)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 19)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 217)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2323)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 218)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 20)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 218)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2324)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 297)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 27)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 297)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2331)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 298)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 28)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 298)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2332)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 299)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 29)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 299)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2333)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((rc.outer.inner*324) + floormod(threadIdx.x, 7))]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 72)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[((rc.outer.inner*324) + floormod(threadIdx.x, 7))]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2376)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 73)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 1)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2377)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 74)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 2)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2378)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 81)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 81)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 81)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2385)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 82)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 82)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 82)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2386)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 83)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 83)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 83)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2387)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 162)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 90)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 162)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2394)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 163)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 91)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 163)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2395)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 164)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 92)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 164)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2396)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 243)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 99)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 243)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2403)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 244)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 100)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 244)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2404)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 245)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 101)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 245)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2405)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 9)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 72)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 9)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2376)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 10)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 73)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 10)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2377)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 11)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 74)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 11)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2378)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 90)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 81)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 90)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2385)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 91)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 82)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 91)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2386)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 92)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 83)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 92)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2387)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 171)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 90)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 171)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2394)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 172)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 91)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 172)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2395)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 173)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 92)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 173)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2396)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 99)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2403)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 100)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2404)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 101)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2405)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 72)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2376)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 73)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2377)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 74)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2378)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 81)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2385)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 82)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2386)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 83)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2387)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 90)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2394)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 91)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2395)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 92)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2396)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 99)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2403)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 100)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2404)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 101)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2405)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 27)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 72)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 27)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2376)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 28)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 73)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 28)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2377)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 29)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 74)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 29)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2378)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 108)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 81)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 108)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2385)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 109)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 82)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 109)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2386)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 110)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 83)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 110)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2387)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 90)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2394)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 91)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2395)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 92)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2396)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 270)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 99)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 270)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2403)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 271)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 100)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 271)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2404)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 272)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 101)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 272)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2405)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 36)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 72)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 36)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2376)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 37)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 73)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 37)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2377)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 38)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 74)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 38)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2378)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 117)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 81)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 117)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2385)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 118)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 82)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 118)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2386)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 119)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 83)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 119)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2387)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 198)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 90)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 198)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2394)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 199)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 91)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 199)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2395)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 200)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 92)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 200)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2396)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 279)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 99)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 279)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2403)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 280)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 100)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 280)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2404)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 281)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 101)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 281)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2405)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 45)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 72)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 45)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2376)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 46)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 73)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 46)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2377)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 47)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 74)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 47)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2378)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 81)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2385)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 82)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2386)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 83)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2387)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 207)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 90)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 207)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2394)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 208)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 91)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 208)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2395)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 209)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 92)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 209)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2396)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 288)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 99)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 288)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2403)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 289)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 100)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 289)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2404)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 290)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 101)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 290)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2405)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 54)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 72)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 54)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2376)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 55)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 73)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 55)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2377)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 56)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 74)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 56)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2378)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 135)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 81)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 135)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2385)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 136)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 82)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 136)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2386)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 137)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 83)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 137)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2387)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 216)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 90)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 216)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2394)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 217)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 91)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 217)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2395)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 218)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 92)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 218)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2396)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 297)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 99)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 297)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2403)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 298)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 100)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 298)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2404)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 299)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 101)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 299)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2405)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 9)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 3)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 9)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2307)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 10)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 4)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 10)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2308)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 11)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 5)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 11)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2309)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 90)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 12)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 90)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2316)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 91)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 13)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 91)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2317)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 92)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 14)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 92)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2318)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 171)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 21)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 171)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2325)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 172)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 22)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 172)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2326)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 173)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 23)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 173)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2327)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 30)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2334)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 31)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2335)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 32)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2336)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 3)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2307)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 4)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2308)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 5)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2309)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 12)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2316)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 13)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2317)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 14)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2318)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 21)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2325)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 22)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2326)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 23)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2327)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 30)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2334)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 31)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2335)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 32)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2336)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 27)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 3)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 27)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2307)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 28)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 4)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 28)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2308)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 29)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 5)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 29)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2309)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 108)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 12)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 108)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2316)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 109)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 13)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 109)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2317)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 110)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 14)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 110)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2318)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 21)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2325)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 22)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2326)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 23)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2327)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 270)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 30)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 270)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2334)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 271)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 31)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 271)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2335)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 272)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 32)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 272)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2336)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 36)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 3)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 36)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2307)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 37)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 4)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 37)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2308)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 38)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 5)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 38)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2309)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 117)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 12)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 117)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2316)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 118)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 13)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 118)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2317)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 119)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 14)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 119)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2318)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 198)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 21)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 198)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2325)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 199)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 22)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 199)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2326)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 200)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 23)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 200)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2327)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 279)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 30)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 279)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2334)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 280)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 31)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 280)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2335)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 281)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 32)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 281)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2336)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 45)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 3)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 45)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2307)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 46)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 4)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 46)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2308)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 47)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 5)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 47)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2309)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 12)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2316)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 13)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2317)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 14)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2318)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 207)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 21)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 207)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2325)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 208)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 22)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 208)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2326)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 209)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 23)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 209)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2327)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 288)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 30)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 288)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2334)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 289)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 31)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 289)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2335)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 290)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 32)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 290)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2336)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 54)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 3)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 54)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2307)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 55)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 4)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 55)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2308)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 56)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 5)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 56)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2309)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 135)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 12)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 135)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2316)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 136)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 13)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 136)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2317)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 137)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 14)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 137)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2318)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 216)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 21)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 216)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2325)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 217)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 22)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 217)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2326)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 218)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 23)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 218)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2327)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 297)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 30)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 297)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2334)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 298)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 31)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 298)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2335)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 299)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 32)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 299)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2336)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 3)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2307)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 4)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2308)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 5)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2309)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 144)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 12)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 144)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2316)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 145)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 13)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 145)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2317)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 146)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 14)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 146)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2318)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 225)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 21)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 225)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2325)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 226)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 22)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 226)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2326)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 227)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 23)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 227)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2327)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 306)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 30)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 306)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2334)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 307)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 31)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 307)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2335)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 308)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 32)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 308)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2336)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 9)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 75)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 9)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2379)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 10)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 76)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 10)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2380)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 11)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 77)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 11)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2381)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 90)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 84)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 90)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2388)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 91)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 85)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 91)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2389)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 92)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 86)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 92)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2390)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 171)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 93)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 171)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2397)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 172)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 94)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 172)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2398)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 173)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 95)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 173)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2399)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 102)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 252)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2406)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 103)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 253)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2407)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 104)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 254)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2408)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 75)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2379)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 76)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2380)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 77)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2381)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 84)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2388)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 85)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2389)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 86)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2390)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 93)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2397)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 94)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2398)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 95)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2399)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 102)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2406)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 103)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2407)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 104)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2408)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 27)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 75)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 27)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2379)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 28)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 76)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 28)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2380)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 29)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 77)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 29)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2381)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 108)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 84)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 108)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2388)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 109)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 85)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 109)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2389)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 110)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 86)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 110)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2390)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 93)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2397)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 94)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2398)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 95)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2399)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 270)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 102)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 270)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2406)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 271)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 103)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 271)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2407)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 272)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 104)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 272)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2408)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 36)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 75)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 36)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2379)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 37)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 76)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 37)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2380)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 38)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 77)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 38)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2381)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 117)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 84)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 117)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2388)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 118)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 85)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 118)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2389)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 119)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 86)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 119)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2390)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 198)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 93)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 198)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2397)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 199)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 94)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 199)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2398)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 200)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 95)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 200)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2399)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 279)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 102)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 279)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2406)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 280)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 103)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 280)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2407)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 281)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 104)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 281)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2408)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 45)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 75)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 45)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2379)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 46)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 76)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 46)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2380)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 47)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 77)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 47)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2381)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 84)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2388)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 85)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2389)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 86)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2390)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 207)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 93)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 207)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2397)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 208)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 94)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 208)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2398)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 209)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 95)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 209)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2399)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 288)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 102)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 288)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2406)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 289)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 103)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 289)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2407)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 290)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 104)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 290)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2408)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 54)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 75)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 54)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2379)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 55)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 76)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 55)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2380)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 56)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 77)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 56)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2381)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 135)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 84)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 135)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2388)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 136)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 85)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 136)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2389)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 137)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 86)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 137)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2390)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 216)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 93)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 216)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2397)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 217)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 94)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 217)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2398)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 218)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 95)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 218)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2399)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 297)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 102)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 297)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2406)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 298)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 103)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 298)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2407)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 299)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 104)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 299)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2408)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 75)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2379)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 76)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2380)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 77)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2381)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 144)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 84)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 144)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2388)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 145)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 85)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 145)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2389)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 146)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 86)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 146)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2390)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 225)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 93)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 225)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2397)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 226)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 94)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 226)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2398)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 227)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 95)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 227)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2399)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 306)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 102)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 306)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2406)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 307)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 103)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 307)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2407)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 308)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 104)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 308)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2408)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 6)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2310)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 7)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2311)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 8)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2312)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 15)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2319)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 16)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2320)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 17)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2321)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 24)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2328)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 25)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2329)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 26)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2330)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 33)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2337)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 34)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2338)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 35)]))
+ conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2339)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 27)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 6)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 27)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2310)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 28)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 7)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 28)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2311)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 29)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 8)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 29)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2312)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 108)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 15)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 108)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2319)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 109)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 16)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 109)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2320)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 110)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 17)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 110)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2321)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 24)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2328)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 25)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2329)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 26)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2330)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 270)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 33)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 270)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2337)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 271)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 34)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 271)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2338)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 272)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 35)]))
+ conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 272)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2339)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 36)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 6)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 36)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2310)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 37)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 7)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 37)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2311)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 38)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 8)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 38)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2312)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 117)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 15)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 117)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2319)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 118)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 16)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 118)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2320)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 119)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 17)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 119)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2321)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 198)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 24)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 198)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2328)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 199)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 25)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 199)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2329)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 200)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 26)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 200)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2330)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 279)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 33)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 279)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2337)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 280)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 34)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 280)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2338)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 281)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 35)]))
+ conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 281)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2339)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 45)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 6)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 45)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2310)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 46)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 7)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 46)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2311)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 47)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 8)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 47)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2312)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 15)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2319)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 16)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2320)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 17)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2321)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 207)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 24)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 207)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2328)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 208)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 25)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 208)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2329)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 209)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 26)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 209)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2330)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 288)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 33)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 288)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2337)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 289)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 34)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 289)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2338)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 290)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 35)]))
+ conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 290)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2339)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 54)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 6)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 54)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2310)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 55)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 7)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 55)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2311)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 56)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 8)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 56)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2312)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 135)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 15)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 135)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2319)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 136)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 16)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 136)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2320)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 137)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 17)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 137)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2321)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 216)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 24)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 216)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2328)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 217)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 25)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 217)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2329)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 218)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 26)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 218)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2330)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 297)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 33)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 297)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2337)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 298)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 34)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 298)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2338)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 299)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 35)]))
+ conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 299)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2339)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 6)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2310)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 7)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2311)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 8)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2312)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 144)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 15)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 144)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2319)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 145)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 16)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 145)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2320)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 146)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 17)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 146)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2321)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 225)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 24)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 225)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2328)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 226)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 25)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 226)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2329)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 227)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 26)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 227)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2330)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 306)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 33)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 306)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2337)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 307)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 34)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 307)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2338)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 308)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 35)]))
+ conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 308)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2339)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 72)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 6)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 72)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2310)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 73)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 7)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 73)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2311)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 74)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 8)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 74)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2312)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 153)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 15)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 153)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2319)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 154)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 16)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 154)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2320)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 155)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 17)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 155)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2321)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 234)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 24)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 234)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2328)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 235)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 25)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 235)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2329)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 236)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 26)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 236)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2330)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 315)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 33)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 315)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2337)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 316)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 34)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 316)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2338)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 317)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 35)]))
+ conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 317)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2339)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 78)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 18)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2382)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 79)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 19)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2383)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 80)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 20)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2384)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 87)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 99)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2391)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 88)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 100)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2392)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 89)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 101)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2393)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 96)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 180)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2400)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 97)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 181)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2401)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 98)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 182)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2402)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 105)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 261)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2409)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 106)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 262)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2410)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 107)]))
+ conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 263)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2411)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 27)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 78)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 27)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2382)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 28)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 79)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 28)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2383)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 29)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 80)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 29)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2384)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 108)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 87)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 108)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2391)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 109)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 88)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 109)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2392)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 110)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 89)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 110)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2393)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 96)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 189)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2400)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 97)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 190)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2401)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 98)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 191)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2402)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 270)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 105)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 270)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2409)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 271)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 106)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 271)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2410)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 272)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 107)]))
+ conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 272)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2411)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 36)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 78)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 36)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2382)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 37)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 79)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 37)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2383)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 38)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 80)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 38)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2384)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 117)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 87)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 117)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2391)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 118)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 88)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 118)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2392)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 119)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 89)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 119)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2393)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 198)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 96)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 198)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2400)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 199)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 97)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 199)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2401)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 200)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 98)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 200)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2402)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 279)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 105)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 279)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2409)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 280)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 106)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 280)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2410)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 281)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 107)]))
+ conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 281)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2411)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 45)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 78)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 45)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2382)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 46)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 79)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 46)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2383)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 47)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 80)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 47)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2384)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 87)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 126)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2391)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 88)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 127)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2392)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 89)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 128)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2393)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 207)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 96)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 207)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2400)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 208)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 97)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 208)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2401)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 209)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 98)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 209)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2402)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 288)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 105)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 288)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2409)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 289)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 106)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 289)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2410)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 290)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 107)]))
+ conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 290)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2411)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 54)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 78)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 54)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2382)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 55)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 79)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 55)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2383)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 56)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 80)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 56)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2384)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 135)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 87)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 135)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2391)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 136)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 88)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 136)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2392)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 137)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 89)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 137)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2393)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 216)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 96)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 216)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2400)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 217)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 97)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 217)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2401)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 218)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 98)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 218)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2402)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 297)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 105)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 297)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2409)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 298)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 106)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 298)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2410)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 299)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 107)]))
+ conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 299)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2411)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 78)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 63)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2382)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 79)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 64)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2383)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 80)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 65)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2384)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 144)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 87)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 144)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2391)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 145)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 88)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 145)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2392)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 146)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 89)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 146)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2393)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 225)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 96)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 225)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2400)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 226)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 97)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 226)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2401)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 227)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 98)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 227)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2402)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 306)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 105)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 306)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2409)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 307)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 106)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 307)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2410)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 308)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 107)]))
+ conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 308)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2411)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 72)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 78)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 72)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2382)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 73)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 79)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 73)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2383)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 74)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 80)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 74)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2384)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 153)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 87)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 153)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2391)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 154)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 88)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 154)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2392)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 155)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 89)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 155)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2393)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 234)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 96)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 234)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2400)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 235)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 97)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 235)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2401)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 236)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 98)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 236)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2402)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 315)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 105)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 315)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2409)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 316)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 106)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 316)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2410)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 317)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 107)]))
+ conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(((rc.outer.inner*324) + floormod(threadIdx.x, 7)) + 317)]*kernel.shared_1[(((floordiv(threadIdx.x, 7)*144) + (rc.outer.inner*36)) + 2411)]))
}
}
}
- for (i1.inner: int32, 0, 4) {
- compute[(((((floordiv(blockIdx.x, 7)*1568) + (floordiv(threadIdx.x, 7)*196)) + (i1.inner*49)) + (floormod(threadIdx.x, 7)*7)) + floormod(blockIdx.x, 7))] = max((conv2d_nchw_1[i1.inner] + bias[(((floordiv(blockIdx.x, 7)*32) + (floordiv(threadIdx.x, 7)*4)) + i1.inner)]), 0f32)
+ for (i1.inner: int32, 0, 2) {
+ for (i2.inner: int32, 0, 7) {
+ let cse_var_3: int32 = ((i1.inner*7) + i2.inner)
+ {
+ compute[(((((blockIdx.x*3136) + (floordiv(threadIdx.x, 7)*98)) + (i1.inner*49)) + (i2.inner*7)) + floormod(threadIdx.x, 7))] = max((conv2d_nchw_1[cse_var_3] + bias[(((blockIdx.x*64) + (floordiv(threadIdx.x, 7)*2)) + i1.inner)]), 0f32)
+ compute[((((((blockIdx.x*3136) + (floordiv(threadIdx.x, 7)*98)) + (i1.inner*49)) + (i2.inner*7)) + floormod(threadIdx.x, 7)) + 1568)] = max((conv2d_nchw_1[(cse_var_3 + 14)] + bias[((((blockIdx.x*64) + (floordiv(threadIdx.x, 7)*2)) + i1.inner) + 32)]), 0f32)
+ }
+ }
}
}
}
@@ -574,7 +1663,7 @@ cooperative fetching, unrolling and operator fusion.</p>
</pre></div>
</div>
<p class="sphx-glr-script-out">Out:</p>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Execution time of this operator: 0.469 ms
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Execution time of this operator: 0.340 ms
</pre></div>
</div>
</div>
@@ -605,35 +1694,35 @@ conv2d_nchw_nn_o_o_i, conv2d_nchw_nn_o_i = s[conv2d_nchw].split(conv2d_nchw_nn_o
conv2d_nchw_nn_o_o_o_i, conv2d_nchw_nn_o_o_i = s[conv2d_nchw].split(conv2d_nchw_nn_o_o_i, factor=1)
conv2d_nchw_nn_o_o_o_o, conv2d_nchw_nn_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_nn_o_o_o_i, factor=1)
conv2d_nchw_ff_o_i, conv2d_nchw_ff_i = s[conv2d_nchw].split(conv2d_nchw_ff, factor=1)
-conv2d_nchw_ff_o_o_i, conv2d_nchw_ff_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_i, factor=4)
-conv2d_nchw_ff_o_o_o_i, conv2d_nchw_ff_o_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_o_i, factor=8)
-conv2d_nchw_ff_o_o_o_o, conv2d_nchw_ff_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_o_o_i, factor=1)
+conv2d_nchw_ff_o_o_i, conv2d_nchw_ff_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_i, factor=2)
+conv2d_nchw_ff_o_o_o_i, conv2d_nchw_ff_o_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_o_i, factor=16)
+conv2d_nchw_ff_o_o_o_o, conv2d_nchw_ff_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_o_o_i, factor=2)
conv2d_nchw_yy_o_i, conv2d_nchw_yy_i = s[conv2d_nchw].split(conv2d_nchw_yy, factor=1)
-conv2d_nchw_yy_o_o_i, conv2d_nchw_yy_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_i, factor=1)
-conv2d_nchw_yy_o_o_o_i, conv2d_nchw_yy_o_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_o_i, factor=7)
+conv2d_nchw_yy_o_o_i, conv2d_nchw_yy_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_i, factor=7)
+conv2d_nchw_yy_o_o_o_i, conv2d_nchw_yy_o_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_o_i, factor=1)
conv2d_nchw_yy_o_o_o_o, conv2d_nchw_yy_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_o_o_i, factor=1)
conv2d_nchw_xx_o_i, conv2d_nchw_xx_i = s[conv2d_nchw].split(conv2d_nchw_xx, factor=1)
conv2d_nchw_xx_o_o_i, conv2d_nchw_xx_o_i = s[conv2d_nchw].split(conv2d_nchw_xx_o_i, factor=1)
-conv2d_nchw_xx_o_o_o_i, conv2d_nchw_xx_o_o_i = s[conv2d_nchw].split(conv2d_nchw_xx_o_o_i, factor=1)
+conv2d_nchw_xx_o_o_o_i, conv2d_nchw_xx_o_o_i = s[conv2d_nchw].split(conv2d_nchw_xx_o_o_i, factor=7)
conv2d_nchw_xx_o_o_o_o, conv2d_nchw_xx_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_xx_o_o_o_i, factor=1)
-conv2d_nchw_rc_o_i, conv2d_nchw_rc_i = s[conv2d_nchw].split(conv2d_nchw_rc, factor=1)
-conv2d_nchw_rc_o_o, conv2d_nchw_rc_o_i = s[conv2d_nchw].split(conv2d_nchw_rc_o_i, factor=8)
-conv2d_nchw_ry_o_i, conv2d_nchw_ry_i = s[conv2d_nchw].split(conv2d_nchw_ry, factor=3)
-conv2d_nchw_ry_o_o, conv2d_nchw_ry_o_i = s[conv2d_nchw].split(conv2d_nchw_ry_o_i, factor=1)
-conv2d_nchw_rx_o_i, conv2d_nchw_rx_i = s[conv2d_nchw].split(conv2d_nchw_rx, factor=1)
+conv2d_nchw_rc_o_i, conv2d_nchw_rc_i = s[conv2d_nchw].split(conv2d_nchw_rc, factor=4)
+conv2d_nchw_rc_o_o, conv2d_nchw_rc_o_i = s[conv2d_nchw].split(conv2d_nchw_rc_o_i, factor=2)
+conv2d_nchw_ry_o_i, conv2d_nchw_ry_i = s[conv2d_nchw].split(conv2d_nchw_ry, factor=1)
+conv2d_nchw_ry_o_o, conv2d_nchw_ry_o_i = s[conv2d_nchw].split(conv2d_nchw_ry_o_i, factor=3)
+conv2d_nchw_rx_o_i, conv2d_nchw_rx_i = s[conv2d_nchw].split(conv2d_nchw_rx, factor=3)
conv2d_nchw_rx_o_o, conv2d_nchw_rx_o_i = s[conv2d_nchw].split(conv2d_nchw_rx_o_i, factor=1)
s[conv2d_nchw].reorder(conv2d_nchw_nn_o_o_o_o, conv2d_nchw_ff_o_o_o_o, conv2d_nchw_yy_o_o_o_o, conv2d_nchw_xx_o_o_o_o, conv2d_nchw_nn_o_o_o_i, conv2d_nchw_ff_o_o_o_i, conv2d_nchw_yy_o_o_o_i, conv2d_nchw_xx_o_o_o_i, conv2d_nchw_nn_o_o_i, conv2d_nchw_ff_o_o_i, conv2d_nchw_yy_o_o_i, conv2d_nchw_xx_o_o_i, conv2d_nchw_rc_o_o, conv2d_nchw_ry_o_o, conv2d_nchw_rx_o_o, conv2d_nchw_rc_o_i, conv2d_nchw_ry_o_i, conv2d_nchw_rx_o_i, conv2d_nchw_nn_o_i, conv2d_nchw_ff_o_i, conv2d_nchw_yy_o_i, conv2d_nc [...]
compute_i0_o_i, compute_i0_i = s[compute].split(compute_i0, factor=1)
compute_i0_o_o_i, compute_i0_o_i = s[compute].split(compute_i0_o_i, factor=1)
compute_i0_o_o_o, compute_i0_o_o_i = s[compute].split(compute_i0_o_o_i, factor=1)
-compute_i1_o_i, compute_i1_i = s[compute].split(compute_i1, factor=4)
-compute_i1_o_o_i, compute_i1_o_i = s[compute].split(compute_i1_o_i, factor=8)
-compute_i1_o_o_o, compute_i1_o_o_i = s[compute].split(compute_i1_o_o_i, factor=1)
-compute_i2_o_i, compute_i2_i = s[compute].split(compute_i2, factor=1)
-compute_i2_o_o_i, compute_i2_o_i = s[compute].split(compute_i2_o_i, factor=7)
+compute_i1_o_i, compute_i1_i = s[compute].split(compute_i1, factor=2)
+compute_i1_o_o_i, compute_i1_o_i = s[compute].split(compute_i1_o_i, factor=16)
+compute_i1_o_o_o, compute_i1_o_o_i = s[compute].split(compute_i1_o_o_i, factor=2)
+compute_i2_o_i, compute_i2_i = s[compute].split(compute_i2, factor=7)
+compute_i2_o_o_i, compute_i2_o_i = s[compute].split(compute_i2_o_i, factor=1)
compute_i2_o_o_o, compute_i2_o_o_i = s[compute].split(compute_i2_o_o_i, factor=1)
compute_i3_o_i, compute_i3_i = s[compute].split(compute_i3, factor=1)
-compute_i3_o_o_i, compute_i3_o_i = s[compute].split(compute_i3_o_i, factor=1)
+compute_i3_o_o_i, compute_i3_o_i = s[compute].split(compute_i3_o_i, factor=7)
compute_i3_o_o_o, compute_i3_o_o_i = s[compute].split(compute_i3_o_o_i, factor=1)
s[compute].reorder(compute_i0_o_o_o, compute_i1_o_o_o, compute_i2_o_o_o, compute_i3_o_o_o, compute_i0_o_o_i, compute_i1_o_o_i, compute_i2_o_o_i, compute_i3_o_o_i, compute_i0_o_i, compute_i1_o_i, compute_i2_o_i, compute_i3_o_i, compute_i0_i, compute_i1_i, compute_i2_i, compute_i3_i)
s[conv2d_nchw].compute_at(s[compute], compute_i3_o_i)
@@ -653,14 +1742,14 @@ s[compute].bind(compute_i0_o_i_i1_o_i_fused_i2_o_i_fused_i3_o_i_fused, te.thread
kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused = s[kernel_shared].fuse(kernel_shared_ax0, kernel_shared_ax1, kernel_shared_ax2, kernel_shared_ax3)
kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_i = s[kernel_shared].split(kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused, factor=1)
s[kernel_shared].vectorize(kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_i)
-kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_o, kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i = s[kernel_shared].split(kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, factor=56)
+kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_o, kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i = s[kernel_shared].split(kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, factor=112)
s[kernel_shared].bind(kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i, te.thread_axis("threadIdx.x"))
pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused = s[pad_temp_shared].fuse(pad_temp_shared_ax0, pad_temp_shared_ax1, pad_temp_shared_ax2, pad_temp_shared_ax3)
pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_i = s[pad_temp_shared].split(pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused, factor=1)
s[pad_temp_shared].vectorize(pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_i)
-pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_o, pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i = s[pad_temp_shared].split(pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, factor=56)
+pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_o, pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i = s[pad_temp_shared].split(pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, factor=112)
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", 64)
+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:
@@ -678,56 +1767,1109 @@ CUDA source code:
#define int64_t long long
#define uint64_t unsigned long long
#endif
-extern "C" __global__ void __launch_bounds__(56) default_function_kernel0(float* __restrict__ data, float* __restrict__ kernel, float* __restrict__ compute, float* __restrict__ bias) {
- float conv2d_nchw[4];
- __shared__ float pad_temp_shared[72];
- __shared__ float kernel_shared[768];
+extern "C" __global__ void __launch_bounds__(112) default_function_kernel0(float* __restrict__ data, float* __restrict__ kernel, float* __restrict__ compute, float* __restrict__ bias) {
+ float conv2d_nchw[28];
+ __shared__ float pad_temp_shared[648];
+ __shared__ float kernel_shared[4608];
conv2d_nchw[0] = 0.000000e+00f;
+ conv2d_nchw[14] = 0.000000e+00f;
conv2d_nchw[1] = 0.000000e+00f;
+ conv2d_nchw[15] = 0.000000e+00f;
conv2d_nchw[2] = 0.000000e+00f;
+ conv2d_nchw[16] = 0.000000e+00f;
conv2d_nchw[3] = 0.000000e+00f;
+ conv2d_nchw[17] = 0.000000e+00f;
+ conv2d_nchw[4] = 0.000000e+00f;
+ conv2d_nchw[18] = 0.000000e+00f;
+ conv2d_nchw[5] = 0.000000e+00f;
+ conv2d_nchw[19] = 0.000000e+00f;
+ conv2d_nchw[6] = 0.000000e+00f;
+ conv2d_nchw[20] = 0.000000e+00f;
+ conv2d_nchw[7] = 0.000000e+00f;
+ conv2d_nchw[21] = 0.000000e+00f;
+ conv2d_nchw[8] = 0.000000e+00f;
+ conv2d_nchw[22] = 0.000000e+00f;
+ conv2d_nchw[9] = 0.000000e+00f;
+ conv2d_nchw[23] = 0.000000e+00f;
+ conv2d_nchw[10] = 0.000000e+00f;
+ conv2d_nchw[24] = 0.000000e+00f;
+ conv2d_nchw[11] = 0.000000e+00f;
+ conv2d_nchw[25] = 0.000000e+00f;
+ conv2d_nchw[12] = 0.000000e+00f;
+ conv2d_nchw[26] = 0.000000e+00f;
+ conv2d_nchw[13] = 0.000000e+00f;
+ conv2d_nchw[27] = 0.000000e+00f;
for (int rc_outer_outer = 0; rc_outer_outer < 64; ++rc_outer_outer) {
- for (int rx_outer_outer = 0; rx_outer_outer < 3; ++rx_outer_outer) {
- __syncthreads();
- pad_temp_shared[((int)threadIdx.x)] = (((((1 <= (((int)threadIdx.x) % 9)) && ((((int)threadIdx.x) % 9) < 8)) && (1 <= (rx_outer_outer + (((int)blockIdx.x) % 7)))) && ((rx_outer_outer + (((int)blockIdx.x) % 7)) < 8)) ? data[((((((rc_outer_outer * 392) + ((((int)threadIdx.x) / 9) * 49)) + ((((int)threadIdx.x) % 9) * 7)) + rx_outer_outer) + (((int)blockIdx.x) % 7)) - 8)] : 0.000000e+00f);
- if (((int)threadIdx.x) < 16) {
- pad_temp_shared[(((int)threadIdx.x) + 56)] = (((((1 <= ((((int)threadIdx.x) + 2) % 9)) && (((((int)threadIdx.x) + 2) % 9) < 8)) && (1 <= (rx_outer_outer + (((int)blockIdx.x) % 7)))) && ((rx_outer_outer + (((int)blockIdx.x) % 7)) < 8)) ? data[((((((rc_outer_outer * 392) + (((((int)threadIdx.x) + 56) / 9) * 49)) + (((((int)threadIdx.x) + 2) % 9) * 7)) + rx_outer_outer) + (((int)blockIdx.x) % 7)) - 8)] : 0.000000e+00f);
- }
- kernel_shared[((int)threadIdx.x)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + ((((int)threadIdx.x) / 24) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) % 24) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 56)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 56) / 24) * 4608)) + (rc_outer_outer * 72)) + (((((int)threadIdx.x) + 8) % 24) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 112)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 112) / 24) * 4608)) + (rc_outer_outer * 72)) + (((((int)threadIdx.x) + 16) % 24) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 168)] = kernel[(((((((((int)blockIdx.x) / 7) * 147456) + ((((int)threadIdx.x) / 24) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) % 24) * 3)) + rx_outer_outer) + 32256)];
- kernel_shared[(((int)threadIdx.x) + 224)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 224) / 24) * 4608)) + (rc_outer_outer * 72)) + (((((int)threadIdx.x) + 8) % 24) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 280)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 280) / 24) * 4608)) + (rc_outer_outer * 72)) + (((((int)threadIdx.x) + 16) % 24) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 336)] = kernel[(((((((((int)blockIdx.x) / 7) * 147456) + ((((int)threadIdx.x) / 24) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) % 24) * 3)) + rx_outer_outer) + 64512)];
- kernel_shared[(((int)threadIdx.x) + 392)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 392) / 24) * 4608)) + (rc_outer_outer * 72)) + (((((int)threadIdx.x) + 8) % 24) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 448)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 448) / 24) * 4608)) + (rc_outer_outer * 72)) + (((((int)threadIdx.x) + 16) % 24) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 504)] = kernel[(((((((((int)blockIdx.x) / 7) * 147456) + ((((int)threadIdx.x) / 24) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) % 24) * 3)) + rx_outer_outer) + 96768)];
- kernel_shared[(((int)threadIdx.x) + 560)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 560) / 24) * 4608)) + (rc_outer_outer * 72)) + (((((int)threadIdx.x) + 8) % 24) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 616)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 616) / 24) * 4608)) + (rc_outer_outer * 72)) + (((((int)threadIdx.x) + 16) % 24) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 672)] = kernel[(((((((((int)blockIdx.x) / 7) * 147456) + ((((int)threadIdx.x) / 24) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) % 24) * 3)) + rx_outer_outer) + 129024)];
- if (((int)threadIdx.x) < 40) {
- kernel_shared[(((int)threadIdx.x) + 728)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 728) / 24) * 4608)) + (rc_outer_outer * 72)) + (((((int)threadIdx.x) + 8) % 24) * 3)) + rx_outer_outer)];
- }
- __syncthreads();
- for (int rc_outer_inner = 0; rc_outer_inner < 8; ++rc_outer_inner) {
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((rc_outer_inner * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[(((((int)threadIdx.x) / 7) * 96) + (rc_outer_inner * 3))]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[((((((int)threadIdx.x) / 7) * 96) + (rc_outer_inner * 3)) + 1)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[((((((int)threadIdx.x) / 7) * 96) + (rc_outer_inner * 3)) + 2)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((rc_outer_inner * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[((((((int)threadIdx.x) / 7) * 96) + (rc_outer_inner * 3)) + 24)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[((((((int)threadIdx.x) / 7) * 96) + (rc_outer_inner * 3)) + 25)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[((((((int)threadIdx.x) / 7) * 96) + (rc_outer_inner * 3)) + 26)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((rc_outer_inner * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[((((((int)threadIdx.x) / 7) * 96) + (rc_outer_inner * 3)) + 48)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[((((((int)threadIdx.x) / 7) * 96) + (rc_outer_inner * 3)) + 49)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[((((((int)threadIdx.x) / 7) * 96) + (rc_outer_inner * 3)) + 50)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((rc_outer_inner * 9) + (((int)threadIdx.x) % 7))] * kernel_shared[((((((int)threadIdx.x) / 7) * 96) + (rc_outer_inner * 3)) + 72)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 9) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[((((((int)threadIdx.x) / 7) * 96) + (rc_outer_inner * 3)) + 73)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 9) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[((((((int)threadIdx.x) / 7) * 96) + (rc_outer_inner * 3)) + 74)]));
- }
+ __syncthreads();
+ pad_temp_shared[((int)threadIdx.x)] = (((((9 <= (((int)threadIdx.x) % 81)) && ((((int)threadIdx.x) % 81) < 72)) && (1 <= (((int)threadIdx.x) % 9))) && ((((int)threadIdx.x) % 9) < 8)) ? data[(((((rc_outer_outer * 392) + ((((int)threadIdx.x) / 81) * 49)) + (((((int)threadIdx.x) % 81) / 9) * 7)) + (((int)threadIdx.x) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 112)] = (((((9 <= ((((int)threadIdx.x) + 31) % 81)) && (((((int)threadIdx.x) + 31) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 4) % 9))) && (((((int)threadIdx.x) + 4) % 9) < 8)) ? data[(((((rc_outer_outer * 392) + (((((int)threadIdx.x) + 112) / 81) * 49)) + ((((((int)threadIdx.x) + 31) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 4) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 224)] = (((((9 <= ((((int)threadIdx.x) + 62) % 81)) && (((((int)threadIdx.x) + 62) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 8) % 9))) && (((((int)threadIdx.x) + 8) % 9) < 8)) ? data[(((((rc_outer_outer * 392) + (((((int)threadIdx.x) + 224) / 81) * 49)) + ((((((int)threadIdx.x) + 62) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 8) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 336)] = (((((9 <= ((((int)threadIdx.x) + 12) % 81)) && (((((int)threadIdx.x) + 12) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 3) % 9))) && (((((int)threadIdx.x) + 3) % 9) < 8)) ? data[(((((rc_outer_outer * 392) + (((((int)threadIdx.x) + 336) / 81) * 49)) + ((((((int)threadIdx.x) + 12) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 3) % 9)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 448)] = (((((9 <= ((((int)threadIdx.x) + 43) % 81)) && (((((int)threadIdx.x) + 43) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 7) % 9))) && (((((int)threadIdx.x) + 7) % 9) < 8)) ? data[(((((rc_outer_outer * 392) + (((((int)threadIdx.x) + 448) / 81) * 49)) + ((((((int)threadIdx.x) + 43) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 7) % 9)) - 8)] : 0.000000e+00f);
+ if (((int)threadIdx.x) < 88) {
+ pad_temp_shared[(((int)threadIdx.x) + 560)] = (((((9 <= ((((int)threadIdx.x) + 74) % 81)) && (((((int)threadIdx.x) + 74) % 81) < 72)) && (1 <= ((((int)threadIdx.x) + 2) % 9))) && (((((int)threadIdx.x) + 2) % 9) < 8)) ? data[(((((rc_outer_outer * 392) + (((((int)threadIdx.x) + 560) / 81) * 49)) + ((((((int)threadIdx.x) + 74) % 81) / 9) * 7)) + ((((int)threadIdx.x) + 2) % 9)) - 8)] : 0.000000e+00f);
+ }
+ kernel_shared[((int)threadIdx.x)] = kernel[((((((int)blockIdx.x) * 294912) + ((((int)threadIdx.x) / 72) * 4608)) + (rc_outer_outer * 72)) + (((int)threadIdx.x) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 112)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 112) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 40) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 224)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 224) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 8) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 336)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 336) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 48) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 448)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 448) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 16) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 560)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 560) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 56) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 672)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 672) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 24) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 784)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 784) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 64) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 896)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 896) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 32) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 1008)] = kernel[(((((((int)blockIdx.x) * 294912) + ((((int)threadIdx.x) / 72) * 4608)) + (rc_outer_outer * 72)) + (((int)threadIdx.x) % 72)) + 64512)];
+ kernel_shared[(((int)threadIdx.x) + 1120)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 1120) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 40) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 1232)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 1232) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 8) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 1344)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 1344) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 48) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 1456)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 1456) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 16) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 1568)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 1568) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 56) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 1680)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 1680) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 24) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 1792)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 1792) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 64) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 1904)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 1904) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 32) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 2016)] = kernel[(((((((int)blockIdx.x) * 294912) + ((((int)threadIdx.x) / 72) * 4608)) + (rc_outer_outer * 72)) + (((int)threadIdx.x) % 72)) + 129024)];
+ kernel_shared[(((int)threadIdx.x) + 2128)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 2128) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 40) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 2240)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 2240) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 8) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 2352)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 2352) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 48) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 2464)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 2464) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 16) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 2576)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 2576) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 56) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 2688)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 2688) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 24) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 2800)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 2800) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 64) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 2912)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 2912) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 32) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 3024)] = kernel[(((((((int)blockIdx.x) * 294912) + ((((int)threadIdx.x) / 72) * 4608)) + (rc_outer_outer * 72)) + (((int)threadIdx.x) % 72)) + 193536)];
+ kernel_shared[(((int)threadIdx.x) + 3136)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 3136) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 40) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 3248)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 3248) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 8) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 3360)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 3360) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 48) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 3472)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 3472) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 16) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 3584)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 3584) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 56) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 3696)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 3696) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 24) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 3808)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 3808) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 64) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 3920)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 3920) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 32) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 4032)] = kernel[(((((((int)blockIdx.x) * 294912) + ((((int)threadIdx.x) / 72) * 4608)) + (rc_outer_outer * 72)) + (((int)threadIdx.x) % 72)) + 258048)];
+ kernel_shared[(((int)threadIdx.x) + 4144)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 4144) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 40) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 4256)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 4256) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 8) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 4368)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 4368) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 48) % 72))];
+ kernel_shared[(((int)threadIdx.x) + 4480)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 4480) / 72) * 4608)) + (rc_outer_outer * 72)) + ((((int)threadIdx.x) + 16) % 72))];
+ if (((int)threadIdx.x) < 16) {
+ kernel_shared[(((int)threadIdx.x) + 4592)] = kernel[((((((int)blockIdx.x) * 294912) + (((((int)threadIdx.x) + 4592) / 72) * 4608)) + (rc_outer_outer * 72)) + (((int)threadIdx.x) + 56))];
+ }
+ __syncthreads();
+ for (int rc_outer_inner = 0; rc_outer_inner < 2; ++rc_outer_inner) {
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((rc_outer_inner * 324) + (((int)threadIdx.x) % 7))] * kernel_shared[(((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36))]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[((rc_outer_inner * 324) + (((int)threadIdx.x) % 7))] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2304)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 1)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2305)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2306)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 81)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 9)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 81)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2313)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 82)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 10)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 82)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2314)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 83)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 11)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 83)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2315)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 162)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 18)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 162)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2322)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 163)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 19)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 163)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2323)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 164)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 20)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 164)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2324)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 243)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 27)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 243)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2331)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 244)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 28)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 244)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2332)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 245)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 29)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 245)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2333)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[(((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36))]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2304)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 1)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2305)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2306)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 9)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2313)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 10)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2314)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 11)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2315)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 18)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2322)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 19)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2323)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 20)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2324)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 27)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2331)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 28)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2332)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 29)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2333)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[(((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36))]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2304)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 1)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2305)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2306)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 9)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2313)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 10)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2314)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 11)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2315)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 18)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2322)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 19)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2323)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 20)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2324)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 27)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2331)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 28)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2332)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 29)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2333)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[(((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36))]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2304)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 28)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 1)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 28)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2305)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 29)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 29)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2306)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 9)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2313)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 109)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 10)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 109)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2314)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 110)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 11)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 110)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2315)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 18)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2322)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 19)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2323)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 20)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2324)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 27)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2331)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 271)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 28)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 271)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2332)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 272)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 29)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 272)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2333)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[(((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36))]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2304)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 37)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 1)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 37)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2305)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 38)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 38)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2306)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 9)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2313)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 118)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 10)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 118)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2314)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 119)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 11)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 119)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2315)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 18)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2322)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 199)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 19)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 199)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2323)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 200)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 20)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 200)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2324)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 27)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2331)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 280)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 28)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 280)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2332)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 281)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 29)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 281)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2333)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[(((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36))]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2304)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 46)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 1)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 46)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2305)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 47)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 47)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2306)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 9)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2313)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 10)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2314)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 11)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2315)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 18)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2322)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 208)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 19)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 208)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2323)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 209)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 20)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 209)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2324)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 27)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2331)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 289)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 28)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 289)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2332)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 290)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 29)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 290)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2333)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[(((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36))]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2304)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 1)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2305)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 56)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 56)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2306)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 9)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2313)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 136)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 10)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 136)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2314)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 137)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 11)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 137)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2315)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 18)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2322)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 217)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 19)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 217)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2323)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 218)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 20)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 218)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2324)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 27)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2331)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 298)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 28)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 298)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2332)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 299)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 29)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 299)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2333)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((rc_outer_inner * 324) + (((int)threadIdx.x) % 7))] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 72)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[((rc_outer_inner * 324) + (((int)threadIdx.x) % 7))] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2376)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 73)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 1)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2377)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 74)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 2)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2378)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 81)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 81)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 81)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2385)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 82)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 82)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 82)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2386)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 83)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 83)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 83)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2387)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 162)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 90)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 162)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2394)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 163)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 91)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 163)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2395)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 164)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 92)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 164)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2396)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 243)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 99)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 243)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2403)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 244)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 100)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 244)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2404)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 245)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 101)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 245)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2405)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 72)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2376)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 73)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2377)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 74)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2378)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 81)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2385)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 82)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2386)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 83)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2387)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 90)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2394)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 91)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2395)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 92)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2396)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 99)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2403)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 100)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2404)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 101)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2405)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 72)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2376)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 73)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2377)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 74)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2378)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 81)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2385)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 82)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2386)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 83)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2387)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 90)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2394)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 91)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2395)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 92)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2396)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 99)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2403)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 100)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2404)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 101)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2405)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 72)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2376)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 28)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 73)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 28)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2377)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 29)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 74)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 29)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2378)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 81)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2385)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 109)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 82)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 109)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2386)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 110)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 83)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 110)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2387)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 90)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2394)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 91)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2395)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 92)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2396)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 99)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2403)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 271)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 100)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 271)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2404)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 272)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 101)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 272)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2405)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 72)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2376)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 37)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 73)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 37)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2377)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 38)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 74)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 38)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2378)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 81)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2385)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 118)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 82)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 118)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2386)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 119)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 83)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 119)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2387)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 90)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2394)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 199)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 91)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 199)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2395)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 200)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 92)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 200)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2396)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 99)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2403)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 280)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 100)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 280)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2404)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 281)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 101)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 281)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2405)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 72)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2376)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 46)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 73)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 46)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2377)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 47)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 74)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 47)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2378)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 81)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2385)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 82)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2386)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 83)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2387)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 90)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2394)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 208)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 91)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 208)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2395)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 209)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 92)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 209)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2396)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 99)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2403)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 289)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 100)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 289)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2404)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 290)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 101)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 290)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2405)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 72)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2376)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 73)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2377)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 56)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 74)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 56)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2378)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 81)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2385)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 136)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 82)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 136)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2386)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 137)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 83)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 137)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2387)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 90)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2394)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 217)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 91)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 217)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2395)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 218)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 92)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 218)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2396)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 99)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2403)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 298)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 100)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 298)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2404)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 299)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 101)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 299)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2405)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 3)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2307)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 4)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2308)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 5)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2309)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 12)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2316)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 13)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2317)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 14)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2318)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 21)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2325)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 22)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2326)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 23)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2327)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 30)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2334)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 31)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2335)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 32)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2336)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 3)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2307)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 4)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2308)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 5)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2309)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 12)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2316)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 13)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2317)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 14)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2318)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 21)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2325)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 22)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2326)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 23)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2327)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 30)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2334)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 31)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2335)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 32)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2336)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 3)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2307)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 28)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 4)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 28)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2308)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 29)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 5)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 29)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2309)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 12)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2316)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 109)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 13)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 109)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2317)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 110)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 14)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 110)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2318)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 21)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2325)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 22)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2326)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 23)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2327)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 30)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2334)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 271)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 31)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 271)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2335)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 272)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 32)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 272)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2336)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 3)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2307)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 37)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 4)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 37)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2308)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 38)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 5)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 38)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2309)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 12)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2316)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 118)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 13)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 118)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2317)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 119)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 14)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 119)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2318)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 21)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2325)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 199)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 22)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 199)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2326)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 200)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 23)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 200)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2327)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 30)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2334)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 280)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 31)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 280)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2335)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 281)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 32)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 281)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2336)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 3)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2307)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 46)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 4)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 46)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2308)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 47)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 5)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 47)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2309)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 12)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2316)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 13)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2317)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 14)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2318)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 21)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2325)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 208)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 22)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 208)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2326)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 209)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 23)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 209)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2327)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 30)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2334)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 289)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 31)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 289)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2335)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 290)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 32)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 290)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2336)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 3)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2307)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 4)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2308)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 56)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 5)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 56)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2309)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 12)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2316)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 136)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 13)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 136)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2317)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 137)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 14)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 137)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2318)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 21)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2325)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 217)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 22)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 217)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2326)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 218)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 23)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 218)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2327)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 30)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2334)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 298)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 31)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 298)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2335)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 299)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 32)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 299)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2336)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 3)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2307)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 4)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2308)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 5)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2309)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 144)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 12)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 144)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2316)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 145)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 13)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 145)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2317)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 146)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 14)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 146)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2318)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 225)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 21)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 225)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2325)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 226)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 22)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 226)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2326)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 227)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 23)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 227)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2327)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 306)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 30)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 306)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2334)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 307)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 31)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 307)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2335)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 308)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 32)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 308)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2336)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 75)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 9)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2379)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 76)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 10)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2380)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 77)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 11)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2381)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 84)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 90)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2388)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 85)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 91)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2389)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 86)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 92)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2390)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 93)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 171)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2397)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 94)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 172)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2398)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 95)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 173)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2399)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 102)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2406)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 103)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 253)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2407)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 104)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 254)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2408)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 75)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2379)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 76)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2380)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 77)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2381)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 84)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2388)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 85)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2389)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 86)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2390)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 93)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2397)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 94)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2398)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 95)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2399)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 102)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2406)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 103)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2407)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 104)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2408)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 75)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2379)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 28)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 76)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 28)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2380)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 29)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 77)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 29)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2381)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 84)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2388)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 109)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 85)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 109)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2389)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 110)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 86)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 110)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2390)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 93)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2397)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 94)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2398)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 95)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2399)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 102)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2406)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 271)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 103)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 271)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2407)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 272)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 104)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 272)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2408)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 75)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2379)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 37)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 76)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 37)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2380)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 38)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 77)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 38)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2381)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 84)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2388)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 118)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 85)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 118)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2389)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 119)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 86)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 119)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2390)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 93)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2397)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 199)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 94)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 199)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2398)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 200)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 95)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 200)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2399)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 102)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2406)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 280)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 103)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 280)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2407)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 281)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 104)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 281)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2408)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 75)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2379)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 46)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 76)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 46)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2380)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 47)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 77)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 47)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2381)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 84)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2388)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 85)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2389)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 86)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2390)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 93)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2397)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 208)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 94)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 208)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2398)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 209)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 95)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 209)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2399)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 102)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2406)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 289)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 103)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 289)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2407)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 290)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 104)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 290)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2408)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 75)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2379)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 76)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2380)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 56)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 77)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 56)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2381)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 84)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2388)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 136)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 85)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 136)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2389)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 137)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 86)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 137)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2390)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 93)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2397)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 217)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 94)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 217)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2398)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 218)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 95)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 218)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2399)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 102)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2406)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 298)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 103)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 298)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2407)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 299)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 104)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 299)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2408)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 75)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2379)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 76)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2380)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 77)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2381)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 144)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 84)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 144)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2388)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 145)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 85)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 145)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2389)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 146)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 86)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 146)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2390)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 225)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 93)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 225)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2397)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 226)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 94)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 226)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2398)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 227)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 95)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 227)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2399)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 306)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 102)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 306)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2406)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 307)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 103)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 307)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2407)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 308)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 104)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 308)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2408)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 6)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2310)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 7)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2311)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 8)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2312)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 15)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2319)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 16)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2320)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 17)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2321)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 24)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2328)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 25)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2329)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 26)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2330)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 33)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2337)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 34)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2338)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 35)]));
+ conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2339)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 6)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2310)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 28)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 7)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 28)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2311)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 29)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 8)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 29)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2312)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 15)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2319)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 109)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 16)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 109)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2320)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 110)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 17)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 110)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2321)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 24)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2328)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 25)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2329)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 26)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2330)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 33)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2337)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 271)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 34)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 271)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2338)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 272)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 35)]));
+ conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 272)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2339)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 6)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2310)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 37)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 7)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 37)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2311)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 38)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 8)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 38)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2312)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 15)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2319)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 118)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 16)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 118)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2320)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 119)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 17)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 119)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2321)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 24)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2328)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 199)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 25)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 199)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2329)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 200)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 26)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 200)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2330)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 33)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2337)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 280)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 34)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 280)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2338)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 281)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 35)]));
+ conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 281)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2339)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 6)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2310)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 46)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 7)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 46)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2311)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 47)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 8)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 47)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2312)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 15)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2319)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 16)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2320)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 17)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2321)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 24)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2328)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 208)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 25)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 208)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2329)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 209)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 26)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 209)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2330)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 33)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2337)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 289)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 34)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 289)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2338)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 290)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 35)]));
+ conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 290)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2339)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 6)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2310)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 7)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2311)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 56)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 8)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 56)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2312)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 15)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2319)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 136)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 16)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 136)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2320)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 137)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 17)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 137)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2321)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 24)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2328)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 217)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 25)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 217)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2329)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 218)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 26)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 218)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2330)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 33)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2337)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 298)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 34)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 298)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2338)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 299)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 35)]));
+ conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 299)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2339)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 6)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2310)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 7)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2311)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 8)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2312)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 144)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 15)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 144)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2319)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 145)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 16)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 145)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2320)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 146)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 17)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 146)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2321)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 225)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 24)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 225)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2328)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 226)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 25)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 226)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2329)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 227)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 26)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 227)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2330)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 306)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 33)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 306)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2337)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 307)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 34)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 307)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2338)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 308)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 35)]));
+ conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 308)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2339)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 72)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 6)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 72)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2310)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 73)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 7)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 73)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2311)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 74)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 8)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 74)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2312)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 153)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 15)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 153)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2319)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 154)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 16)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 154)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2320)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 155)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 17)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 155)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2321)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 234)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 24)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 234)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2328)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 235)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 25)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 235)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2329)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 236)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 26)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 236)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2330)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 33)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2337)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 316)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 34)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 316)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2338)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 317)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 35)]));
+ conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 317)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2339)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 78)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 18)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2382)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 79)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 19)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2383)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 80)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 20)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2384)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 87)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 99)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2391)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 88)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 100)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2392)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 89)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 101)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2393)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 96)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 180)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2400)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 97)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 181)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2401)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 98)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 182)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2402)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 105)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 261)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2409)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 106)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 262)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2410)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 107)]));
+ conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 263)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2411)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 78)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 27)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2382)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 28)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 79)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 28)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2383)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 29)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 80)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 29)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2384)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 87)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 108)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2391)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 109)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 88)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 109)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2392)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 110)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 89)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 110)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2393)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 96)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2400)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 97)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 190)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2401)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 98)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 191)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2402)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 105)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 270)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2409)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 271)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 106)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 271)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2410)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 272)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 107)]));
+ conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 272)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2411)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 78)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 36)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2382)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 37)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 79)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 37)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2383)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 38)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 80)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 38)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2384)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 87)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 117)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2391)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 118)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 88)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 118)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2392)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 119)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 89)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 119)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2393)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 96)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 198)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2400)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 199)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 97)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 199)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2401)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 200)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 98)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 200)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2402)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 105)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 279)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2409)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 280)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 106)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 280)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2410)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 281)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 107)]));
+ conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 281)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2411)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 78)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 45)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2382)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 46)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 79)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 46)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2383)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 47)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 80)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 47)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2384)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 87)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2391)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 88)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 127)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2392)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 89)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 128)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2393)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 96)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 207)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2400)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 208)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 97)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 208)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2401)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 209)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 98)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 209)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2402)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 105)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 288)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2409)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 289)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 106)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 289)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2410)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 290)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 107)]));
+ conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 290)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2411)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 78)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 54)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2382)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 79)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2383)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 56)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 80)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 56)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2384)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 87)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 135)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2391)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 136)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 88)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 136)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2392)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 137)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 89)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 137)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2393)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 96)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 216)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2400)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 217)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 97)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 217)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2401)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 218)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 98)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 218)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2402)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 105)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 297)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2409)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 298)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 106)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 298)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2410)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 299)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 107)]));
+ conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 299)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2411)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 78)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2382)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 79)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 64)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2383)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 80)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 65)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2384)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 144)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 87)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 144)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2391)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 145)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 88)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 145)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2392)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 146)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 89)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 146)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2393)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 225)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 96)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 225)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2400)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 226)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 97)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 226)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2401)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 227)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 98)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 227)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2402)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 306)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 105)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 306)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2409)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 307)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 106)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 307)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2410)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 308)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 107)]));
+ conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 308)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2411)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 72)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 78)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 72)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2382)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 73)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 79)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 73)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2383)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 74)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 80)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 74)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2384)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 153)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 87)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 153)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2391)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 154)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 88)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 154)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2392)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 155)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 89)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 155)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2393)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 234)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 96)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 234)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2400)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 235)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 97)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 235)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2401)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 236)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 98)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 236)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2402)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 105)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 315)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2409)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 316)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 106)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 316)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2410)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 317)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 107)]));
+ conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((rc_outer_inner * 324) + (((int)threadIdx.x) % 7)) + 317)] * kernel_shared[((((((int)threadIdx.x) / 7) * 144) + (rc_outer_inner * 36)) + 2411)]));
}
}
- for (int i1_inner = 0; i1_inner < 4; ++i1_inner) {
- compute[((((((((int)blockIdx.x) / 7) * 1568) + ((((int)threadIdx.x) / 7) * 196)) + (i1_inner * 49)) + ((((int)threadIdx.x) % 7) * 7)) + (((int)blockIdx.x) % 7))] = max((conv2d_nchw[i1_inner] + bias[((((((int)blockIdx.x) / 7) * 32) + ((((int)threadIdx.x) / 7) * 4)) + i1_inner)]), 0.000000e+00f);
+ for (int i1_inner = 0; i1_inner < 2; ++i1_inner) {
+ for (int i2_inner = 0; i2_inner < 7; ++i2_inner) {
+ compute[(((((((int)blockIdx.x) * 3136) + ((((int)threadIdx.x) / 7) * 98)) + (i1_inner * 49)) + (i2_inner * 7)) + (((int)threadIdx.x) % 7))] = max((conv2d_nchw[((i1_inner * 7) + i2_inner)] + bias[(((((int)blockIdx.x) * 64) + ((((int)threadIdx.x) / 7) * 2)) + i1_inner)]), 0.000000e+00f);
+ compute[((((((((int)blockIdx.x) * 3136) + ((((int)threadIdx.x) / 7) * 98)) + (i1_inner * 49)) + (i2_inner * 7)) + (((int)threadIdx.x) % 7)) + 1568)] = max((conv2d_nchw[(((i1_inner * 7) + i2_inner) + 14)] + bias[((((((int)blockIdx.x) * 64) + ((((int)threadIdx.x) / 7) * 2)) + i1_inner) + 32)]), 0.000000e+00f);
+ }
}
}
</pre></div>
@@ -765,7 +2907,7 @@ 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 21.084 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 2 minutes 21.977 seconds)</p>
<div class="sphx-glr-footer class 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 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 301610d..282e5dd 100644
--- a/docs/how_to/tune_with_autoscheduler/tune_network_cuda.html
+++ b/docs/how_to/tune_with_autoscheduler/tune_network_cuda.html
@@ -876,7 +876,7 @@ 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)
- 9.6230 9.6119 9.6561 9.6010 0.0238
+ 9.6781 9.7014 9.7153 9.6174 0.0433
</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 7b7ab46..57ea731 100644
--- a/docs/how_to/tune_with_autoscheduler/tune_network_x86.html
+++ b/docs/how_to/tune_with_autoscheduler/tune_network_x86.html
@@ -895,7 +895,7 @@ 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)
- 763.7144 765.4548 767.6819 758.0064 4.1373
+ 759.9460 761.3004 763.9244 754.6133 3.9200
</pre></div>
</div>
</div>
@@ -917,7 +917,7 @@ 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 19.840 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 19.340 seconds)</p>
<div class="sphx-glr-footer class sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-tune-with-autoscheduler-tune-network-x86-py">
<div class="sphx-glr-download 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 db9e782..adc576b 100644
--- a/docs/how_to/tune_with_autoscheduler/tune_sparse_x86.html
+++ b/docs/how_to/tune_with_autoscheduler/tune_sparse_x86.html
@@ -600,73 +600,120 @@ 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} {
- for (i0.outer.i1.outer.fused: int32, 0, 128) "parallel" {
- allocate(compute_3: Pointer(global float32), float32, [512]), storage_scope = global {
- for (i.inner.init: int32, 0, 32) {
- let cse_var_1: int32 = (i.inner.init*16)
- {
- compute_4: Buffer(compute_3, float32, [512], [])[cse_var_1] = 0f32
- compute_4[(cse_var_1 + 1)] = 0f32
- compute_4[(cse_var_1 + 2)] = 0f32
- compute_4[(cse_var_1 + 3)] = 0f32
- compute_4[(cse_var_1 + 4)] = 0f32
- compute_4[(cse_var_1 + 5)] = 0f32
- compute_4[(cse_var_1 + 6)] = 0f32
- compute_4[(cse_var_1 + 7)] = 0f32
- compute_4[(cse_var_1 + 8)] = 0f32
- compute_4[(cse_var_1 + 9)] = 0f32
- compute_4[(cse_var_1 + 10)] = 0f32
- compute_4[(cse_var_1 + 11)] = 0f32
- compute_4[(cse_var_1 + 12)] = 0f32
- compute_4[(cse_var_1 + 13)] = 0f32
- compute_4[(cse_var_1 + 14)] = 0f32
- compute_4[(cse_var_1 + 15)] = 0f32
- }
- }
- for (elem_idx: int32, 0, let cse_var_2: int32 = floormod(i0.outer.i1.outer.fused, 32) in (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])) {
- for (i.inner: int32, 0, 32) {
- let cse_var_21: int32 = floormod(i0.outer.i1.outer.fused, 32)
- let cse_var_20: int32 = (i.inner*16)
- let cse_var_19: int32 = (elem_idx*16)
- let cse_var_18: int32 = (cse_var_20 + 10)
- let cse_var_17: int32 = (cse_var_20 + 11)
- let cse_var_16: int32 = (cse_var_20 + 12)
- let cse_var_15: int32 = (cse_var_20 + 13)
- let cse_var_14: int32 = (cse_var_20 + 14)
- let cse_var_13: int32 = (cse_var_20 + 15)
- let cse_var_12: int32 = (cse_var_20 + 2)
- let cse_var_11: int32 = (cse_var_20 + 3)
- let cse_var_10: int32 = (cse_var_20 + 4)
- let cse_var_9: int32 = (cse_var_20 + 5)
- let cse_var_8: int32 = (cse_var_20 + 6)
- let cse_var_7: int32 = (cse_var_20 + 7)
- let cse_var_6: int32 = (cse_var_20 + 8)
- let cse_var_5: int32 = (cse_var_20 + 9)
- let cse_var_4: int32 = (cse_var_20 + 1)
- let cse_var_3: int32 = ((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.inner*256))
+ for (i0.outer.i1.outer.fused: int32, 0, 32) "parallel" {
+ allocate(compute_3: Pointer(global float32), float32, [2048]), storage_scope = global {
+ for (i.outer.inner: int32, 0, 32) {
+ for (nb_j.inner: int32, 0, 2) {
+ let cse_var_2: int32 = ((i.outer.inner*64) + (nb_j.inner*16))
+ let cse_var_1: int32 = ((floormod(i0.outer.i1.outer.fused, 16)*2) + nb_j.inner)
{
- compute_4[cse_var_20] = (compute_4[cse_var_20] + (placeholder_1[((placeholder_3[cse_var_21]*16) + cse_var_19)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_21] + elem_idx)])], 0f32)))
- compute_4[cse_var_4] = (compute_4[cse_var_4] + (placeholder_1[(((placeholder_3[cse_var_21]*16) + cse_var_19) + 1)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_21] + elem_idx)])], 0f32)))
- compute_4[cse_var_12] = (compute_4[cse_var_12] + (placeholder_1[(((placeholder_3[cse_var_21]*16) + cse_var_19) + 2)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_21] + elem_idx)])], 0f32)))
- compute_4[cse_var_11] = (compute_4[cse_var_11] + (placeholder_1[(((placeholder_3[cse_var_21]*16) + cse_var_19) + 3)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_21] + elem_idx)])], 0f32)))
- compute_4[cse_var_10] = (compute_4[cse_var_10] + (placeholder_1[(((placeholder_3[cse_var_21]*16) + cse_var_19) + 4)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_21] + elem_idx)])], 0f32)))
- compute_4[cse_var_9] = (compute_4[cse_var_9] + (placeholder_1[(((placeholder_3[cse_var_21]*16) + cse_var_19) + 5)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_21] + elem_idx)])], 0f32)))
- compute_4[cse_var_8] = (compute_4[cse_var_8] + (placeholder_1[(((placeholder_3[cse_var_21]*16) + cse_var_19) + 6)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_21] + elem_idx)])], 0f32)))
- compute_4[cse_var_7] = (compute_4[cse_var_7] + (placeholder_1[(((placeholder_3[cse_var_21]*16) + cse_var_19) + 7)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_21] + elem_idx)])], 0f32)))
- compute_4[cse_var_6] = (compute_4[cse_var_6] + (placeholder_1[(((placeholder_3[cse_var_21]*16) + cse_var_19) + 8)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_21] + elem_idx)])], 0f32)))
- compute_4[cse_var_5] = (compute_4[cse_var_5] + (placeholder_1[(((placeholder_3[cse_var_21]*16) + cse_var_19) + 9)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_21] + elem_idx)])], 0f32)))
- compute_4[cse_var_18] = (compute_4[cse_var_18] + (placeholder_1[(((placeholder_3[cse_var_21]*16) + cse_var_19) + 10)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_21] + elem_idx)])], 0f32)))
- compute_4[cse_var_17] = (compute_4[cse_var_17] + (placeholder_1[(((placeholder_3[cse_var_21]*16) + cse_var_19) + 11)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_21] + elem_idx)])], 0f32)))
- compute_4[cse_var_16] = (compute_4[cse_var_16] + (placeholder_1[(((placeholder_3[cse_var_21]*16) + cse_var_19) + 12)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_21] + elem_idx)])], 0f32)))
- compute_4[cse_var_15] = (compute_4[cse_var_15] + (placeholder_1[(((placeholder_3[cse_var_21]*16) + cse_var_19) + 13)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_21] + elem_idx)])], 0f32)))
- compute_4[cse_var_14] = (compute_4[cse_var_14] + (placeholder_1[(((placeholder_3[cse_var_21]*16) + cse_var_19) + 14)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_21] + elem_idx)])], 0f32)))
- compute_4[cse_var_13] = (compute_4[cse_var_13] + (placeholder_1[(((placeholder_3[cse_var_21]*16) + cse_var_19) + 15)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_21] + elem_idx)])], 0f32)))
+ compute_4: Buffer(compute_3, float32, [2048], [])[cse_var_2] = 0f32
+ compute_4[(cse_var_2 + 1)] = 0f32
+ compute_4[(cse_var_2 + 2)] = 0f32
+ compute_4[(cse_var_2 + 3)] = 0f32
+ compute_4[(cse_var_2 + 4)] = 0f32
+ compute_4[(cse_var_2 + 5)] = 0f32
+ compute_4[(cse_var_2 + 6)] = 0f32
+ compute_4[(cse_var_2 + 7)] = 0f32
+ compute_4[(cse_var_2 + 8)] = 0f32
+ compute_4[(cse_var_2 + 9)] = 0f32
+ compute_4[(cse_var_2 + 10)] = 0f32
+ compute_4[(cse_var_2 + 11)] = 0f32
+ compute_4[(cse_var_2 + 12)] = 0f32
+ compute_4[(cse_var_2 + 13)] = 0f32
+ compute_4[(cse_var_2 + 14)] = 0f32
+ compute_4[(cse_var_2 + 15)] = 0f32
+ compute_4[(cse_var_2 + 32)] = 0f32
+ compute_4[(cse_var_2 + 33)] = 0f32
+ compute_4[(cse_var_2 + 34)] = 0f32
+ compute_4[(cse_var_2 + 35)] = 0f32
+ compute_4[(cse_var_2 + 36)] = 0f32
+ compute_4[(cse_var_2 + 37)] = 0f32
+ compute_4[(cse_var_2 + 38)] = 0f32
+ compute_4[(cse_var_2 + 39)] = 0f32
+ compute_4[(cse_var_2 + 40)] = 0f32
+ compute_4[(cse_var_2 + 41)] = 0f32
+ compute_4[(cse_var_2 + 42)] = 0f32
+ compute_4[(cse_var_2 + 43)] = 0f32
+ compute_4[(cse_var_2 + 44)] = 0f32
+ compute_4[(cse_var_2 + 45)] = 0f32
+ compute_4[(cse_var_2 + 46)] = 0f32
+ compute_4[(cse_var_2 + 47)] = 0f32
+ for (elem_idx: int32, 0, (placeholder_3[(cse_var_1 + 1)] - placeholder_3[cse_var_1])) {
+ let cse_var_35: int32 = (cse_var_2 + 1)
+ let cse_var_34: int32 = (cse_var_2 + 10)
+ let cse_var_33: int32 = (cse_var_2 + 11)
+ let cse_var_32: int32 = (cse_var_2 + 12)
+ let cse_var_31: int32 = (cse_var_2 + 13)
+ let cse_var_30: int32 = (cse_var_2 + 14)
+ let cse_var_29: int32 = (cse_var_2 + 15)
+ let cse_var_28: int32 = (cse_var_2 + 2)
+ let cse_var_27: int32 = (cse_var_2 + 3)
+ let cse_var_26: int32 = (cse_var_2 + 32)
+ let cse_var_25: int32 = (cse_var_2 + 33)
+ let cse_var_24: int32 = (cse_var_2 + 34)
+ let cse_var_23: int32 = (cse_var_2 + 35)
+ let cse_var_22: int32 = (cse_var_2 + 36)
+ let cse_var_21: int32 = (cse_var_2 + 37)
+ let cse_var_20: int32 = (cse_var_2 + 39)
+ let cse_var_19: int32 = (elem_idx*16)
+ let cse_var_18: int32 = (cse_var_2 + 9)
+ let cse_var_17: int32 = (cse_var_2 + 8)
+ let cse_var_16: int32 = (cse_var_2 + 7)
+ let cse_var_15: int32 = (cse_var_2 + 6)
+ let cse_var_14: int32 = (cse_var_2 + 5)
+ let cse_var_13: int32 = (cse_var_2 + 47)
+ let cse_var_12: int32 = (cse_var_2 + 38)
+ let cse_var_11: int32 = (cse_var_2 + 45)
+ let cse_var_10: int32 = (cse_var_2 + 44)
+ let cse_var_9: int32 = (cse_var_2 + 43)
+ let cse_var_8: int32 = (cse_var_2 + 42)
+ let cse_var_7: int32 = (cse_var_2 + 41)
+ let cse_var_6: int32 = (cse_var_2 + 40)
+ let cse_var_5: int32 = (cse_var_2 + 4)
+ let cse_var_4: int32 = (cse_var_2 + 46)
+ let cse_var_3: int32 = ((floordiv(i0.outer.i1.outer.fused, 16)*16384) + (i.outer.inner*512))
+ {
+ compute_4[cse_var_2] = (compute_4[cse_var_2] + (placeholder_1[((placeholder_3[cse_var_1]*16) + cse_var_19)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)])], 0f32)))
+ compute_4[cse_var_35] = (compute_4[cse_var_35] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 1)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)])], 0f32)))
+ compute_4[cse_var_28] = (compute_4[cse_var_28] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 2)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)])], 0f32)))
+ compute_4[cse_var_27] = (compute_4[cse_var_27] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 3)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)])], 0f32)))
+ compute_4[cse_var_5] = (compute_4[cse_var_5] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 4)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)])], 0f32)))
+ compute_4[cse_var_14] = (compute_4[cse_var_14] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 5)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)])], 0f32)))
+ compute_4[cse_var_15] = (compute_4[cse_var_15] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 6)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)])], 0f32)))
+ compute_4[cse_var_16] = (compute_4[cse_var_16] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 7)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)])], 0f32)))
+ compute_4[cse_var_17] = (compute_4[cse_var_17] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 8)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)])], 0f32)))
+ compute_4[cse_var_18] = (compute_4[cse_var_18] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 9)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)])], 0f32)))
+ compute_4[cse_var_34] = (compute_4[cse_var_34] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 10)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)])], 0f32)))
+ compute_4[cse_var_33] = (compute_4[cse_var_33] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 11)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)])], 0f32)))
+ compute_4[cse_var_32] = (compute_4[cse_var_32] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 12)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)])], 0f32)))
+ compute_4[cse_var_31] = (compute_4[cse_var_31] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 13)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)])], 0f32)))
+ compute_4[cse_var_30] = (compute_4[cse_var_30] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 14)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)])], 0f32)))
+ compute_4[cse_var_29] = (compute_4[cse_var_29] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 15)]*max(placeholder[(cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)])], 0f32)))
+ compute_4[cse_var_26] = (compute_4[cse_var_26] + (placeholder_1[((placeholder_3[cse_var_1]*16) + cse_var_19)]*max(placeholder[((cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)]) + 256)], 0f32)))
+ compute_4[cse_var_25] = (compute_4[cse_var_25] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 1)]*max(placeholder[((cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)]) + 256)], 0f32)))
+ compute_4[cse_var_24] = (compute_4[cse_var_24] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 2)]*max(placeholder[((cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)]) + 256)], 0f32)))
+ compute_4[cse_var_23] = (compute_4[cse_var_23] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 3)]*max(placeholder[((cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)]) + 256)], 0f32)))
+ compute_4[cse_var_22] = (compute_4[cse_var_22] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 4)]*max(placeholder[((cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)]) + 256)], 0f32)))
+ compute_4[cse_var_21] = (compute_4[cse_var_21] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 5)]*max(placeholder[((cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)]) + 256)], 0f32)))
+ compute_4[cse_var_12] = (compute_4[cse_var_12] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 6)]*max(placeholder[((cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)]) + 256)], 0f32)))
+ compute_4[cse_var_20] = (compute_4[cse_var_20] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 7)]*max(placeholder[((cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)]) + 256)], 0f32)))
+ compute_4[cse_var_6] = (compute_4[cse_var_6] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 8)]*max(placeholder[((cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)]) + 256)], 0f32)))
+ compute_4[cse_var_7] = (compute_4[cse_var_7] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 9)]*max(placeholder[((cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)]) + 256)], 0f32)))
+ compute_4[cse_var_8] = (compute_4[cse_var_8] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 10)]*max(placeholder[((cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)]) + 256)], 0f32)))
+ compute_4[cse_var_9] = (compute_4[cse_var_9] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 11)]*max(placeholder[((cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)]) + 256)], 0f32)))
+ compute_4[cse_var_10] = (compute_4[cse_var_10] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 12)]*max(placeholder[((cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)]) + 256)], 0f32)))
+ compute_4[cse_var_11] = (compute_4[cse_var_11] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 13)]*max(placeholder[((cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)]) + 256)], 0f32)))
+ compute_4[cse_var_4] = (compute_4[cse_var_4] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 14)]*max(placeholder[((cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)]) + 256)], 0f32)))
+ compute_4[cse_var_13] = (compute_4[cse_var_13] + (placeholder_1[(((placeholder_3[cse_var_1]*16) + cse_var_19) + 15)]*max(placeholder[((cse_var_3 + placeholder_2[(placeholder_3[cse_var_1] + elem_idx)]) + 256)], 0f32)))
+ }
+ }
}
}
}
- for (i0.inner: int32, 0, 32) {
- let cse_var_22: int32 = (((floordiv(i0.outer.i1.outer.fused, 32)*16384) + (i0.inner*512)) + (floormod(i0.outer.i1.outer.fused, 32)*16))
- compute[ramp(cse_var_22, 1, 16)] = max((compute_4[ramp((i0.inner*16), 1, 16)] + placeholder_4[ramp(cse_var_22, 1, 16)]), broadcast(0f32, 16))
+ for (i0.inner: int32, 0, 64) {
+ let cse_var_36: int32 = (((floordiv(i0.outer.i1.outer.fused, 16)*32768) + (i0.inner*512)) + (floormod(i0.outer.i1.outer.fused, 16)*32))
+ compute[ramp(cse_var_36, 1, 32)] = max((compute_4[ramp((i0.inner*32), 1, 32)] + placeholder_4[ramp(cse_var_36, 1, 32)]), broadcast(0f32, 32))
}
}
}
@@ -705,7 +752,7 @@ layout transformation, parallelization, vectorization, unrolling, and operator f
</pre></div>
</div>
<p class="sphx-glr-script-out">Out:</p>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Execution time of this operator: 1.733 ms
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Execution time of this operator: 3.501 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 e619500..317a870 100644
--- a/docs/how_to/tune_with_autotvm/sg_execution_times.html
+++ b/docs/how_to/tune_with_autotvm/sg_execution_times.html
@@ -300,13 +300,13 @@
<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>00:44.405</strong> total execution time for <strong>how_to_tune_with_autotvm</strong> files:</p>
+<p><strong>00:44.528</strong> total execution time for <strong>how_to_tune_with_autotvm</strong> files:</p>
<ul class="simple">
-<li><p><strong>00:43.515</strong>: <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></li>
-<li><p><strong>00:00.232</strong>: <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></li>
-<li><p><strong>00:00.223</strong>: <a class="reference internal" href="tune_relay_mobile_gpu.html#sphx-glr-how-to-tune-with-autotvm-tune-relay-mobile-gpu-py"><span class="std std-ref">Auto-tuning a Convolutional Network for Mobile GPU</span></a> (<code class="docutils literal notranslate"><span class="pre">tune_relay_mobile_gpu.py</span></code>)</p></li>
-<li><p><strong>00:00.219</strong>: <a class="reference internal" href="tune_relay_arm.html#sphx-glr-how-to-tune-with-autotvm-tune-relay-arm-py"><span class="std std-ref">Auto-tuning a Convolutional Network for ARM CPU</span></a> (<code class="docutils literal notranslate"><span class="pre">tune_relay_arm.py</span></code>)</p></li>
-<li><p><strong>00:00.216</strong>: <a class="reference internal" href="tune_relay_cuda.html#sphx-glr-how-to-tune-with-autotvm-tune-relay-cuda-py"><span class="std std-ref">Auto-tuning a Convolutional Network for NVIDIA GPU</span></a> (<code class="docutils literal notranslate"><span class="pre">tune_relay_cuda.py</span></code>)</p></li>
+<li><p><strong>00:43.676</strong>: <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></li>
+<li><p><strong>00:00.222</strong>: <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></li>
+<li><p><strong>00:00.215</strong>: <a class="reference internal" href="tune_relay_mobile_gpu.html#sphx-glr-how-to-tune-with-autotvm-tune-relay-mobile-gpu-py"><span class="std std-ref">Auto-tuning a Convolutional Network for Mobile GPU</span></a> (<code class="docutils literal notranslate"><span class="pre">tune_relay_mobile_gpu.py</span></code>)</p></li>
+<li><p><strong>00:00.212</strong>: <a class="reference internal" href="tune_relay_arm.html#sphx-glr-how-to-tune-with-autotvm-tune-relay-arm-py"><span class="std std-ref">Auto-tuning a Convolutional Network for ARM CPU</span></a> (<code class="docutils literal notranslate"><span class="pre">tune_relay_arm.py</span></code>)</p></li>
+<li><p><strong>00:00.202</strong>: <a class="reference internal" href="tune_relay_cuda.html#sphx-glr-how-to-tune-with-autotvm-tune-relay-cuda-py"><span class="std std-ref">Auto-tuning a Convolutional Network for NVIDIA GPU</span></a> (<code class="docutils literal notranslate"><span class="pre">tune_relay_cuda.py</span></code>)</p></li>
</ul>
</div>
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 1f7d149..e89eef8 100644
--- a/docs/how_to/tune_with_autotvm/tune_conv2d_cuda.html
+++ b/docs/how_to/tune_with_autotvm/tune_conv2d_cuda.html
@@ -1142,8 +1142,8 @@ Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 854, 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: 112.57/112.57 result: MeasureResult(costs=(0.002056472038961039,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.8501737117767334, timestamp=1648696103.9259007) [('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: 0.00/112.57 result: Traceback (most recent call last):
+No: 6 GFLOPS: 97.59/97.59 result: MeasureResult(costs=(0.002372302375,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.609654426574707, timestamp=1648705752.068103) [('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: 0.00/97.59 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 571, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 523, in _build_func_common
@@ -1266,7 +1266,7 @@ Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 854, 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: 0.00/112.57 result: Traceback (most recent call last):
+No: 8 GFLOPS: 0.00/97.59 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 571, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 523, in _build_func_common
@@ -1389,7 +1389,7 @@ Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 854, 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: 0.00/112.57 result: Traceback (most recent call last):
+No: 9 GFLOPS: 0.00/97.59 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 571, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 523, in _build_func_common
@@ -1512,7 +1512,7 @@ Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 854, 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: 0.00/112.57 result: Traceback (most recent call last):
+No: 10 GFLOPS: 0.00/97.59 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
@@ -1530,7 +1530,7 @@ No: 10 GFLOPS: 0.00/112.57 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: 0.00/112.57 result: Traceback (most recent call last):
+No: 11 GFLOPS: 0.00/97.59 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 571, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 523, in _build_func_common
@@ -1653,7 +1653,7 @@ Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 854, 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: 0.00/112.57 result: Traceback (most recent call last):
+No: 12 GFLOPS: 0.00/97.59 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 571, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 523, in _build_func_common
@@ -1776,7 +1776,7 @@ Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 854, 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: 0.00/112.57 result: Traceback (most recent call last):
+No: 13 GFLOPS: 0.00/97.59 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 571, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 523, in _build_func_common
@@ -1899,7 +1899,7 @@ Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 854, 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: 0.00/112.57 result: Traceback (most recent call last):
+No: 14 GFLOPS: 0.00/97.59 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 571, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 523, in _build_func_common
@@ -2022,7 +2022,7 @@ Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 854, 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: 0.00/112.57 result: Traceback (most recent call last):
+No: 15 GFLOPS: 0.00/97.59 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 571, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 523, in _build_func_common
@@ -2145,7 +2145,7 @@ Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 854, 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: 0.00/112.57 result: Traceback (most recent call last):
+No: 16 GFLOPS: 0.00/97.59 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 571, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 523, in _build_func_common
@@ -2268,7 +2268,7 @@ Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 854, 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: 0.00/112.57 result: Traceback (most recent call last):
+No: 17 GFLOPS: 0.00/97.59 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 571, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 523, in _build_func_common
@@ -2391,7 +2391,7 @@ Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 854, 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: 0.00/112.57 result: Traceback (most recent call last):
+No: 18 GFLOPS: 0.00/97.59 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 571, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 523, in _build_func_common
@@ -2514,7 +2514,7 @@ Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 854, 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: 0.00/112.57 result: Traceback (most recent call last):
+No: 19 GFLOPS: 0.00/97.59 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 721, in __call__
yield remote, remote.load_module(os.path.split(build_result.filename)[1])
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 685, in run_through_rpc
@@ -2602,7 +2602,7 @@ tvm._ffi.base.TVMError: Traceback (most recent call last):
15: _PyEval_EvalFrameDefault
14: 0x0000000000537c30
13: _PyObject_FastCallKeywords
- 12: 0x00007ff01a5f6fa2
+ 12: 0x00007faf2622dfa2
11: _ctypes_callproc
10: ffi_call
9: ffi_call_unix64
@@ -2667,7 +2667,7 @@ 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: 144.81/144.81 result: MeasureResult(costs=(0.0015986197799999998,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.4213595390319824, timestamp=1648696130.3614411) [('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: 144.71/144.71 result: MeasureResult(costs=(0.00159980322,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.4247119426727295, timestamp=1648705778.465009) [('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,
@@ -2706,7 +2706,7 @@ and measure running time.</p>
<p class="sphx-glr-script-out">Out:</p>
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>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
-Time cost of this operator: 0.002018
+Time cost of this operator: 0.002029
</pre></div>
</div>
<div class="sphx-glr-footer class 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 4f0d190..e293169 100644
--- a/docs/how_to/work_with_microtvm/micro_autotune.html
+++ b/docs/how_to/work_with_microtvm/micro_autotune.html
@@ -553,10 +553,10 @@ the tuned operator.</p>
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>########## Build without Autotuning ##########
Node Name Ops Time(us) Time(%) Shape Inputs Outputs
--------- --- -------- ------- ----- ------ -------
-tvmgen_default_fused_nn_contrib_conv2d_NCHWc tvmgen_default_fused_nn_contrib_conv2d_NCHWc 310.8 98.675 (1, 2, 10, 10, 3) 2 1
-tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 3.228 1.025 (1, 6, 10, 10) 1 1
-tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.944 0.3 (1, 1, 10, 10, 3) 1 1
-Total_time - 314.972 - - - -
+tvmgen_default_fused_nn_contrib_conv2d_NCHWc tvmgen_default_fused_nn_contrib_conv2d_NCHWc 307.3 98.698 (1, 2, 10, 10, 3) 2 1
+tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 3.141 1.009 (1, 6, 10, 10) 1 1
+tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.912 0.293 (1, 1, 10, 10, 3) 1 1
+Total_time - 311.352 - - - -
</pre></div>
</div>
</div>
@@ -608,10 +608,10 @@ Total_time -
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>########## Build with Autotuning ##########
Node Name Ops Time(us) Time(%) Shape Inputs Outputs
--------- --- -------- ------- ----- ------ -------
-tvmgen_default_fused_nn_contrib_conv2d_NCHWc tvmgen_default_fused_nn_contrib_conv2d_NCHWc 197.1 98.678 (1, 6, 10, 10, 1) 2 1
-tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 1.74 0.871 (1, 6, 10, 10) 1 1
-tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.9 0.451 (1, 3, 10, 10, 1) 1 1
-Total_time - 199.74 - - - -
+tvmgen_default_fused_nn_contrib_conv2d_NCHWc tvmgen_default_fused_nn_contrib_conv2d_NCHWc 214.3 98.668 (1, 1, 10, 10, 6) 2 1
+tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 1.944 0.895 (1, 6, 10, 10) 1 1
+tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.948 0.437 (1, 3, 10, 10, 1) 1 1
+Total_time - 217.192 - - - -
</pre></div>
</div>
<div class="sphx-glr-footer class 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/sg_execution_times.html b/docs/how_to/work_with_microtvm/sg_execution_times.html
index a62b70c..315482c 100644
--- a/docs/how_to/work_with_microtvm/sg_execution_times.html
+++ b/docs/how_to/work_with_microtvm/sg_execution_times.html
@@ -300,13 +300,13 @@
<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>00:44.623</strong> total execution time for <strong>how_to_work_with_microtvm</strong> files:</p>
+<p><strong>00:44.518</strong> total execution time for <strong>how_to_work_with_microtvm</strong> files:</p>
<ul class="simple">
-<li><p><strong>00:40.445</strong>: <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></li>
-<li><p><strong>00:03.560</strong>: <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></li>
-<li><p><strong>00:00.218</strong>: <a class="reference internal" href="micro_tvmc.html#sphx-glr-how-to-work-with-microtvm-micro-tvmc-py"><span class="std std-ref">Executing a Tiny Model with TVMC Micro</span></a> (<code class="docutils literal notranslate"><span class="pre">micro_tvmc.py</span></code>)</p></li>
-<li><p><strong>00:00.200</strong>: <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</span></a> (<code class="docutils literal notranslate"><span class="pre">micro_ethosu.py</span></code>)</p></li>
-<li><p><strong>00:00.200</strong>: <a class="reference internal" href="micro_reference_vm.html#sphx-glr-how-to-work-with-microtvm-micro-reference-vm-py"><span class="std std-ref">microTVM Reference Virtual Machines</span></a> (<code class="docutils literal notranslate"><span class="pre">micro_reference_vm.py</span></code>)</p></li>
+<li><p><strong>00:40.235</strong>: <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></li>
+<li><p><strong>00:03.671</strong>: <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></li>
+<li><p><strong>00:00.211</strong>: <a class="reference internal" href="micro_tvmc.html#sphx-glr-how-to-work-with-microtvm-micro-tvmc-py"><span class="std std-ref">Executing a Tiny Model with TVMC Micro</span></a> (<code class="docutils literal notranslate"><span class="pre">micro_tvmc.py</span></code>)</p></li>
+<li><p><strong>00:00.202</strong>: <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</span></a> (<code class="docutils literal notranslate"><span class="pre">micro_ethosu.py</span></code>)</p></li>
+<li><p><strong>00:00.199</strong>: <a class="reference internal" href="micro_reference_vm.html#sphx-glr-how-to-work-with-microtvm-micro-reference-vm-py"><span class="std std-ref">microTVM Reference Virtual Machines</span></a> (<code class="docutils literal notranslate"><span class="pre">micro_reference_vm.py</span></code>)</p></li>
</ul>
</div>
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 f3b9d2b..cad9e69 100644
--- a/docs/how_to/work_with_relay/sg_execution_times.html
+++ b/docs/how_to/work_with_relay/sg_execution_times.html
@@ -300,11 +300,11 @@
<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>00:04.438</strong> total execution time for <strong>how_to_work_with_relay</strong> files:</p>
+<p><strong>00:04.538</strong> total execution time for <strong>how_to_work_with_relay</strong> files:</p>
<ul class="simple">
-<li><p><strong>00:02.526</strong>: <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></li>
-<li><p><strong>00:01.691</strong>: <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></li>
-<li><p><strong>00:00.221</strong>: <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></li>
+<li><p><strong>00:02.532</strong>: <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></li>
+<li><p><strong>00:01.809</strong>: <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></li>
+<li><p><strong>00:00.197</strong>: <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></li>
</ul>
</div>
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 90dfc06..1137de6 100644
--- a/docs/how_to/work_with_schedules/sg_execution_times.html
+++ b/docs/how_to/work_with_schedules/sg_execution_times.html
@@ -300,16 +300,16 @@
<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>00:06.049</strong> total execution time for <strong>how_to_work_with_schedules</strong> files:</p>
+<p><strong>00:06.173</strong> total execution time for <strong>how_to_work_with_schedules</strong> files:</p>
<ul class="simple">
-<li><p><strong>00:02.233</strong>: <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></li>
-<li><p><strong>00:01.223</strong>: <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></li>
-<li><p><strong>00:00.772</strong>: <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></li>
-<li><p><strong>00:00.764</strong>: <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></li>
-<li><p><strong>00:00.332</strong>: <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></li>
-<li><p><strong>00:00.249</strong>: <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></li>
-<li><p><strong>00:00.242</strong>: <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></li>
-<li><p><strong>00:00.233</strong>: <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></li>
+<li><p><strong>00:02.272</strong>: <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></li>
+<li><p><strong>00:01.333</strong>: <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></li>
+<li><p><strong>00:00.771</strong>: <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></li>
+<li><p><strong>00:00.763</strong>: <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></li>
+<li><p><strong>00:00.324</strong>: <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></li>
+<li><p><strong>00:00.239</strong>: <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></li>
+<li><p><strong>00:00.238</strong>: <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></li>
+<li><p><strong>00:00.231</strong>: <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></li>
</ul>
</div>
diff --git a/docs/how_to/work_with_schedules/tensorize.html b/docs/how_to/work_with_schedules/tensorize.html
index 065a219..d6910ca 100644
--- a/docs/how_to/work_with_schedules/tensorize.html
+++ b/docs/how_to/work_with_schedules/tensorize.html
@@ -548,8 +548,8 @@ The importing needs to happen before the tensorized GEMV being executed.</p>
B: Buffer(B_2: Pointer(float32), float32, [32768], []),
C: Buffer(C_2: Pointer(float32), float32, [524288], [])}
buffer_map = {A_1: A, B_1: B, C_1: C} {
- attr [IterVar(i: int32, (nullptr), "DataPar", "")] "pragma_import_llvm" = "; ModuleID = '/tmp/tmp768o3bb6/input0.cc'
-source_filename = "/tmp/tmp768o3bb6/input0.cc"
+ attr [IterVar(i: int32, (nullptr), "DataPar", "")] "pragma_import_llvm" = "; ModuleID = '/tmp/tmpy9m74646/input0.cc'
+source_filename = "/tmp/tmpy9m74646/input0.cc"
target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
target triple = "x86_64-pc-linux-gnu"
diff --git a/docs/reference/api/python/auto_scheduler.html b/docs/reference/api/python/auto_scheduler.html
index 772ebb9..0e65471 100644
--- a/docs/reference/api/python/auto_scheduler.html
+++ b/docs/reference/api/python/auto_scheduler.html
@@ -1713,7 +1713,7 @@ Can be the a function or the function name.</p></li>
<dl class="py function">
<dt class="sig sig-object py" id="tvm.auto_scheduler.auto_schedule">
-<span class="sig-prename descclassname"><span class="pre">tvm.auto_scheduler.</span></span><span class="sig-name descname"><span class="pre">auto_schedule</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">task</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">search_policy</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em>, <em clas [...]
+<span class="sig-prename descclassname"><span class="pre">tvm.auto_scheduler.</span></span><span class="sig-name descname"><span class="pre">auto_schedule</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">task</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">search_policy</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em>, <em clas [...]
<dd><p>THIS API IS DEPRECATED.</p>
<p>Run auto scheduling search for a task.</p>
<dl class="field-list simple">
@@ -1750,7 +1750,7 @@ the initial naive schedule (state).</p>
<dl class="py class">
<dt class="sig sig-object py" id="tvm.auto_scheduler.SketchPolicy">
-<em class="property"><span class="pre">class</span> </em><span class="sig-prename descclassname"><span class="pre">tvm.auto_scheduler.</span></span><span class="sig-name descname"><span class="pre">SketchPolicy</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">task</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">program_cost_model</span></span><span class="o"><span class="pre">=</span></span><span class="defau [...]
+<em class="property"><span class="pre">class</span> </em><span class="sig-prename descclassname"><span class="pre">tvm.auto_scheduler.</span></span><span class="sig-name descname"><span class="pre">SketchPolicy</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">task</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">program_cost_model</span></span><span class="o"><span class="pre">=</span></span><span class="defau [...]
<dd><p>The search policy that searches in a hierarchical search space defined by sketches.
The policy randomly samples programs from the space defined by sketches and use evolutionary
search to fine-tune them.</p>
diff --git a/docs/reference/api/typedoc/classes/bytestreamreader.html b/docs/reference/api/typedoc/classes/bytestreamreader.html
index b796a4e..abbe435 100644
--- a/docs/reference/api/typedoc/classes/bytestreamreader.html
+++ b/docs/reference/api/typedoc/classes/bytestreamreader.html
@@ -119,7 +119,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/rpc_server.ts#L43">rpc_server.ts:43</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/rpc_server.ts#L43">rpc_server.ts:43</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -141,7 +141,7 @@
<div class="tsd-signature tsd-kind-icon">bytes<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">Uint8Array</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/rpc_server.ts#L43">rpc_server.ts:43</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/rpc_server.ts#L43">rpc_server.ts:43</a></li>
</ul>
</aside>
</section>
@@ -151,7 +151,7 @@
<div class="tsd-signature tsd-kind-icon">offset<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">number</span><span class="tsd-signature-symbol"> = 0</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/rpc_server.ts#L42">rpc_server.ts:42</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/rpc_server.ts#L42">rpc_server.ts:42</a></li>
</ul>
</aside>
</section>
@@ -168,7 +168,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/rpc_server.ts#L63">rpc_server.ts:63</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/rpc_server.ts#L63">rpc_server.ts:63</a></li>
</ul>
</aside>
<h4 class="tsd-returns-title">Returns <span class="tsd-signature-type">Uint8Array</span></h4>
@@ -185,7 +185,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/rpc_server.ts#L49">rpc_server.ts:49</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/rpc_server.ts#L49">rpc_server.ts:49</a></li>
</ul>
</aside>
<h4 class="tsd-returns-title">Returns <span class="tsd-signature-type">number</span></h4>
@@ -202,7 +202,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/rpc_server.ts#L57">rpc_server.ts:57</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/rpc_server.ts#L57">rpc_server.ts:57</a></li>
</ul>
</aside>
<h4 class="tsd-returns-title">Returns <span class="tsd-signature-type">number</span></h4>
diff --git a/docs/reference/api/typedoc/classes/cachedcallstack.html b/docs/reference/api/typedoc/classes/cachedcallstack.html
index 2152047..b11813b 100644
--- a/docs/reference/api/typedoc/classes/cachedcallstack.html
+++ b/docs/reference/api/typedoc/classes/cachedcallstack.html
@@ -144,7 +144,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/memory.ts#L223">memory.ts:223</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/memory.ts#L223">memory.ts:223</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -172,7 +172,7 @@
<div class="tsd-signature tsd-kind-icon">temp<wbr>Args<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">Array</span><span class="tsd-signature-symbol"><</span><a href="../interfaces/disposable.html" class="tsd-signature-type">Disposable</a><span class="tsd-signature-symbol">></span><span class="tsd-signature-symbol"> = []</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/memory.ts#L208">memory.ts:208</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/memory.ts#L208">memory.ts:208</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -194,7 +194,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/memory.ts#L312">memory.ts:312</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/memory.ts#L312">memory.ts:312</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -226,7 +226,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/memory.ts#L284">memory.ts:284</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/memory.ts#L284">memory.ts:284</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -262,7 +262,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/memory.ts#L388">memory.ts:388</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/memory.ts#L388">memory.ts:388</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -300,7 +300,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/memory.ts#L376">memory.ts:376</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/memory.ts#L376">memory.ts:376</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -340,7 +340,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/memory.ts#L267">memory.ts:267</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/memory.ts#L267">memory.ts:267</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -373,7 +373,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/memory.ts#L243">memory.ts:243</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/memory.ts#L243">memory.ts:243</a></li>
</ul>
</aside>
<h4 class="tsd-returns-title">Returns <span class="tsd-signature-type">void</span></h4>
@@ -390,7 +390,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/memory.ts#L321">memory.ts:321</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/memory.ts#L321">memory.ts:321</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -422,7 +422,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/memory.ts#L252">memory.ts:252</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/memory.ts#L252">memory.ts:252</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -444,7 +444,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/memory.ts#L359">memory.ts:359</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/memory.ts#L359">memory.ts:359</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -470,7 +470,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/memory.ts#L342">memory.ts:342</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/memory.ts#L342">memory.ts:342</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -496,7 +496,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/memory.ts#L350">memory.ts:350</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/memory.ts#L350">memory.ts:350</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -522,7 +522,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/memory.ts#L326">memory.ts:326</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/memory.ts#L326">memory.ts:326</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -548,7 +548,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/memory.ts#L363">memory.ts:363</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/memory.ts#L363">memory.ts:363</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -574,7 +574,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/memory.ts#L346">memory.ts:346</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/memory.ts#L346">memory.ts:346</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -600,7 +600,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/memory.ts#L334">memory.ts:334</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/memory.ts#L334">memory.ts:334</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
diff --git a/docs/reference/api/typedoc/classes/dldatatype.html b/docs/reference/api/typedoc/classes/dldatatype.html
index e12b69d..d6e477e 100644
--- a/docs/reference/api/typedoc/classes/dldatatype.html
+++ b/docs/reference/api/typedoc/classes/dldatatype.html
@@ -119,7 +119,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L262">runtime.ts:262</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L262">runtime.ts:262</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -147,7 +147,7 @@
<div class="tsd-signature tsd-kind-icon">bits<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">number</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L260">runtime.ts:260</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L260">runtime.ts:260</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -162,7 +162,7 @@
<div class="tsd-signature tsd-kind-icon">code<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">number</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L258">runtime.ts:258</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L258">runtime.ts:258</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -177,7 +177,7 @@
<div class="tsd-signature tsd-kind-icon">lanes<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">number</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L262">runtime.ts:262</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L262">runtime.ts:262</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -199,7 +199,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L279">runtime.ts:279</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L279">runtime.ts:279</a></li>
</ul>
</aside>
<h4 class="tsd-returns-title">Returns <span class="tsd-signature-type">number</span></h4>
@@ -216,7 +216,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L270">runtime.ts:270</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L270">runtime.ts:270</a></li>
</ul>
</aside>
<h4 class="tsd-returns-title">Returns <span class="tsd-signature-type">string</span></h4>
diff --git a/docs/reference/api/typedoc/classes/dldevice.html b/docs/reference/api/typedoc/classes/dldevice.html
index a86506b..2828d24 100644
--- a/docs/reference/api/typedoc/classes/dldevice.html
+++ b/docs/reference/api/typedoc/classes/dldevice.html
@@ -118,7 +118,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L202">runtime.ts:202</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L202">runtime.ts:202</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -146,7 +146,7 @@
<div class="tsd-signature tsd-kind-icon">device<wbr>Id<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">number</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L200">runtime.ts:200</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L200">runtime.ts:200</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -161,7 +161,7 @@
<div class="tsd-signature tsd-kind-icon">device<wbr>Type<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">number</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L198">runtime.ts:198</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L198">runtime.ts:198</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -183,7 +183,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L223">runtime.ts:223</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L223">runtime.ts:223</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -205,7 +205,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L230">runtime.ts:230</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L230">runtime.ts:230</a></li>
</ul>
</aside>
<h4 class="tsd-returns-title">Returns <span class="tsd-signature-type">string</span></h4>
diff --git a/docs/reference/api/typedoc/classes/environment.html b/docs/reference/api/typedoc/classes/environment.html
index 164dc5a..794446e 100644
--- a/docs/reference/api/typedoc/classes/environment.html
+++ b/docs/reference/api/typedoc/classes/environment.html
@@ -125,7 +125,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/environment.ts#L86">environment.ts:86</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/environment.ts#L86">environment.ts:86</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -169,7 +169,7 @@
<aside class="tsd-sources">
<p>Implementation of <a href="../interfaces/libraryprovider.html">LibraryProvider</a>.<a href="../interfaces/libraryprovider.html#imports">imports</a></p>
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/environment.ts#L70">environment.ts:70</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/environment.ts#L70">environment.ts:70</a></li>
</ul>
</aside>
</section>
@@ -179,7 +179,7 @@
<div class="tsd-signature tsd-kind-icon">logger<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol">(</span>msg<span class="tsd-signature-symbol">: </span><span class="tsd-signature-type">string</span><span class="tsd-signature-symbol">)</span><span class="tsd-signature-symbol"> => </span><span class="tsd-signature-type">void</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/environment.ts#L69">environment.ts:69</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/environment.ts#L69">environment.ts:69</a></li>
</ul>
</aside>
<div class="tsd-type-declaration">
@@ -210,7 +210,7 @@
<div class="tsd-signature tsd-kind-icon">packedCFunc<wbr>Table<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">Array</span><span class="tsd-signature-symbol"><</span><span class="tsd-signature-type">ctypes.FTVMWasmPackedCFunc</span><span class="tsd-signature-symbol"> | </span><span class="tsd-signature-type">undefined</span><span class="tsd-signature-symbol">></span><span class="tsd-signature-symbol"> = [undefined,]</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/environment.ts#L78">environment.ts:78</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/environment.ts#L78">environment.ts:78</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -228,7 +228,7 @@
<div class="tsd-signature tsd-kind-icon">packedCFunc<wbr>Table<wbr>Free<wbr>Id<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">Array</span><span class="tsd-signature-symbol"><</span><span class="tsd-signature-type">number</span><span class="tsd-signature-symbol">></span><span class="tsd-signature-symbol"> = []</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/environment.ts#L84">environment.ts:84</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/environment.ts#L84">environment.ts:84</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -250,7 +250,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/environment.ts#L105">environment.ts:105</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/environment.ts#L105">environment.ts:105</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
diff --git a/docs/reference/api/typedoc/classes/ffilibrary.html b/docs/reference/api/typedoc/classes/ffilibrary.html
index 271ce3c..1757bc5 100644
--- a/docs/reference/api/typedoc/classes/ffilibrary.html
+++ b/docs/reference/api/typedoc/classes/ffilibrary.html
@@ -131,7 +131,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L49">runtime.ts:49</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L49">runtime.ts:49</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -156,7 +156,7 @@
<div class="tsd-signature tsd-kind-icon">exports<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">Record</span><span class="tsd-signature-symbol"><</span><span class="tsd-signature-type">string</span><span class="tsd-signature-symbol">, </span><span class="tsd-signature-type">Function</span><span class="tsd-signature-symbol">></span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L46">runtime.ts:46</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L46">runtime.ts:46</a></li>
</ul>
</aside>
</section>
@@ -166,7 +166,7 @@
<div class="tsd-signature tsd-kind-icon">memory<span class="tsd-signature-symbol">:</span> <a href="memory.html" class="tsd-signature-type">Memory</a></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L45">runtime.ts:45</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L45">runtime.ts:45</a></li>
</ul>
</aside>
</section>
@@ -176,7 +176,7 @@
<div class="tsd-signature tsd-kind-icon">wasm32<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">boolean</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L44">runtime.ts:44</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L44">runtime.ts:44</a></li>
</ul>
</aside>
</section>
@@ -186,7 +186,7 @@
<div class="tsd-signature tsd-kind-icon">webGPUContext<span class="tsd-signature-symbol">:</span> <a href="webgpucontext.html" class="tsd-signature-type">WebGPUContext</a></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L47">runtime.ts:47</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L47">runtime.ts:47</a></li>
</ul>
</aside>
</section>
@@ -203,7 +203,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L76">runtime.ts:76</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L76">runtime.ts:76</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -226,7 +226,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L66">runtime.ts:66</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L66">runtime.ts:66</a></li>
</ul>
</aside>
<h4 class="tsd-returns-title">Returns <span class="tsd-signature-type">void</span></h4>
@@ -243,7 +243,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L84">runtime.ts:84</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L84">runtime.ts:84</a></li>
</ul>
</aside>
<h4 class="tsd-returns-title">Returns <a href="cachedcallstack.html" class="tsd-signature-type">CachedCallStack</a></h4>
@@ -260,7 +260,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L95">runtime.ts:95</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L95">runtime.ts:95</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -283,7 +283,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L72">runtime.ts:72</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L72">runtime.ts:72</a></li>
</ul>
</aside>
<h4 class="tsd-returns-title">Returns <span class="tsd-signature-type">number</span></h4>
diff --git a/docs/reference/api/typedoc/classes/graphexecutor.html b/docs/reference/api/typedoc/classes/graphexecutor.html
index 49b6e33..f7a0de8 100644
--- a/docs/reference/api/typedoc/classes/graphexecutor.html
+++ b/docs/reference/api/typedoc/classes/graphexecutor.html
@@ -130,7 +130,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L583">runtime.ts:583</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L583">runtime.ts:583</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -162,7 +162,7 @@
<div class="tsd-signature tsd-kind-icon">module<span class="tsd-signature-symbol">:</span> <a href="module.html" class="tsd-signature-type">Module</a></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L579">runtime.ts:579</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L579">runtime.ts:579</a></li>
</ul>
</aside>
</section>
@@ -179,7 +179,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L654">runtime.ts:654</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L654">runtime.ts:654</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -224,7 +224,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L597">runtime.ts:597</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L597">runtime.ts:597</a></li>
</ul>
</aside>
<h4 class="tsd-returns-title">Returns <span class="tsd-signature-type">void</span></h4>
@@ -241,7 +241,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L631">runtime.ts:631</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L631">runtime.ts:631</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -279,7 +279,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L644">runtime.ts:644</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L644">runtime.ts:644</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -310,7 +310,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L621">runtime.ts:621</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L621">runtime.ts:621</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -332,7 +332,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L609">runtime.ts:609</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L609">runtime.ts:609</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
diff --git a/docs/reference/api/typedoc/classes/instance.html b/docs/reference/api/typedoc/classes/instance.html
index 8d5af13..6393c74 100644
--- a/docs/reference/api/typedoc/classes/instance.html
+++ b/docs/reference/api/typedoc/classes/instance.html
@@ -139,7 +139,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L692">runtime.ts:692</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L692">runtime.ts:692</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -202,7 +202,7 @@
<div class="tsd-signature tsd-kind-icon">exports<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">Record</span><span class="tsd-signature-symbol"><</span><span class="tsd-signature-type">string</span><span class="tsd-signature-symbol">, </span><span class="tsd-signature-type">Function</span><span class="tsd-signature-symbol">></span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L684">runtime.ts:684</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L684">runtime.ts:684</a></li>
</ul>
</aside>
</section>
@@ -212,7 +212,7 @@
<div class="tsd-signature tsd-kind-icon">memory<span class="tsd-signature-symbol">:</span> <a href="memory.html" class="tsd-signature-type">Memory</a></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L683">runtime.ts:683</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L683">runtime.ts:683</a></li>
</ul>
</aside>
</section>
@@ -229,7 +229,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L932">runtime.ts:932</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L932">runtime.ts:932</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -260,7 +260,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L994">runtime.ts:994</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L994">runtime.ts:994</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -303,7 +303,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L924">runtime.ts:924</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L924">runtime.ts:924</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -341,7 +341,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L732">runtime.ts:732</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L732">runtime.ts:732</a></li>
</ul>
</aside>
<h4 class="tsd-returns-title">Returns <span class="tsd-signature-type">void</span></h4>
@@ -358,7 +358,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L952">runtime.ts:952</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L952">runtime.ts:952</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -402,7 +402,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L816">runtime.ts:816</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L816">runtime.ts:816</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -434,7 +434,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L1033">runtime.ts:1033</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L1033">runtime.ts:1033</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -465,7 +465,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L846">runtime.ts:846</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L846">runtime.ts:846</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -497,7 +497,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L750">runtime.ts:750</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L750">runtime.ts:750</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -520,7 +520,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L1013">runtime.ts:1013</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L1013">runtime.ts:1013</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -568,7 +568,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L789">runtime.ts:789</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L789">runtime.ts:789</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -608,7 +608,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L914">runtime.ts:914</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L914">runtime.ts:914</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -646,7 +646,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L1134">runtime.ts:1134</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L1134">runtime.ts:1134</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -698,7 +698,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L740">runtime.ts:740</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L740">runtime.ts:740</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -722,7 +722,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L868">runtime.ts:868</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L868">runtime.ts:868</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -754,7 +754,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L857">runtime.ts:857</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L857">runtime.ts:857</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -786,7 +786,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L940">runtime.ts:940</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L940">runtime.ts:940</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
diff --git a/docs/reference/api/typedoc/classes/memory.html b/docs/reference/api/typedoc/classes/memory.html
index 1e16ff3..21223ee 100644
--- a/docs/reference/api/typedoc/classes/memory.html
+++ b/docs/reference/api/typedoc/classes/memory.html
@@ -130,7 +130,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/memory.ts#L40">memory.ts:40</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/memory.ts#L40">memory.ts:40</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -152,7 +152,7 @@
<div class="tsd-signature tsd-kind-icon">memory<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">Memory</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/memory.ts#L32">memory.ts:32</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/memory.ts#L32">memory.ts:32</a></li>
</ul>
</aside>
</section>
@@ -162,7 +162,7 @@
<div class="tsd-signature tsd-kind-icon">wasm32<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">boolean</span><span class="tsd-signature-symbol"> = true</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/memory.ts#L33">memory.ts:33</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/memory.ts#L33">memory.ts:33</a></li>
</ul>
</aside>
</section>
@@ -179,7 +179,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/memory.ts#L154">memory.ts:154</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/memory.ts#L154">memory.ts:154</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -210,7 +210,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/memory.ts#L90">memory.ts:90</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/memory.ts#L90">memory.ts:90</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -233,7 +233,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/memory.ts#L97">memory.ts:97</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/memory.ts#L97">memory.ts:97</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -256,7 +256,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/memory.ts#L74">memory.ts:74</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/memory.ts#L74">memory.ts:74</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -279,7 +279,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/memory.ts#L81">memory.ts:81</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/memory.ts#L81">memory.ts:81</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -302,7 +302,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/memory.ts#L104">memory.ts:104</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/memory.ts#L104">memory.ts:104</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -325,7 +325,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/memory.ts#L132">memory.ts:132</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/memory.ts#L132">memory.ts:132</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -362,7 +362,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/memory.ts#L145">memory.ts:145</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/memory.ts#L145">memory.ts:145</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -393,7 +393,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/memory.ts#L60">memory.ts:60</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/memory.ts#L60">memory.ts:60</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -416,7 +416,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/memory.ts#L67">memory.ts:67</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/memory.ts#L67">memory.ts:67</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -439,7 +439,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/memory.ts#L53">memory.ts:53</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/memory.ts#L53">memory.ts:53</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -462,7 +462,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/memory.ts#L114">memory.ts:114</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/memory.ts#L114">memory.ts:114</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -485,7 +485,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/memory.ts#L124">memory.ts:124</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/memory.ts#L124">memory.ts:124</a></li>
</ul>
</aside>
<h4 class="tsd-returns-title">Returns <span class="tsd-signature-type">number</span></h4>
@@ -502,7 +502,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/memory.ts#L175">memory.ts:175</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/memory.ts#L175">memory.ts:175</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
diff --git a/docs/reference/api/typedoc/classes/module.html b/docs/reference/api/typedoc/classes/module.html
index 9fed910..0088c45 100644
--- a/docs/reference/api/typedoc/classes/module.html
+++ b/docs/reference/api/typedoc/classes/module.html
@@ -124,7 +124,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L504">runtime.ts:504</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L504">runtime.ts:504</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -170,7 +170,7 @@
<div class="tsd-signature tsd-kind-icon">handle<span class="tsd-signature-symbol">:</span> <a href="../index.html#pointer" class="tsd-signature-type">Pointer</a></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L502">runtime.ts:502</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L502">runtime.ts:502</a></li>
</ul>
</aside>
</section>
@@ -187,7 +187,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L516">runtime.ts:516</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L516">runtime.ts:516</a></li>
</ul>
</aside>
<h4 class="tsd-returns-title">Returns <span class="tsd-signature-type">void</span></h4>
@@ -204,7 +204,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L530">runtime.ts:530</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L530">runtime.ts:530</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -236,7 +236,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L561">runtime.ts:561</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L561">runtime.ts:561</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
diff --git a/docs/reference/api/typedoc/classes/ndarray.html b/docs/reference/api/typedoc/classes/ndarray.html
index 0ee1f23..5b9d7eb 100644
--- a/docs/reference/api/typedoc/classes/ndarray.html
+++ b/docs/reference/api/typedoc/classes/ndarray.html
@@ -130,7 +130,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L304">runtime.ts:304</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L304">runtime.ts:304</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -158,7 +158,7 @@
<div class="tsd-signature tsd-kind-icon">device<span class="tsd-signature-symbol">:</span> <a href="dldevice.html" class="tsd-signature-type">DLDevice</a></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L297">runtime.ts:297</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L297">runtime.ts:297</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -173,7 +173,7 @@
<div class="tsd-signature tsd-kind-icon">dtype<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">string</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L293">runtime.ts:293</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L293">runtime.ts:293</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -188,7 +188,7 @@
<div class="tsd-signature tsd-kind-icon">handle<span class="tsd-signature-symbol">:</span> <a href="../index.html#pointer" class="tsd-signature-type">Pointer</a></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L289">runtime.ts:289</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L289">runtime.ts:289</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -203,7 +203,7 @@
<div class="tsd-signature tsd-kind-icon">ndim<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">number</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L291">runtime.ts:291</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L291">runtime.ts:291</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -218,7 +218,7 @@
<div class="tsd-signature tsd-kind-icon">shape<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">Array</span><span class="tsd-signature-symbol"><</span><span class="tsd-signature-type">number</span><span class="tsd-signature-symbol">></span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L295">runtime.ts:295</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L295">runtime.ts:295</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -240,7 +240,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L370">runtime.ts:370</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L370">runtime.ts:370</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -273,7 +273,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L414">runtime.ts:414</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L414">runtime.ts:414</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -305,7 +305,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L355">runtime.ts:355</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L355">runtime.ts:355</a></li>
</ul>
</aside>
<h4 class="tsd-returns-title">Returns <span class="tsd-signature-type">void</span></h4>
@@ -322,7 +322,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L474">runtime.ts:474</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L474">runtime.ts:474</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -346,7 +346,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L443">runtime.ts:443</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L443">runtime.ts:443</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
diff --git a/docs/reference/api/typedoc/classes/packedfunccell.html b/docs/reference/api/typedoc/classes/packedfunccell.html
index 6b085af..8a27e0b 100644
--- a/docs/reference/api/typedoc/classes/packedfunccell.html
+++ b/docs/reference/api/typedoc/classes/packedfunccell.html
@@ -122,7 +122,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L158">runtime.ts:158</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L158">runtime.ts:158</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -147,7 +147,7 @@
<div class="tsd-signature tsd-kind-icon">handle<span class="tsd-signature-symbol">:</span> <a href="../index.html#pointer" class="tsd-signature-type">Pointer</a></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L157">runtime.ts:157</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L157">runtime.ts:157</a></li>
</ul>
</aside>
</section>
@@ -164,7 +164,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L165">runtime.ts:165</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L165">runtime.ts:165</a></li>
</ul>
</aside>
<h4 class="tsd-returns-title">Returns <span class="tsd-signature-type">void</span></h4>
diff --git a/docs/reference/api/typedoc/classes/rpcserver.html b/docs/reference/api/typedoc/classes/rpcserver.html
index ace04b3..97f4efc 100644
--- a/docs/reference/api/typedoc/classes/rpcserver.html
+++ b/docs/reference/api/typedoc/classes/rpcserver.html
@@ -115,7 +115,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/rpc_server.ts#L92">rpc_server.ts:92</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/rpc_server.ts#L92">rpc_server.ts:92</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -176,7 +176,7 @@
<div class="tsd-signature tsd-kind-icon">get<wbr>Imports<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol">(</span><span class="tsd-signature-symbol">)</span><span class="tsd-signature-symbol"> => </span><span class="tsd-signature-type">Record</span><span class="tsd-signature-symbol"><</span><span class="tsd-signature-type">string</span><span class="tsd-signature-symbol">, </span><span class="tsd-signature-type">unknown</span><span class="tsd-signat [...]
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/rpc_server.ts#L82">rpc_server.ts:82</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/rpc_server.ts#L82">rpc_server.ts:82</a></li>
</ul>
</aside>
<div class="tsd-type-declaration">
@@ -201,7 +201,7 @@
<div class="tsd-signature tsd-kind-icon">key<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">string</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/rpc_server.ts#L78">rpc_server.ts:78</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/rpc_server.ts#L78">rpc_server.ts:78</a></li>
</ul>
</aside>
</section>
@@ -211,7 +211,7 @@
<div class="tsd-signature tsd-kind-icon">logger<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol">(</span>msg<span class="tsd-signature-symbol">: </span><span class="tsd-signature-type">string</span><span class="tsd-signature-symbol">)</span><span class="tsd-signature-symbol"> => </span><span class="tsd-signature-type">void</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/rpc_server.ts#L81">rpc_server.ts:81</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/rpc_server.ts#L81">rpc_server.ts:81</a></li>
</ul>
</aside>
<div class="tsd-type-declaration">
@@ -242,7 +242,7 @@
<div class="tsd-signature tsd-kind-icon">socket<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">WebSocket</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/rpc_server.ts#L79">rpc_server.ts:79</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/rpc_server.ts#L79">rpc_server.ts:79</a></li>
</ul>
</aside>
</section>
@@ -252,7 +252,7 @@
<div class="tsd-signature tsd-kind-icon">state<span class="tsd-signature-symbol">:</span> <a href="../enums/rpcserverstate.html" class="tsd-signature-type">RPCServerState</a><span class="tsd-signature-symbol"> = RPCServerState.InitHeader</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/rpc_server.ts#L80">rpc_server.ts:80</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/rpc_server.ts#L80">rpc_server.ts:80</a></li>
</ul>
</aside>
</section>
@@ -262,7 +262,7 @@
<div class="tsd-signature tsd-kind-icon">url<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">string</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/rpc_server.ts#L77">rpc_server.ts:77</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/rpc_server.ts#L77">rpc_server.ts:77</a></li>
</ul>
</aside>
</section>
diff --git a/docs/reference/api/typedoc/classes/scalar.html b/docs/reference/api/typedoc/classes/scalar.html
index ea2a6b3..e018786 100644
--- a/docs/reference/api/typedoc/classes/scalar.html
+++ b/docs/reference/api/typedoc/classes/scalar.html
@@ -112,7 +112,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L145">runtime.ts:145</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L145">runtime.ts:145</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -137,7 +137,7 @@
<div class="tsd-signature tsd-kind-icon">dtype<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">string</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L145">runtime.ts:145</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L145">runtime.ts:145</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -152,7 +152,7 @@
<div class="tsd-signature tsd-kind-icon">value<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">number</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L143">runtime.ts:143</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L143">runtime.ts:143</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
diff --git a/docs/reference/api/typedoc/classes/webgpucontext.html b/docs/reference/api/typedoc/classes/webgpucontext.html
index 16cf52e..ce2920a 100644
--- a/docs/reference/api/typedoc/classes/webgpucontext.html
+++ b/docs/reference/api/typedoc/classes/webgpucontext.html
@@ -120,7 +120,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/webgpu.ts#L57">webgpu.ts:57</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/webgpu.ts#L57">webgpu.ts:57</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -145,7 +145,7 @@
<div class="tsd-signature tsd-kind-icon">device<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">GPUDevice</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/webgpu.ts#L50">webgpu.ts:50</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/webgpu.ts#L50">webgpu.ts:50</a></li>
</ul>
</aside>
</section>
@@ -155,7 +155,7 @@
<div class="tsd-signature tsd-kind-icon">memory<span class="tsd-signature-symbol">:</span> <a href="memory.html" class="tsd-signature-type">Memory</a></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/webgpu.ts#L51">webgpu.ts:51</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/webgpu.ts#L51">webgpu.ts:51</a></li>
</ul>
</aside>
</section>
@@ -172,7 +172,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/webgpu.ts#L84">webgpu.ts:84</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/webgpu.ts#L84">webgpu.ts:84</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -209,7 +209,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/webgpu.ts#L170">webgpu.ts:170</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/webgpu.ts#L170">webgpu.ts:170</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -238,7 +238,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/webgpu.ts#L67">webgpu.ts:67</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/webgpu.ts#L67">webgpu.ts:67</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
diff --git a/docs/reference/api/typedoc/enums/argtypecode.html b/docs/reference/api/typedoc/enums/argtypecode.html
index 7549405..9c54cc1 100644
--- a/docs/reference/api/typedoc/enums/argtypecode.html
+++ b/docs/reference/api/typedoc/enums/argtypecode.html
@@ -106,7 +106,7 @@
<div class="tsd-signature tsd-kind-icon">DLDevice<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol"> = 6</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L220">ctypes.ts:220</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L220">ctypes.ts:220</a></li>
</ul>
</aside>
</section>
@@ -116,7 +116,7 @@
<div class="tsd-signature tsd-kind-icon">Float<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol"> = 2</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L216">ctypes.ts:216</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L216">ctypes.ts:216</a></li>
</ul>
</aside>
</section>
@@ -126,7 +126,7 @@
<div class="tsd-signature tsd-kind-icon">Int<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol"> = 0</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L214">ctypes.ts:214</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L214">ctypes.ts:214</a></li>
</ul>
</aside>
</section>
@@ -136,7 +136,7 @@
<div class="tsd-signature tsd-kind-icon">Null<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol"> = 4</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L218">ctypes.ts:218</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L218">ctypes.ts:218</a></li>
</ul>
</aside>
</section>
@@ -146,7 +146,7 @@
<div class="tsd-signature tsd-kind-icon">TVMBytes<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol"> = 12</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L226">ctypes.ts:226</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L226">ctypes.ts:226</a></li>
</ul>
</aside>
</section>
@@ -156,7 +156,7 @@
<div class="tsd-signature tsd-kind-icon">TVMDLTensor<wbr>Handle<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol"> = 7</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L221">ctypes.ts:221</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L221">ctypes.ts:221</a></li>
</ul>
</aside>
</section>
@@ -166,7 +166,7 @@
<div class="tsd-signature tsd-kind-icon">TVMData<wbr>Type<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol"> = 5</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L219">ctypes.ts:219</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L219">ctypes.ts:219</a></li>
</ul>
</aside>
</section>
@@ -176,7 +176,7 @@
<div class="tsd-signature tsd-kind-icon">TVMModule<wbr>Handle<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol"> = 9</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L223">ctypes.ts:223</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L223">ctypes.ts:223</a></li>
</ul>
</aside>
</section>
@@ -186,7 +186,7 @@
<div class="tsd-signature tsd-kind-icon">TVMNDArray<wbr>Handle<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol"> = 13</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L227">ctypes.ts:227</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L227">ctypes.ts:227</a></li>
</ul>
</aside>
</section>
@@ -196,7 +196,7 @@
<div class="tsd-signature tsd-kind-icon">TVMObject<wbr>Handle<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol"> = 8</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L222">ctypes.ts:222</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L222">ctypes.ts:222</a></li>
</ul>
</aside>
</section>
@@ -206,7 +206,7 @@
<div class="tsd-signature tsd-kind-icon">TVMObjectRValue<wbr>Ref<wbr>Arg<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol"> = 14</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L228">ctypes.ts:228</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L228">ctypes.ts:228</a></li>
</ul>
</aside>
</section>
@@ -216,7 +216,7 @@
<div class="tsd-signature tsd-kind-icon">TVMOpaque<wbr>Handle<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol"> = 3</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L217">ctypes.ts:217</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L217">ctypes.ts:217</a></li>
</ul>
</aside>
</section>
@@ -226,7 +226,7 @@
<div class="tsd-signature tsd-kind-icon">TVMPacked<wbr>Func<wbr>Handle<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol"> = 10</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L224">ctypes.ts:224</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L224">ctypes.ts:224</a></li>
</ul>
</aside>
</section>
@@ -236,7 +236,7 @@
<div class="tsd-signature tsd-kind-icon">TVMStr<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol"> = 11</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L225">ctypes.ts:225</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L225">ctypes.ts:225</a></li>
</ul>
</aside>
</section>
@@ -246,7 +246,7 @@
<div class="tsd-signature tsd-kind-icon">UInt<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol"> = 1</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L215">ctypes.ts:215</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L215">ctypes.ts:215</a></li>
</ul>
</aside>
</section>
diff --git a/docs/reference/api/typedoc/enums/aynccallbackcode.html b/docs/reference/api/typedoc/enums/aynccallbackcode.html
index fe69541..5322b7a 100644
--- a/docs/reference/api/typedoc/enums/aynccallbackcode.html
+++ b/docs/reference/api/typedoc/enums/aynccallbackcode.html
@@ -93,7 +93,7 @@
<div class="tsd-signature tsd-kind-icon">k<wbr>Exception<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol"> = 5</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L676">runtime.ts:676</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L676">runtime.ts:676</a></li>
</ul>
</aside>
</section>
@@ -103,7 +103,7 @@
<div class="tsd-signature tsd-kind-icon">k<wbr>Return<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol"> = 4</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L675">runtime.ts:675</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L675">runtime.ts:675</a></li>
</ul>
</aside>
</section>
diff --git a/docs/reference/api/typedoc/enums/dldatatypecode.html b/docs/reference/api/typedoc/enums/dldatatypecode.html
index 670fcad..a8abba6 100644
--- a/docs/reference/api/typedoc/enums/dldatatypecode.html
+++ b/docs/reference/api/typedoc/enums/dldatatypecode.html
@@ -95,7 +95,7 @@
<div class="tsd-signature tsd-kind-icon">Float<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol"> = 2</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L242">runtime.ts:242</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L242">runtime.ts:242</a></li>
</ul>
</aside>
</section>
@@ -105,7 +105,7 @@
<div class="tsd-signature tsd-kind-icon">Int<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol"> = 0</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L240">runtime.ts:240</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L240">runtime.ts:240</a></li>
</ul>
</aside>
</section>
@@ -115,7 +115,7 @@
<div class="tsd-signature tsd-kind-icon">Opaque<wbr>Handle<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol"> = 3</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L243">runtime.ts:243</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L243">runtime.ts:243</a></li>
</ul>
</aside>
</section>
@@ -125,7 +125,7 @@
<div class="tsd-signature tsd-kind-icon">UInt<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol"> = 1</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L241">runtime.ts:241</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L241">runtime.ts:241</a></li>
</ul>
</aside>
</section>
diff --git a/docs/reference/api/typedoc/enums/rpcserverstate.html b/docs/reference/api/typedoc/enums/rpcserverstate.html
index ecb9970..1ad0aed 100644
--- a/docs/reference/api/typedoc/enums/rpcserverstate.html
+++ b/docs/reference/api/typedoc/enums/rpcserverstate.html
@@ -90,7 +90,7 @@
<div class="tsd-signature tsd-kind-icon">Init<wbr>Header<span class="tsd-signature-symbol">:</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/rpc_server.ts#L27">rpc_server.ts:27</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/rpc_server.ts#L27">rpc_server.ts:27</a></li>
</ul>
</aside>
</section>
@@ -100,7 +100,7 @@
<div class="tsd-signature tsd-kind-icon">Init<wbr>Header<wbr>Key<span class="tsd-signature-symbol">:</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/rpc_server.ts#L28">rpc_server.ts:28</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/rpc_server.ts#L28">rpc_server.ts:28</a></li>
</ul>
</aside>
</section>
@@ -110,7 +110,7 @@
<div class="tsd-signature tsd-kind-icon">Init<wbr>Server<span class="tsd-signature-symbol">:</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/rpc_server.ts#L29">rpc_server.ts:29</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/rpc_server.ts#L29">rpc_server.ts:29</a></li>
</ul>
</aside>
</section>
@@ -120,7 +120,7 @@
<div class="tsd-signature tsd-kind-icon">Receive<wbr>Packet<wbr>Body<span class="tsd-signature-symbol">:</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/rpc_server.ts#L32">rpc_server.ts:32</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/rpc_server.ts#L32">rpc_server.ts:32</a></li>
</ul>
</aside>
</section>
@@ -130,7 +130,7 @@
<div class="tsd-signature tsd-kind-icon">Receive<wbr>Packet<wbr>Header<span class="tsd-signature-symbol">:</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/rpc_server.ts#L31">rpc_server.ts:31</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/rpc_server.ts#L31">rpc_server.ts:31</a></li>
</ul>
</aside>
</section>
@@ -140,7 +140,7 @@
<div class="tsd-signature tsd-kind-icon">Wait<wbr>For<wbr>Callback<span class="tsd-signature-symbol">:</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/rpc_server.ts#L30">rpc_server.ts:30</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/rpc_server.ts#L30">rpc_server.ts:30</a></li>
</ul>
</aside>
</section>
diff --git a/docs/reference/api/typedoc/enums/sizeof.html b/docs/reference/api/typedoc/enums/sizeof.html
index 6099f7c..6168e3f 100644
--- a/docs/reference/api/typedoc/enums/sizeof.html
+++ b/docs/reference/api/typedoc/enums/sizeof.html
@@ -100,7 +100,7 @@
<div class="tsd-signature tsd-kind-icon">DLData<wbr>Type<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol"> = I32</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L206">ctypes.ts:206</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L206">ctypes.ts:206</a></li>
</ul>
</aside>
</section>
@@ -110,7 +110,7 @@
<div class="tsd-signature tsd-kind-icon">DLDevice<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol"> = I32 + I32</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L207">ctypes.ts:207</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L207">ctypes.ts:207</a></li>
</ul>
</aside>
</section>
@@ -120,7 +120,7 @@
<div class="tsd-signature tsd-kind-icon">F32<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol"> = 4</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L203">ctypes.ts:203</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L203">ctypes.ts:203</a></li>
</ul>
</aside>
</section>
@@ -130,7 +130,7 @@
<div class="tsd-signature tsd-kind-icon">F64<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol"> = 8</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L204">ctypes.ts:204</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L204">ctypes.ts:204</a></li>
</ul>
</aside>
</section>
@@ -140,7 +140,7 @@
<div class="tsd-signature tsd-kind-icon">I32<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol"> = 4</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L201">ctypes.ts:201</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L201">ctypes.ts:201</a></li>
</ul>
</aside>
</section>
@@ -150,7 +150,7 @@
<div class="tsd-signature tsd-kind-icon">I64<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol"> = 8</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L202">ctypes.ts:202</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L202">ctypes.ts:202</a></li>
</ul>
</aside>
</section>
@@ -160,7 +160,7 @@
<div class="tsd-signature tsd-kind-icon">TVMValue<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol"> = 8</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L205">ctypes.ts:205</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L205">ctypes.ts:205</a></li>
</ul>
</aside>
</section>
@@ -170,7 +170,7 @@
<div class="tsd-signature tsd-kind-icon">U16<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol"> = 2</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L200">ctypes.ts:200</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L200">ctypes.ts:200</a></li>
</ul>
</aside>
</section>
@@ -180,7 +180,7 @@
<div class="tsd-signature tsd-kind-icon">U8<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol"> = 1</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L199">ctypes.ts:199</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L199">ctypes.ts:199</a></li>
</ul>
</aside>
</section>
diff --git a/docs/reference/api/typedoc/index.html b/docs/reference/api/typedoc/index.html
index d8b7770..60e2dcb 100644
--- a/docs/reference/api/typedoc/index.html
+++ b/docs/reference/api/typedoc/index.html
@@ -174,7 +174,7 @@
<div class="tsd-signature tsd-kind-icon">FTVMArray<wbr>Alloc<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol">(</span>shape<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a>, ndim<span class="tsd-signature-symbol">: </span><span class="tsd-signature-type">number</span>, dtypeCode<span class="tsd-signature-symbol">: </span><span class="tsd-signature-type">number</span>, dtypeBits<span class="tsd [...]
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L112">ctypes.ts:112</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L112">ctypes.ts:112</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -238,7 +238,7 @@
<div class="tsd-signature tsd-kind-icon">FTVMArray<wbr>Copy<wbr>From<wbr>Bytes<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol">(</span>handle<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a>, data<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a>, nbytes<span class="tsd-signature-symbol">: </span><span class="tsd-signature-type">num [...]
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L128">ctypes.ts:128</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L128">ctypes.ts:128</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -282,7 +282,7 @@
<div class="tsd-signature tsd-kind-icon">FTVMArray<wbr>Copy<wbr>From<wbr>To<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol">(</span>from<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a>, to<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a>, stream<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-sig [...]
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L144">ctypes.ts:144</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L144">ctypes.ts:144</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -326,7 +326,7 @@
<div class="tsd-signature tsd-kind-icon">FTVMArray<wbr>Copy<wbr>ToBytes<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol">(</span>handle<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a>, data<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a>, nbytes<span class="tsd-signature-symbol">: </span><span class="tsd-signature-type">number</sp [...]
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L136">ctypes.ts:136</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L136">ctypes.ts:136</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -370,7 +370,7 @@
<div class="tsd-signature tsd-kind-icon">FTVMArray<wbr>Free<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol">(</span>handle<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a><span class="tsd-signature-symbol">)</span><span class="tsd-signature-symbol"> => </span><span class="tsd-signature-type">number</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L121">ctypes.ts:121</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L121">ctypes.ts:121</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -406,7 +406,7 @@
<div class="tsd-signature tsd-kind-icon">FTVMBackend<wbr>PackedCFunc<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol">(</span>argValues<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a>, argCodes<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a>, nargs<span class="tsd-signature-symbol">: </span><span class="tsd-signature-type">number< [...]
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L160">ctypes.ts:160</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L160">ctypes.ts:160</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -458,7 +458,7 @@
<div class="tsd-signature tsd-kind-icon">FTVMCFunc<wbr>Set<wbr>Return<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol">(</span>ret<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a>, value<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a>, typeCode<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signa [...]
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L77">ctypes.ts:77</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L77">ctypes.ts:77</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -506,7 +506,7 @@
<div class="tsd-signature tsd-kind-icon">FTVMCb<wbr>Arg<wbr>ToReturn<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol">(</span>value<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a>, code<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a><span class="tsd-signature-symbol">)</span><span class="tsd-signature-symbol"> => </span><span c [...]
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L83">ctypes.ts:83</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L83">ctypes.ts:83</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -545,7 +545,7 @@
<div class="tsd-signature tsd-kind-icon">FTVMFunc<wbr>Call<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol">(</span>func<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a>, argValues<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a>, typeCode<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signature-t [...]
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L67">ctypes.ts:67</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L67">ctypes.ts:67</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -601,7 +601,7 @@
<div class="tsd-signature tsd-kind-icon">FTVMFunc<wbr>Free<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol">(</span>func<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a><span class="tsd-signature-symbol">)</span><span class="tsd-signature-symbol"> => </span><span class="tsd-signature-type">number</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L57">ctypes.ts:57</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L57">ctypes.ts:57</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -637,7 +637,7 @@
<div class="tsd-signature tsd-kind-icon">FTVMFunc<wbr>Get<wbr>Global<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol">(</span>name<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a>, out<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a><span class="tsd-signature-symbol">)</span><span class="tsd-signature-symbol"> => </span><span cla [...]
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L100">ctypes.ts:100</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L100">ctypes.ts:100</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -676,7 +676,7 @@
<div class="tsd-signature tsd-kind-icon">FTVMFunc<wbr>List<wbr>Global<wbr>Names<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol">(</span>outSize<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a>, outArray<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a><span class="tsd-signature-symbol">)</span><span class="tsd-signature-symbol"> =&g [...]
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L88">ctypes.ts:88</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L88">ctypes.ts:88</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -715,7 +715,7 @@
<div class="tsd-signature tsd-kind-icon">FTVMFunc<wbr>Register<wbr>Global<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol">(</span>name<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a>, f<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a>, override<span class="tsd-signature-symbol">: </span><span class="tsd-signature-type">number</spa [...]
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L94">ctypes.ts:94</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L94">ctypes.ts:94</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -758,7 +758,7 @@
<div class="tsd-signature tsd-kind-icon">FTVMGet<wbr>Last<wbr>Error<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol">(</span><span class="tsd-signature-symbol">)</span><span class="tsd-signature-symbol"> => </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L34">ctypes.ts:34</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L34">ctypes.ts:34</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -788,7 +788,7 @@
<div class="tsd-signature tsd-kind-icon">FTVMMod<wbr>Free<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol">(</span>mod<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a><span class="tsd-signature-symbol">)</span><span class="tsd-signature-symbol"> => </span><span class="tsd-signature-type">number</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L52">ctypes.ts:52</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L52">ctypes.ts:52</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -824,7 +824,7 @@
<div class="tsd-signature tsd-kind-icon">FTVMMod<wbr>Get<wbr>Function<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol">(</span>mod<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a>, funcName<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a>, queryImports<span class="tsd-signature-symbol">: </span><span class="tsd-signature-type">numbe [...]
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L42">ctypes.ts:42</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L42">ctypes.ts:42</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -872,7 +872,7 @@
<div class="tsd-signature tsd-kind-icon">FTVMMod<wbr>Import<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol">(</span>mod<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a>, dep<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a><span class="tsd-signature-symbol">)</span><span class="tsd-signature-symbol"> => </span><span class="tsd-si [...]
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L48">ctypes.ts:48</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L48">ctypes.ts:48</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -912,7 +912,7 @@
<div class="tsd-signature tsd-kind-icon">FTVMSynchronize<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol">(</span>deviceType<span class="tsd-signature-symbol">: </span><span class="tsd-signature-type">number</span>, deviceId<span class="tsd-signature-symbol">: </span><span class="tsd-signature-type">number</span>, stream<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a><span class="tsd-signatur [...]
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L150">ctypes.ts:150</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L150">ctypes.ts:150</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -954,7 +954,7 @@
<div class="tsd-signature tsd-kind-icon">FTVMWasm<wbr>Alloc<wbr>Space<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol">(</span>size<span class="tsd-signature-symbol">: </span><span class="tsd-signature-type">number</span><span class="tsd-signature-symbol">)</span><span class="tsd-signature-symbol"> => </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L167">ctypes.ts:167</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L167">ctypes.ts:167</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -990,7 +990,7 @@
<div class="tsd-signature tsd-kind-icon">FTVMWasm<wbr>Free<wbr>Space<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol">(</span>ptr<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a><span class="tsd-signature-symbol">)</span><span class="tsd-signature-symbol"> => </span><span class="tsd-signature-type">void</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L170">ctypes.ts:170</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L170">ctypes.ts:170</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -1026,7 +1026,7 @@
<div class="tsd-signature tsd-kind-icon">FTVMWasm<wbr>Func<wbr>Create<wbr>FromCFunc<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol">(</span>resource<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a>, out<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a><span class="tsd-signature-symbol">)</span><span class="tsd-signature-symbol"> =&g [...]
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L187">ctypes.ts:187</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L187">ctypes.ts:187</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -1066,7 +1066,7 @@
<div class="tsd-signature tsd-kind-icon">FTVMWasm<wbr>PackedCFunc<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol">(</span>args<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a>, typeCodes<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a>, nargs<span class="tsd-signature-symbol">: </span><span class="tsd-signature-type">number</span>, [...]
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L179">ctypes.ts:179</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L179">ctypes.ts:179</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -1118,7 +1118,7 @@
<div class="tsd-signature tsd-kind-icon">FTVMWasm<wbr>PackedCFunc<wbr>Finalizer<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol">(</span>resourceHandle<span class="tsd-signature-symbol">: </span><a href="index.html#pointer" class="tsd-signature-type">Pointer</a><span class="tsd-signature-symbol">)</span><span class="tsd-signature-symbol"> => </span><span class="tsd-signature-type">void</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L193">ctypes.ts:193</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L193">ctypes.ts:193</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -1154,7 +1154,7 @@
<div class="tsd-signature tsd-kind-icon">GPUPointer<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">number</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/webgpu.ts#L25">webgpu.ts:25</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/webgpu.ts#L25">webgpu.ts:25</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -1169,7 +1169,7 @@
<div class="tsd-signature tsd-kind-icon">Packed<wbr>Func<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol">(</span><span class="tsd-signature-symbol">...</span>args<span class="tsd-signature-symbol">: </span><span class="tsd-signature-type">any</span><span class="tsd-signature-symbol">)</span><span class="tsd-signature-symbol"> => </span><span class="tsd-signature-type">any</span><span class="tsd-signature-symbol"> & </span><a href="interfaces/disp [...]
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L36">runtime.ts:36</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L36">runtime.ts:36</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -1184,7 +1184,7 @@
<div class="tsd-signature tsd-kind-icon">Pointer<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">number</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L25">ctypes.ts:25</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L25">ctypes.ts:25</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -1199,7 +1199,7 @@
<div class="tsd-signature tsd-kind-icon">Ptr<wbr>Offset<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">number</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/ctypes.ts#L28">ctypes.ts:28</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/ctypes.ts#L28">ctypes.ts:28</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -1217,7 +1217,7 @@
<div class="tsd-signature tsd-kind-icon">RPC_<wbr>MAGIC<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">1045105</span><span class="tsd-signature-symbol"> = 1045105</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/rpc_server.ts#L36">rpc_server.ts:36</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/rpc_server.ts#L36">rpc_server.ts:36</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -1239,7 +1239,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/support.ts#L25">support.ts:25</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/support.ts#L25">support.ts:25</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -1271,7 +1271,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/support.ts#L39">support.ts:39</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/support.ts#L39">support.ts:39</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -1300,7 +1300,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/support.ts#L52">support.ts:52</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/support.ts#L52">support.ts:52</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -1337,7 +1337,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/compact.ts#L38">compact.ts:38</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/compact.ts#L38">compact.ts:38</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -1368,7 +1368,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/webgpu.ts#L30">webgpu.ts:30</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/webgpu.ts#L30">webgpu.ts:30</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -1390,7 +1390,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/environment.ts#L32">environment.ts:32</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/environment.ts#L32">environment.ts:32</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -1421,7 +1421,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/compact.ts#L24">compact.ts:24</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/compact.ts#L24">compact.ts:24</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -1443,7 +1443,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/runtime.ts#L1356">runtime.ts:1356</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/runtime.ts#L1356">runtime.ts:1356</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -1508,7 +1508,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/72c761c80/web/src/support.ts#L62">support.ts:62</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/42985c372/web/src/support.ts#L62">support.ts:62</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -1530,7 +1530,7 @@
<div class="tsd-signature tsd-kind-icon">DLData<wbr>Type<wbr>Code<wbr>ToStr<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">object</span></div>
<aside class="tsd-sources">
... 807 lines suppressed ...