[tvm-site] branch asf-site updated: deploying docs (apache/tvm@19e5ec65760e0edc3ae1c6e0a05cb9e78a139fd1)

[tq...@apache.org]

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 37168bfaa deploying docs (apache/tvm@19e5ec65760e0edc3ae1c6e0a05cb9e78a139fd1)
37168bfaa is described below

commit 37168bfaa5fba11abed7255f609ab44c4af7cfb7
Author: tvm-bot <95...@users.noreply.github.com>
AuthorDate: Tue Jul 26 01:51:29 2022 +0000

    deploying docs (apache/tvm@19e5ec65760e0edc3ae1c6e0a05cb9e78a139fd1)
---
 .../how_to/compile_models/from_darknet.rst.txt     |    2 +-
 .../how_to/compile_models/from_mxnet.rst.txt       |    2 +-
 .../how_to/compile_models/from_oneflow.rst.txt     |    2 +-
 .../how_to/compile_models/from_pytorch.rst.txt     |    2 +-
 .../how_to/compile_models/from_tensorflow.rst.txt  |    2 +-
 .../compile_models/sg_execution_times.rst.txt      |   22 +-
 .../deploy_models/deploy_model_on_android.rst.txt  |    2 +-
 .../deploy_object_detection_pytorch.rst.txt        |    4 +-
 .../deploy_models/deploy_prequantized.rst.txt      |    6 +-
 .../deploy_prequantized_tflite.rst.txt             |    4 +-
 .../how_to/deploy_models/deploy_quantized.rst.txt  |    2 +-
 .../deploy_models/deploy_ssd_gluoncv.rst.txt       |    4 +-
 .../deploy_models/sg_execution_times.rst.txt       |   16 +-
 .../extend_tvm/bring_your_own_datatypes.rst.txt    |    2 +-
 .../how_to/extend_tvm/sg_execution_times.rst.txt   |    8 +-
 .../how_to/extend_tvm/use_pass_instrument.rst.txt  |   16 +-
 .../optimize_operators/opt_conv_cuda.rst.txt       |    2 +-
 .../optimize_operators/opt_conv_tensorcore.rst.txt |    2 +-
 .../how_to/optimize_operators/opt_gemm.rst.txt     |   16 +-
 .../optimize_operators/sg_execution_times.rst.txt  |    8 +-
 .../sg_execution_times.rst.txt                     |   14 +-
 .../tune_conv2d_layer_cuda.rst.txt                 | 2205 ++++++++------------
 .../tune_network_cuda.rst.txt                      |    2 +-
 .../tune_network_x86.rst.txt                       |    4 +-
 .../tune_sparse_x86.rst.txt                        |  669 +++++-
 .../tune_with_autotvm/sg_execution_times.rst.txt   |    4 +-
 .../tune_with_autotvm/tune_conv2d_cuda.rst.txt     |   26 +-
 .../work_with_microtvm/micro_autotune.rst.txt      |   16 +-
 .../how_to/work_with_microtvm/micro_train.rst.txt  |   16 +-
 .../work_with_microtvm/sg_execution_times.rst.txt  |    8 +-
 .../work_with_relay/sg_execution_times.rst.txt     |   10 +-
 .../how_to/work_with_schedules/intrin_math.rst.txt |    2 +-
 .../work_with_schedules/sg_execution_times.rst.txt |   14 +-
 .../how_to/work_with_schedules/tensorize.rst.txt   |    2 +-
 .../tutorials/autotvm/sg_execution_times.rst.txt   |    6 +-
 .../frontend/deploy_classification.rst.txt         |    2 +-
 .../tutorials/frontend/deploy_detection.rst.txt    |    2 +-
 .../tutorials/frontend/sg_execution_times.rst.txt  |    6 +-
 .../tutorials/optimize/sg_execution_times.rst.txt  |    6 +-
 .../topic/vta/tutorials/sg_execution_times.rst.txt |    4 +-
 .../tutorial/auto_scheduler_matmul_x86.rst.txt     |    2 +-
 docs/_sources/tutorial/autotvm_matmul_x86.rst.txt  |   20 +-
 docs/_sources/tutorial/autotvm_relay_x86.rst.txt   |   56 +-
 .../tutorial/cross_compilation_and_rpc.rst.txt     |    2 +-
 docs/_sources/tutorial/intro_topi.rst.txt          |    2 +-
 docs/_sources/tutorial/sg_execution_times.rst.txt  |   24 +-
 .../tutorial/tensor_expr_get_started.rst.txt       |   47 +-
 docs/commit_hash                                   |    2 +-
 docs/how_to/compile_models/from_darknet.html       |    2 +-
 docs/how_to/compile_models/from_mxnet.html         |    2 +-
 docs/how_to/compile_models/from_oneflow.html       |   14 +-
 docs/how_to/compile_models/from_pytorch.html       |    5 +-
 docs/how_to/compile_models/from_tensorflow.html    |    2 +-
 docs/how_to/compile_models/sg_execution_times.html |   22 +-
 .../deploy_models/deploy_model_on_android.html     |    2 +-
 .../deploy_object_detection_pytorch.html           |   18 +-
 docs/how_to/deploy_models/deploy_prequantized.html |   21 +-
 .../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     |    8 +-
 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                    | 2205 ++++++++------------
 .../tune_with_autoscheduler/tune_network_cuda.html |    2 +-
 .../tune_with_autoscheduler/tune_network_x86.html  |    4 +-
 .../tune_with_autoscheduler/tune_sparse_x86.html   |  669 +++++-
 .../tune_with_autotvm/sg_execution_times.html      |    4 +-
 .../how_to/tune_with_autotvm/tune_conv2d_cuda.html |   26 +-
 docs/how_to/work_with_microtvm/micro_autotune.html |   16 +-
 docs/how_to/work_with_microtvm/micro_train.html    |   16 +-
 .../work_with_microtvm/sg_execution_times.html     |    8 +-
 .../how_to/work_with_relay/sg_execution_times.html |   10 +-
 docs/how_to/work_with_schedules/intrin_math.html   |    2 +-
 .../work_with_schedules/sg_execution_times.html    |   14 +-
 docs/how_to/work_with_schedules/tensorize.html     |    2 +-
 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   |    4 +-
 docs/tutorial/auto_scheduler_matmul_x86.html       |    2 +-
 docs/tutorial/autotvm_matmul_x86.html              |   20 +-
 docs/tutorial/autotvm_relay_x86.html               |  262 +--
 docs/tutorial/cross_compilation_and_rpc.html       |    2 +-
 docs/tutorial/intro_topi.html                      |    2 +-
 docs/tutorial/sg_execution_times.html              |   30 +-
 docs/tutorial/tensor_expr_get_started.html         |   43 +-
 121 files changed, 3900 insertions(+), 3439 deletions(-)

diff --git a/docs/_sources/how_to/compile_models/from_darknet.rst.txt b/docs/_sources/how_to/compile_models/from_darknet.rst.txt
index d2b65c2e2..fd7429e9e 100644
--- a/docs/_sources/how_to/compile_models/from_darknet.rst.txt
+++ b/docs/_sources/how_to/compile_models/from_darknet.rst.txt
@@ -317,7 +317,7 @@ The process is no different from other examples.
 
 .. rst-class:: sphx-glr-timing
 
-   **Total running time of the script:** ( 1 minutes  3.722 seconds)
+   **Total running time of the script:** ( 1 minutes  3.919 seconds)
 
 
 .. _sphx_glr_download_how_to_compile_models_from_darknet.py:
diff --git a/docs/_sources/how_to/compile_models/from_mxnet.rst.txt b/docs/_sources/how_to/compile_models/from_mxnet.rst.txt
index 868bbd27d..8c9ca2083 100644
--- a/docs/_sources/how_to/compile_models/from_mxnet.rst.txt
+++ b/docs/_sources/how_to/compile_models/from_mxnet.rst.txt
@@ -115,7 +115,7 @@ In this section, we download a pretrained imagenet model and classify an image.
 
  .. code-block:: none
 
-    Downloading /workspace/.mxnet/models/resnet18_v1-a0666292.zip8263c9fb-cde9-4e2b-85c1-086c075c285c from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/resnet18_v1-a0666292.zip...
+    Downloading /workspace/.mxnet/models/resnet18_v1-a0666292.zipd59d0974-3f8a-4c88-80a5-58e86eaa5467 from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/resnet18_v1-a0666292.zip...
     x (1, 3, 224, 224)
 
 
diff --git a/docs/_sources/how_to/compile_models/from_oneflow.rst.txt b/docs/_sources/how_to/compile_models/from_oneflow.rst.txt
index b50a6b498..0b90ab199 100644
--- a/docs/_sources/how_to/compile_models/from_oneflow.rst.txt
+++ b/docs/_sources/how_to/compile_models/from_oneflow.rst.txt
@@ -113,7 +113,7 @@ Load a pretrained OneFlow model and save model
  .. code-block:: none
 
     Downloading: "https://oneflow-public.oss-cn-beijing.aliyuncs.com/model_zoo/flowvision/classification/ResNet/resnet18.zip" to /workspace/.oneflow/flowvision_cache/resnet18.zip
-
      0%|          | 0.00/41.5M [00:00<?, ?B/s]
     19%|#9        | 7.99M/41.5M [00:00<00:00, 55.4MB/s]
     35%|###4      | 14.3M/41.5M [00:00<00:00, 52.8MB/s]
     47%|####6     | 19.4M/41.5M [00:00<00:00, 45.0MB/s]
     58%|#####7    | 24.0M/41.5M [00:00<00:00, 41.0MB/s]
     82%|########2 | 34.1M/41.5M [00:00<00:00, 51.3MB/s]
    100%|##########| 41.5M/41.5M [00:00<00:00, 54.4MB/s]
+
      0%|          | 0.00/41.5M [00:00<?, ?B/s]
     19%|#9        | 7.99M/41.5M [00:00<00:00, 44.4MB/s]
     35%|###4      | 14.3M/41.5M [00:00<00:00, 48.0MB/s]
     46%|####5     | 19.0M/41.5M [00:00<00:00, 41.0MB/s]
     55%|#####5    | 22.9M/41.5M [00:00<00:00, 35.5MB/s]
     63%|######3   | 26.3M/41.5M [00:00<00:00, 34.0MB/s]
     77%|#######7  | 32.0M/41.5M [00:00<00:00, 40.2MB/s]
     92%|#########2| 38.3M/41.5M [00:01<00:00, 39.9MB/s]
    100%|##########| 41.5M/41.5M [00:01<00:00, 39.7MB/s]
 
 
 
diff --git a/docs/_sources/how_to/compile_models/from_pytorch.rst.txt b/docs/_sources/how_to/compile_models/from_pytorch.rst.txt
index aaf375f1e..6fad71149 100644
--- a/docs/_sources/how_to/compile_models/from_pytorch.rst.txt
+++ b/docs/_sources/how_to/compile_models/from_pytorch.rst.txt
@@ -94,7 +94,7 @@ Load a pretrained PyTorch model
  .. code-block:: none
 
     Downloading: "https://download.pytorch.org/models/resnet18-f37072fd.pth" to /workspace/.cache/torch/hub/checkpoints/resnet18-f37072fd.pth
-
      0%|          | 0.00/44.7M [00:00<?, ?B/s]
     49%|####9     | 22.0M/44.7M [00:00<00:00, 230MB/s]
    100%|##########| 44.7M/44.7M [00:00<00:00, 262MB/s]
+
      0%|          | 0.00/44.7M [00:00<?, ?B/s]
     46%|####6     | 20.7M/44.7M [00:00<00:00, 217MB/s]
     98%|#########8| 43.9M/44.7M [00:00<00:00, 232MB/s]
    100%|##########| 44.7M/44.7M [00:00<00:00, 230MB/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 968a2842b..5faabe37f 100644
--- a/docs/_sources/how_to/compile_models/from_tensorflow.rst.txt
+++ b/docs/_sources/how_to/compile_models/from_tensorflow.rst.txt
@@ -423,7 +423,7 @@ Run the corresponding model on tensorflow
 
 .. rst-class:: sphx-glr-timing
 
-   **Total running time of the script:** ( 1 minutes  2.917 seconds)
+   **Total running time of the script:** ( 1 minutes  3.019 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 d05b1c3c3..1f645bfd8 100644
--- a/docs/_sources/how_to/compile_models/sg_execution_times.rst.txt
+++ b/docs/_sources/how_to/compile_models/sg_execution_times.rst.txt
@@ -5,26 +5,26 @@
 
 Computation times
 =================
-**05:03.235** total execution time for **how_to_compile_models** files:
+**05:01.855** total execution time for **how_to_compile_models** files:
 
 +-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_darknet.py` (``from_darknet.py``)       | 01:03.722 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_darknet.py` (``from_darknet.py``)       | 01:03.919 | 0.0 MB |
 +-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_tensorflow.py` (``from_tensorflow.py``) | 01:02.917 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_tensorflow.py` (``from_tensorflow.py``) | 01:03.019 | 0.0 MB |
 +-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_paddle.py` (``from_paddle.py``)         | 00:40.483 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_paddle.py` (``from_paddle.py``)         | 00:39.316 | 0.0 MB |
 +-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_oneflow.py` (``from_oneflow.py``)       | 00:27.777 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_oneflow.py` (``from_oneflow.py``)       | 00:27.537 | 0.0 MB |
 +-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_mxnet.py` (``from_mxnet.py``)           | 00:24.542 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_mxnet.py` (``from_mxnet.py``)           | 00:24.488 | 0.0 MB |
 +-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_tflite.py` (``from_tflite.py``)         | 00:24.308 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_tflite.py` (``from_tflite.py``)         | 00:24.201 | 0.0 MB |
 +-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_coreml.py` (``from_coreml.py``)         | 00:22.364 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_coreml.py` (``from_coreml.py``)         | 00:22.408 | 0.0 MB |
 +-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_pytorch.py` (``from_pytorch.py``)       | 00:20.139 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_pytorch.py` (``from_pytorch.py``)       | 00:19.540 | 0.0 MB |
 +-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_keras.py` (``from_keras.py``)           | 00:14.587 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_keras.py` (``from_keras.py``)           | 00:15.148 | 0.0 MB |
 +-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_onnx.py` (``from_onnx.py``)             | 00:02.396 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_onnx.py` (``from_onnx.py``)             | 00:02.279 | 0.0 MB |
 +-----------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/how_to/deploy_models/deploy_model_on_android.rst.txt b/docs/_sources/how_to/deploy_models/deploy_model_on_android.rst.txt
index 9791e1f77..f34da57b0 100644
--- a/docs/_sources/how_to/deploy_models/deploy_model_on_android.rst.txt
+++ b/docs/_sources/how_to/deploy_models/deploy_model_on_android.rst.txt
@@ -441,7 +441,7 @@ Execute on TVM
     Evaluate inference time cost...
     Execution time summary:
      mean (ms)   median (ms)    max (ms)     min (ms)     std (ms)  
-      16.0766      16.0816      16.2853      15.8805       0.1172   
+      16.2731      16.1871      16.7614      15.8627       0.3202   
                
 
 
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 472ce893a..96d1115a1 100644
--- a/docs/_sources/how_to/deploy_models/deploy_object_detection_pytorch.rst.txt
+++ b/docs/_sources/how_to/deploy_models/deploy_object_detection_pytorch.rst.txt
@@ -123,7 +123,7 @@ Load pre-trained maskrcnn from torchvision and do tracing
  .. code-block:: none
 
     Downloading: "https://download.pytorch.org/models/maskrcnn_resnet50_fpn_coco-bf2d0c1e.pth" to /workspace/.cache/torch/hub/checkpoints/maskrcnn_resnet50_fpn_coco-bf2d0c1e.pth
-
      0%|          | 0.00/170M [00:00<?, ?B/s]
      8%|8         | 13.9M/170M [00:00<00:01, 143MB/s]
     16%|#6        | 27.6M/170M [00:00<00:01, 126MB/s]
     30%|###       | 51.2M/170M [00:00<00:00, 178MB/s]
     45%|####4     | 76.2M/170M [00:00<00:00, 209MB/s]
     58%|#####7    | 98.3M/170M [00:00<00:00, 217MB/s]
     72%|#######1  | 121M/170M [00:00<00:00, 226MB/s] 
     86%|########6 | 147M/170M [00:00<00:00, 238MB/s]
    100%|##########| 170M/170M [00:00<00:00, 219MB/s]
+
      0%|          | 0.00/170M [00:00<?, ?B/s]
      8%|8         | 14.4M/170M [00:00<00:01, 151MB/s]
     21%|##        | 35.2M/170M [00:00<00:00, 190MB/s]
     34%|###3      | 57.5M/170M [00:00<00:00, 210MB/s]
     47%|####6     | 79.4M/170M [00:00<00:00, 218MB/s]
     61%|######    | 103M/170M [00:00<00:00, 230MB/s] 
     74%|#######4  | 126M/170M [00:00<00:00, 234MB/s]
     88%|########7 | 149M/170M [00:00<00:00, 230MB/s]
    100%|##########| 170M/170M [00:00<00:00, 224MB/s]
     /usr/local/lib/python3.7/dist-packages/torch/nn/functional.py:3878: UserWarning: To copy construct from a tensor, it is recommended to use sourceTensor.clone().detach() or sourceTensor.clone().detach().requires_grad_(True), rather than torch.tensor(sourceTensor).
       for i in range(dim)
     /usr/local/lib/python3.7/dist-packages/torchvision/models/detection/anchor_utils.py:127: UserWarning: __floordiv__ is deprecated, and its behavior will change in a future version of pytorch. It currently rounds toward 0 (like the 'trunc' function NOT 'floor'). This results in incorrect rounding for negative values. To keep the current behavior, use torch.div(a, b, rounding_mode='trunc'), or for actual floor division, use torch.div(a, b, rounding_mode='floor').
@@ -292,7 +292,7 @@ Get boxes with score larger than 0.9
 
 .. rst-class:: sphx-glr-timing
 
-   **Total running time of the script:** ( 3 minutes  2.841 seconds)
+   **Total running time of the script:** ( 2 minutes  57.665 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 1749b4113..64d9c60a5 100644
--- a/docs/_sources/how_to/deploy_models/deploy_prequantized.rst.txt
+++ b/docs/_sources/how_to/deploy_models/deploy_prequantized.rst.txt
@@ -232,7 +232,7 @@ training. Other models require a full post training calibration.
  .. code-block:: none
 
     Downloading: "https://download.pytorch.org/models/mobilenet_v2-b0353104.pth" to /workspace/.cache/torch/hub/checkpoints/mobilenet_v2-b0353104.pth
-
      0%|          | 0.00/13.6M [00:00<?, ?B/s]
      0%|          | 40.0k/13.6M [00:00<00:48, 290kB/s]
      1%|          | 88.0k/13.6M [00:00<00:37, 373kB/s]
      2%|1         | 216k/13.6M [00:00<00:20, 687kB/s] 
      3%|3         | 464k/13.6M [00:00<00:10, 1.28MB/s]
      7%|6         | 968k/13.6M [00:00<00:05, 2.48MB/s]
     13%|#2        | 1.71M/13.6M [00:00<00:02, 4.20MB/s]
     21%|##        | 2.82M/13.6M [00:00<00:01, 6.48MB/s]
     28%|##7       | 3.78M/13.6M [00:00<00:01, 7.57MB/s]
     37%|###6      | 4.97M/13.6M [00:01<00:01, 8.84MB/s]
     47%|####7     | 6.39M/13.6M [00:01<00:00, 10.3MB/s]
     58%|#####7    | 7.82M/13.6M [00:01<00:00, 11.5MB/s]
     67%|######7   | 9.13M/13.6M [00:01<00:00, 12.2MB/s]
     76%|#######6  | 10.3M/13.6M [00:01<00:00, 11.9MB/s]
     84%|########4 | 11.5M/13.6M [00:01<00:00, 11.7MB/s]
     93%|#########2| 12.6M/13.6M [00:01<00:00, 11.4MB/s]
    100%|##########| 13.6M/13.6M [00:01<00:00, 8.40MB/s]
+
      0%|          | 0.00/13.6M [00:00<?, ?B/s]
    100%|##########| 13.6M/13.6M [00:00<00:00, 174MB/s]
 
 
 
@@ -412,7 +412,7 @@ Here we give an example of how to measure performance of TVM compiled models.
 
     Execution time summary:
      mean (ms)   median (ms)    max (ms)     min (ms)     std (ms)  
-      90.3297      90.2381      93.5955      90.1025       0.3959   
+      90.2922      90.2391      92.0132      90.0752       0.2443   
                
 
 
@@ -461,7 +461,7 @@ TODO
 
 .. rst-class:: sphx-glr-timing
 
-   **Total running time of the script:** ( 1 minutes  11.268 seconds)
+   **Total running time of the script:** ( 1 minutes  9.087 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 aa26810b3..c6e789022 100644
--- a/docs/_sources/how_to/deploy_models/deploy_prequantized_tflite.rst.txt
+++ b/docs/_sources/how_to/deploy_models/deploy_prequantized_tflite.rst.txt
@@ -439,7 +439,7 @@ Here we give an example of how to measure performance of TVM compiled models.
 
     Execution time summary:
      mean (ms)   median (ms)    max (ms)     min (ms)     std (ms)  
-      120.7101     120.6807     121.4579     120.0139      0.3205   
+      119.9342     119.9348     120.8328     119.1156      0.3515   
                
 
 
@@ -476,7 +476,7 @@ Here we give an example of how to measure performance of TVM compiled models.
 
 .. rst-class:: sphx-glr-timing
 
-   **Total running time of the script:** ( 1 minutes  59.003 seconds)
+   **Total running time of the script:** ( 1 minutes  58.670 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 1a40c9ab9..599d5f830 100644
--- a/docs/_sources/how_to/deploy_models/deploy_quantized.rst.txt
+++ b/docs/_sources/how_to/deploy_models/deploy_quantized.rst.txt
@@ -255,7 +255,7 @@ We create a Relay VM to build and execute the model.
 
 .. rst-class:: sphx-glr-timing
 
-   **Total running time of the script:** ( 1 minutes  32.419 seconds)
+   **Total running time of the script:** ( 1 minutes  35.140 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 3ecce22ee..8fc77afd9 100644
--- a/docs/_sources/how_to/deploy_models/deploy_ssd_gluoncv.rst.txt
+++ b/docs/_sources/how_to/deploy_models/deploy_ssd_gluoncv.rst.txt
@@ -158,7 +158,7 @@ Convert and compile model for CPU.
             data: None
       input_sym_arg_type = in_param.infer_type()[0]
     Downloading /workspace/.mxnet/models/ssd_512_resnet50_v1_voc-9c8b225a.zip from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/ssd_512_resnet50_v1_voc-9c8b225a.zip...
-
      0%|          | 0/132723 [00:00<?, ?KB/s]
      5%|4         | 6047/132723 [00:00<00:02, 60458.67KB/s]
     10%|#         | 13851/132723 [00:00<00:01, 70789.25KB/s]
     16%|#5        | 20930/132723 [00:00<00:02, 38392.67KB/s]
     22%|##1       | 28746/132723 [00:00<00:02, 48780.93KB/s]
     26%|##6       | 34846/132723 [00:00<00:01, 49733.35KB/s]
     32%|###2      | 42737/132723 [00:00<00:01, 57621.30KB/s]
     38%|###8      | 50567/132723 [00:00<00:01, 63381.71KB/s]
     44%|####4     | 58432/132723 [00:01<00:01, 67726.62KB/s]
     49%|####9     | 65663/132723 [00:01<00:01, 66884.73KB/s]
     55%|#####5    | 73245/132723 [00:01<00:00, 69435.86KB/s]
     61%|######1   | 81281/132723 [00:01<00:00, 72601.27KB/s]
     67%|######7   | 89135/132723 [00:01<00:00, 74339.15KB/s]
     73%|#######3  | 97001/132723 [00:01<00:00, 75609.88KB/s]
     79%|#######9  | 104914/132723 [00:01<00:00, 76650.79KB/s]
     85%|########4 | 112774/132723 [00:01<00:00, 77227.12KB/s]
     91%|#########
  | 120546/132723 [00:01<00:00, 75506.82KB/s]
     97%|#########7| 129011/132723 [00:01<00:00, 78189.14KB/s]
    100%|##########| 132723/132723 [00:01<00:00, 67463.49KB/s]
+
      0%|          | 0/132723 [00:00<?, ?KB/s]
      4%|4         | 5693/132723 [00:00<00:02, 56919.33KB/s]
     11%|#         | 14125/132723 [00:00<00:01, 73029.72KB/s]
     16%|#6        | 21428/132723 [00:00<00:02, 45469.78KB/s]
     22%|##2       | 29856/132723 [00:00<00:01, 56677.01KB/s]
     29%|##8       | 38277/132723 [00:00<00:01, 64713.77KB/s]
     35%|###5      | 46823/132723 [00:00<00:01, 70818.86KB/s]
     41%|####1     | 54516/132723 [00:00<00:01, 65459.78KB/s]
     47%|####7     | 63013/132723 [00:00<00:00, 70843.29KB/s]
     54%|#####3    | 71250/132723 [00:01<00:00, 74101.28KB/s]
     60%|######    | 79758/132723 [00:01<00:00, 77257.66KB/s]
     66%|######6   | 87722/132723 [00:01<00:00, 73035.84KB/s]
     73%|#######2  | 96231/132723 [00:01<00:00, 76412.74KB/s]
     78%|#######8  | 104042/132723 [00:01<00:00, 60262.14KB/s]
     85%|########4 | 112624/132723 [00:01<00:00, 66461.17KB/s]
     90%|######### | 119857/132723 [00:01<00:00, 63444.31KB/s]
     97%|########
 #6| 128272/132723 [00:01<00:00, 68732.98KB/s]
    100%|##########| 132723/132723 [00:01<00:00, 67320.76KB/s]
 
 
 
@@ -241,7 +241,7 @@ Display result
 
 .. rst-class:: sphx-glr-timing
 
-   **Total running time of the script:** ( 2 minutes  32.705 seconds)
+   **Total running time of the script:** ( 2 minutes  34.549 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 fe269eb64..b30479a4e 100644
--- a/docs/_sources/how_to/deploy_models/sg_execution_times.rst.txt
+++ b/docs/_sources/how_to/deploy_models/sg_execution_times.rst.txt
@@ -5,22 +5,22 @@
 
 Computation times
 =================
-**11:12.097** total execution time for **how_to_deploy_models** files:
+**11:07.409** total execution time for **how_to_deploy_models** files:
 
 +------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_object_detection_pytorch.py` (``deploy_object_detection_pytorch.py``) | 03:02.841 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_object_detection_pytorch.py` (``deploy_object_detection_pytorch.py``) | 02:57.665 | 0.0 MB |
 +------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_ssd_gluoncv.py` (``deploy_ssd_gluoncv.py``)                           | 02:32.705 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_ssd_gluoncv.py` (``deploy_ssd_gluoncv.py``)                           | 02:34.549 | 0.0 MB |
 +------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_prequantized_tflite.py` (``deploy_prequantized_tflite.py``)           | 01:59.003 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_prequantized_tflite.py` (``deploy_prequantized_tflite.py``)           | 01:58.670 | 0.0 MB |
 +------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_quantized.py` (``deploy_quantized.py``)                               | 01:32.419 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_quantized.py` (``deploy_quantized.py``)                               | 01:35.140 | 0.0 MB |
 +------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_prequantized.py` (``deploy_prequantized.py``)                         | 01:11.268 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_prequantized.py` (``deploy_prequantized.py``)                         | 01:09.087 | 0.0 MB |
 +------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_android.py` (``deploy_model_on_android.py``)                 | 00:30.519 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_android.py` (``deploy_model_on_android.py``)                 | 00:29.455 | 0.0 MB |
 +------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_rasp.py` (``deploy_model_on_rasp.py``)                       | 00:23.335 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_rasp.py` (``deploy_model_on_rasp.py``)                       | 00:22.838 | 0.0 MB |
 +------------------------------------------------------------------------------------------------------------------+-----------+--------+
 | :ref:`sphx_glr_how_to_deploy_models_deploy_sparse.py` (``deploy_sparse.py``)                                     | 00:00.006 | 0.0 MB |
 +------------------------------------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/how_to/extend_tvm/bring_your_own_datatypes.rst.txt b/docs/_sources/how_to/extend_tvm/bring_your_own_datatypes.rst.txt
index 5b2b8bde9..8626c91a8 100644
--- a/docs/_sources/how_to/extend_tvm/bring_your_own_datatypes.rst.txt
+++ b/docs/_sources/how_to/extend_tvm/bring_your_own_datatypes.rst.txt
@@ -476,7 +476,7 @@ First let us define two helper functions to get the mobilenet model and a cat im
 
  .. code-block:: none
 
-    Downloading /workspace/.mxnet/models/mobilenet0.25-9f83e440.zipcc65caf3-e88d-4049-b12f-b94bc5ee1995 from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/mobilenet0.25-9f83e440.zip...
+    Downloading /workspace/.mxnet/models/mobilenet0.25-9f83e440.zip34156840-1b6c-4cb3-ab83-81270d9ed63d 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 b68865bdf..edee625fe 100644
--- a/docs/_sources/how_to/extend_tvm/sg_execution_times.rst.txt
+++ b/docs/_sources/how_to/extend_tvm/sg_execution_times.rst.txt
@@ -5,14 +5,14 @@
 
 Computation times
 =================
-**00:40.059** total execution time for **how_to_extend_tvm** files:
+**00:40.453** total execution time for **how_to_extend_tvm** files:
 
 +-------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_extend_tvm_bring_your_own_datatypes.py` (``bring_your_own_datatypes.py``) | 00:36.917 | 0.0 MB |
+| :ref:`sphx_glr_how_to_extend_tvm_bring_your_own_datatypes.py` (``bring_your_own_datatypes.py``) | 00:37.260 | 0.0 MB |
 +-------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_extend_tvm_use_pass_instrument.py` (``use_pass_instrument.py``)           | 00:02.218 | 0.0 MB |
+| :ref:`sphx_glr_how_to_extend_tvm_use_pass_instrument.py` (``use_pass_instrument.py``)           | 00:02.250 | 0.0 MB |
 +-------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_extend_tvm_use_pass_infra.py` (``use_pass_infra.py``)                     | 00:00.916 | 0.0 MB |
+| :ref:`sphx_glr_how_to_extend_tvm_use_pass_infra.py` (``use_pass_infra.py``)                     | 00:00.934 | 0.0 MB |
 +-------------------------------------------------------------------------------------------------+-----------+--------+
 | :ref:`sphx_glr_how_to_extend_tvm_low_level_custom_pass.py` (``low_level_custom_pass.py``)       | 00:00.008 | 0.0 MB |
 +-------------------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/how_to/extend_tvm/use_pass_instrument.rst.txt b/docs/_sources/how_to/extend_tvm/use_pass_instrument.rst.txt
index cb76a3cab..196732e32 100644
--- a/docs/_sources/how_to/extend_tvm/use_pass_instrument.rst.txt
+++ b/docs/_sources/how_to/extend_tvm/use_pass_instrument.rst.txt
@@ -216,10 +216,10 @@ profile the execution time of each passes.
  .. code-block:: none
 
     Printing results of timing profile...
-    InferType: 6695us [6695us] (45.79%; 45.79%)
-    FoldScaleAxis: 7925us [5us] (54.21%; 54.21%)
-            FoldConstant: 7920us [1593us] (54.17%; 99.93%)
-                    InferType: 6327us [6327us] (43.28%; 79.89%)
+    InferType: 6504us [6504us] (45.81%; 45.81%)
+    FoldScaleAxis: 7694us [5us] (54.19%; 54.19%)
+            FoldConstant: 7688us [1584us] (54.15%; 99.93%)
+                    InferType: 6104us [6104us] (42.99%; 79.39%)
 
 
 
@@ -258,10 +258,10 @@ Refer to following sections and :py:func:`tvm.instrument.pass_instrument` for th
  .. code-block:: none
 
     Printing results of timing profile...
-    InferType: 6308us [6308us] (45.02%; 45.02%)
-    FoldScaleAxis: 7705us [5us] (54.98%; 54.98%)
-            FoldConstant: 7700us [1587us] (54.95%; 99.94%)
-                    InferType: 6113us [6113us] (43.62%; 79.39%)
+    InferType: 6124us [6124us] (44.68%; 44.68%)
+    FoldScaleAxis: 7584us [4us] (55.32%; 55.32%)
+            FoldConstant: 7579us [1566us] (55.29%; 99.94%)
+                    InferType: 6013us [6013us] (43.87%; 79.33%)
 
 
 
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 ec8dedaac..f6e654eb1 100644
--- a/docs/_sources/how_to/optimize_operators/opt_conv_cuda.rst.txt
+++ b/docs/_sources/how_to/optimize_operators/opt_conv_cuda.rst.txt
@@ -340,7 +340,7 @@ latency of convolution.
 
  .. code-block:: none
 
-    Convolution: 54.160442 ms
+    Convolution: 49.836182 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 bbaca8599..0d1b0dba2 100644
--- a/docs/_sources/how_to/optimize_operators/opt_conv_tensorcore.rst.txt
+++ b/docs/_sources/how_to/optimize_operators/opt_conv_tensorcore.rst.txt
@@ -671,7 +671,7 @@ be able to run on our build server
 
  .. code-block:: none
 
-    conv2d with tensor core: 9.090662 ms
+    conv2d with tensor core: 6.863631 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 c9205ad73..76fdac0ac 100644
--- a/docs/_sources/how_to/optimize_operators/opt_gemm.rst.txt
+++ b/docs/_sources/how_to/optimize_operators/opt_gemm.rst.txt
@@ -143,8 +143,8 @@ Then we write a baseline implementation, the simplest way to write a matrix mult
 
  .. code-block:: none
 
-    Numpy running time: 0.018934
-    Baseline: 3.386234
+    Numpy running time: 0.018806
+    Baseline: 3.216682
 
 
 
@@ -239,7 +239,7 @@ fill 32 * 32 * sizeof(float) which is 4KB in the cache whose total size is 32KB
 
  .. code-block:: none
 
-    Opt1: 0.289869
+    Opt1: 0.299593
 
 
 
@@ -342,7 +342,7 @@ In this tutorial, we chose to vectorize the inner loop row data since it is cach
 
  .. code-block:: none
 
-    Opt2: 0.341522
+    Opt2: 0.343425
 
 
 
@@ -438,7 +438,7 @@ the access pattern for A matrix is more cache friendly.
 
  .. code-block:: none
 
-    Opt3: 0.120605
+    Opt3: 0.114249
 
 
 
@@ -563,7 +563,7 @@ flattening.
 
  .. code-block:: none
 
-    Opt4: 0.110875
+    Opt4: 0.109088
 
 
 
@@ -685,7 +685,7 @@ write to C when all the block results are ready.
 
  .. code-block:: none
 
-    Opt5: 0.111340
+    Opt5: 0.110632
 
 
 
@@ -810,7 +810,7 @@ Futhermore, we can also utilize multi-core processors to do the thread-level par
 
  .. code-block:: none
 
-    Opt6: 0.144972
+    Opt6: 0.144890
 
 
 
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 883261c5e..79e339bbf 100644
--- a/docs/_sources/how_to/optimize_operators/sg_execution_times.rst.txt
+++ b/docs/_sources/how_to/optimize_operators/sg_execution_times.rst.txt
@@ -5,12 +5,12 @@
 
 Computation times
 =================
-**00:34.419** total execution time for **how_to_optimize_operators** files:
+**00:33.887** total execution time for **how_to_optimize_operators** files:
 
 +-----------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_optimize_operators_opt_gemm.py` (``opt_gemm.py``)                       | 00:32.168 | 0.0 MB |
+| :ref:`sphx_glr_how_to_optimize_operators_opt_gemm.py` (``opt_gemm.py``)                       | 00:31.654 | 0.0 MB |
 +-----------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_optimize_operators_opt_conv_tensorcore.py` (``opt_conv_tensorcore.py``) | 00:01.253 | 0.0 MB |
+| :ref:`sphx_glr_how_to_optimize_operators_opt_conv_tensorcore.py` (``opt_conv_tensorcore.py``) | 00:01.183 | 0.0 MB |
 +-----------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_optimize_operators_opt_conv_cuda.py` (``opt_conv_cuda.py``)             | 00:00.997 | 0.0 MB |
+| :ref:`sphx_glr_how_to_optimize_operators_opt_conv_cuda.py` (``opt_conv_cuda.py``)             | 00:01.050 | 0.0 MB |
 +-----------------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/how_to/tune_with_autoscheduler/sg_execution_times.rst.txt b/docs/_sources/how_to/tune_with_autoscheduler/sg_execution_times.rst.txt
index b611d5133..31bb66c02 100644
--- a/docs/_sources/how_to/tune_with_autoscheduler/sg_execution_times.rst.txt
+++ b/docs/_sources/how_to/tune_with_autoscheduler/sg_execution_times.rst.txt
@@ -5,18 +5,18 @@
 
 Computation times
 =================
-**05:54.523** total execution time for **how_to_tune_with_autoscheduler** files:
+**05:56.604** total execution time for **how_to_tune_with_autoscheduler** files:
 
 +----------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_conv2d_layer_cuda.py` (``tune_conv2d_layer_cuda.py``) | 03:11.157 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_conv2d_layer_cuda.py` (``tune_conv2d_layer_cuda.py``) | 03:10.386 | 0.0 MB |
 +----------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_x86.py` (``tune_network_x86.py``)             | 01:21.967 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_x86.py` (``tune_network_x86.py``)             | 01:22.515 | 0.0 MB |
 +----------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_cuda.py` (``tune_network_cuda.py``)           | 00:45.629 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_cuda.py` (``tune_network_cuda.py``)           | 00:45.653 | 0.0 MB |
 +----------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_sparse_x86.py` (``tune_sparse_x86.py``)               | 00:18.431 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_sparse_x86.py` (``tune_sparse_x86.py``)               | 00:20.490 | 0.0 MB |
 +----------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_mali.py` (``tune_network_mali.py``)           | 00:08.821 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_mali.py` (``tune_network_mali.py``)           | 00:08.951 | 0.0 MB |
 +----------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_arm.py` (``tune_network_arm.py``)             | 00:08.518 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_arm.py` (``tune_network_arm.py``)             | 00:08.609 | 0.0 MB |
 +----------------------------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/how_to/tune_with_autoscheduler/tune_conv2d_layer_cuda.rst.txt b/docs/_sources/how_to/tune_with_autoscheduler/tune_conv2d_layer_cuda.rst.txt
index dd6dc2bc2..fe2529176 100644
--- a/docs/_sources/how_to/tune_with_autoscheduler/tune_conv2d_layer_cuda.rst.txt
+++ b/docs/_sources/how_to/tune_with_autoscheduler/tune_conv2d_layer_cuda.rst.txt
@@ -240,699 +240,485 @@ cooperative fetching, unrolling and operator fusion.
                  compute: Buffer(compute_2: Pointer(float32), float32, [25088], [])}
       buffer_map = {data_1: data, kernel_1: kernel, bias_1: bias, compute_1: compute}
       preflattened_buffer_map = {data_1: data_3: Buffer(data_2, float32, [1, 512, 7, 7], []), kernel_1: kernel_3: Buffer(kernel_2, float32, [512, 512, 3, 3], []), bias_1: bias_3: Buffer(bias_2, float32, [1, 512, 1, 1], []), compute_1: compute_3: Buffer(compute_2, float32, [1, 512, 7, 7], [])} {
-      attr [IterVar(blockIdx.x: int32, (nullptr), "ThreadIndex", "blockIdx.x")] "thread_extent" = 112;
-      allocate(conv2d_nchw: Pointer(local float32), float32, [14]), storage_scope = local;
-      allocate(pad_temp.shared: Pointer(shared float32), float32, [108]), storage_scope = shared;
-      allocate(kernel.shared: Pointer(shared float32), float32, [1152]), storage_scope = shared;
-      attr [IterVar(threadIdx.x: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16 {
-        conv2d_nchw_1: Buffer(conv2d_nchw, float32, [14], [], scope="local", align=32)[0] = 0f32
-        conv2d_nchw_1[1] = 0f32
+      attr [IterVar(blockIdx.x: int32, (nullptr), "ThreadIndex", "blockIdx.x")] "thread_extent" = 128;
+      allocate(conv2d_nchw: Pointer(local float32), float32, [28]), storage_scope = local;
+      allocate(pad_temp.shared: Pointer(shared float32), float32, [252]), storage_scope = shared;
+      allocate(kernel.shared: Pointer(shared float32), float32, [48]), storage_scope = shared;
+      attr [IterVar(threadIdx.x: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7 {
+        conv2d_nchw_1: Buffer(conv2d_nchw, float32, [4], [], scope="local", align=8)[0] = 0f32
         conv2d_nchw_1[2] = 0f32
-        conv2d_nchw_1[3] = 0f32
         conv2d_nchw_1[4] = 0f32
-        conv2d_nchw_1[5] = 0f32
         conv2d_nchw_1[6] = 0f32
-        conv2d_nchw_1[7] = 0f32
         conv2d_nchw_1[8] = 0f32
-        conv2d_nchw_1[9] = 0f32
         conv2d_nchw_1[10] = 0f32
-        conv2d_nchw_1[11] = 0f32
         conv2d_nchw_1[12] = 0f32
+        conv2d_nchw_1[14] = 0f32
+        conv2d_nchw_1[16] = 0f32
+        conv2d_nchw_1[18] = 0f32
+        conv2d_nchw_1[20] = 0f32
+        conv2d_nchw_1[22] = 0f32
+        conv2d_nchw_1[24] = 0f32
+        conv2d_nchw_1[26] = 0f32
+        conv2d_nchw_1[1] = 0f32
+        conv2d_nchw_1[3] = 0f32
+        conv2d_nchw_1[5] = 0f32
+        conv2d_nchw_1[7] = 0f32
+        conv2d_nchw_1[9] = 0f32
+        conv2d_nchw_1[11] = 0f32
         conv2d_nchw_1[13] = 0f32
+        conv2d_nchw_1[15] = 0f32
+        conv2d_nchw_1[17] = 0f32
+        conv2d_nchw_1[19] = 0f32
+        conv2d_nchw_1[21] = 0f32
+        conv2d_nchw_1[23] = 0f32
+        conv2d_nchw_1[25] = 0f32
+        conv2d_nchw_1[27] = 0f32
         for (rc.outer.outer: int32, 0, 128) {
-          let cse_var_2: int32 = (rc.outer.outer*196)
-          let cse_var_1: int32 = (rc.outer.outer*36)
-           {
-            attr [IterVar(threadIdx.x_1: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            pad_temp.shared_1: Buffer(pad_temp.shared, float32, [108], [], scope="shared")[threadIdx.x_1] = @tir.if_then_else((((3 <= threadIdx.x_1) && (1 <= (floormod(blockIdx.x, 7) + floormod(threadIdx.x_1, 3)))) && ((floormod(blockIdx.x, 7) + floormod(threadIdx.x_1, 3)) < 8)), data[((((cse_var_2 + (floordiv(threadIdx.x_1, 3)*7)) + floormod(blockIdx.x, 7)) + floormod(threadIdx.x_1, 3)) - 8)], 0f32, dtype=float32)
-            attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            pad_temp.shared_1[(threadIdx.x_1 + 16)] = @tir.if_then_else(((((3 <= floormod((threadIdx.x_1 + 16), 27)) && (floormod((threadIdx.x_1 + 16), 27) < 24)) && (1 <= (floormod(blockIdx.x, 7) + floormod((threadIdx.x_1 + 1), 3)))) && ((floormod(blockIdx.x, 7) + floormod((threadIdx.x_1 + 1), 3)) < 8)), data[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 16), 27)*49)) + (floordiv(floormod((threadIdx.x_1 + 16), 27), 3)*7)) + floormod(blockIdx.x, 7)) + floormod((threadIdx.x_1 + 1), 3)) - 8) [...]
-            attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            pad_temp.shared_1[(threadIdx.x_1 + 32)] = @tir.if_then_else(((1 <= (floormod(blockIdx.x, 7) + floormod((threadIdx.x_1 + 2), 3))) && ((floormod(blockIdx.x, 7) + floormod((threadIdx.x_1 + 2), 3)) < 8)), data[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 32), 27)*49)) + (floordiv(floormod((threadIdx.x_1 + 5), 27), 3)*7)) + floormod(blockIdx.x, 7)) + floormod((threadIdx.x_1 + 2), 3)) - 8)], 0f32, dtype=float32)
-            attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            pad_temp.shared_1[(threadIdx.x_1 + 48)] = @tir.if_then_else(((((1 <= floormod((floordiv(threadIdx.x_1, 3) + 7), 9)) && (floormod((threadIdx.x_1 + 21), 27) < 24)) && (1 <= (floormod(blockIdx.x, 7) + floormod(threadIdx.x_1, 3)))) && ((floormod(blockIdx.x, 7) + floormod(threadIdx.x_1, 3)) < 8)), data[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 48), 27)*49)) + (floormod((floordiv(threadIdx.x_1, 3) + 7), 9)*7)) + floormod(blockIdx.x, 7)) + floormod(threadIdx.x_1, 3)) - 8)], 0f32,  [...]
-            attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            pad_temp.shared_1[(threadIdx.x_1 + 64)] = @tir.if_then_else((((threadIdx.x_1 < 14) && (1 <= (floormod(blockIdx.x, 7) + floormod((threadIdx.x_1 + 1), 3)))) && ((floormod(blockIdx.x, 7) + floormod((threadIdx.x_1 + 1), 3)) < 8)), data[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 64), 27)*49)) + (floordiv(floormod((threadIdx.x_1 + 10), 27), 3)*7)) + floormod(blockIdx.x, 7)) + floormod((threadIdx.x_1 + 1), 3)) - 8)], 0f32, dtype=float32)
-            attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            pad_temp.shared_1[(threadIdx.x_1 + 80)] = @tir.if_then_else(((((3 <= floormod((threadIdx.x_1 + 26), 27)) && (floormod((threadIdx.x_1 + 26), 27) < 24)) && (1 <= (floormod(blockIdx.x, 7) + floormod((threadIdx.x_1 + 2), 3)))) && ((floormod(blockIdx.x, 7) + floormod((threadIdx.x_1 + 2), 3)) < 8)), data[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 80), 27)*49)) + (floordiv(floormod((threadIdx.x_1 + 26), 27), 3)*7)) + floormod(blockIdx.x, 7)) + floormod((threadIdx.x_1 + 2), 3)) - 8) [...]
-            attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            if @tir.likely((threadIdx.x_1 < 12), dtype=bool) {
-              pad_temp.shared_1[(threadIdx.x_1 + 96)] = @tir.if_then_else((((threadIdx.x_1 < 9) && (1 <= (floormod(blockIdx.x, 7) + floormod(threadIdx.x_1, 3)))) && ((floormod(blockIdx.x, 7) + floormod(threadIdx.x_1, 3)) < 8)), data[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 96), 27)*49)) + ((floordiv(threadIdx.x_1, 3) + 5)*7)) + floormod(blockIdx.x, 7)) + floormod(threadIdx.x_1, 3)) - 8)], 0f32, dtype=float32)
+          for (rx.outer.outer: int32, 0, 3) {
+            let cse_var_1: int32 = (rc.outer.outer*36)
+             {
+              attr [IterVar(threadIdx.x_1: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              pad_temp.shared_1: Buffer(pad_temp.shared, float32, [252], [], scope="shared")[threadIdx.x_1] = 0f32
+              attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              pad_temp.shared_1[(threadIdx.x_1 + 7)] = @tir.if_then_else(((1 <= (threadIdx.x_1 + rx.outer.outer)) && ((threadIdx.x_1 + rx.outer.outer) < 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) - 1)], 0f32, dtype=float32)
+              attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              pad_temp.shared_1[(threadIdx.x_1 + 14)] = @tir.if_then_else(((1 <= (threadIdx.x_1 + rx.outer.outer)) && ((threadIdx.x_1 + rx.outer.outer) < 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 6)], 0f32, dtype=float32)
+              attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              pad_temp.shared_1[(threadIdx.x_1 + 21)] = @tir.if_then_else(((1 <= (threadIdx.x_1 + rx.outer.outer)) && ((threadIdx.x_1 + rx.outer.outer) < 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 13)], 0f32, dtype=float32)
+              attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              pad_temp.shared_1[(threadIdx.x_1 + 28)] = @tir.if_then_else(((1 <= (threadIdx.x_1 + rx.outer.outer)) && ((threadIdx.x_1 + rx.outer.outer) < 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 20)], 0f32, dtype=float32)
+              attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              pad_temp.shared_1[(threadIdx.x_1 + 35)] = @tir.if_then_else(((1 <= (threadIdx.x_1 + rx.outer.outer)) && ((threadIdx.x_1 + rx.outer.outer) < 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 27)], 0f32, dtype=float32)
+              attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              pad_temp.shared_1[(threadIdx.x_1 + 42)] = @tir.if_then_else(((1 <= (threadIdx.x_1 + rx.outer.outer)) && ((threadIdx.x_1 + rx.outer.outer) < 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 34)], 0f32, dtype=float32)
+              attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              pad_temp.shared_1[(threadIdx.x_1 + 49)] = @tir.if_then_else(((1 <= (threadIdx.x_1 + rx.outer.outer)) && ((threadIdx.x_1 + rx.outer.outer) < 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 41)], 0f32, dtype=float32)
+              attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              pad_temp.shared_1[(threadIdx.x_1 + 56)] = 0f32
+              attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              pad_temp.shared_1[(threadIdx.x_1 + 63)] = 0f32
+              attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              pad_temp.shared_1[(threadIdx.x_1 + 70)] = @tir.if_then_else(((1 <= (threadIdx.x_1 + rx.outer.outer)) && ((threadIdx.x_1 + rx.outer.outer) < 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 48)], 0f32, dtype=float32)
+              attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              pad_temp.shared_1[(threadIdx.x_1 + 77)] = @tir.if_then_else(((1 <= (threadIdx.x_1 + rx.outer.outer)) && ((threadIdx.x_1 + rx.outer.outer) < 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 55)], 0f32, dtype=float32)
+              attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              pad_temp.shared_1[(threadIdx.x_1 + 84)] = @tir.if_then_else(((1 <= (threadIdx.x_1 + rx.outer.outer)) && ((threadIdx.x_1 + rx.outer.outer) < 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 62)], 0f32, dtype=float32)
+              attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              pad_temp.shared_1[(threadIdx.x_1 + 91)] = @tir.if_then_else(((1 <= (threadIdx.x_1 + rx.outer.outer)) && ((threadIdx.x_1 + rx.outer.outer) < 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 69)], 0f32, dtype=float32)
+              attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              pad_temp.shared_1[(threadIdx.x_1 + 98)] = @tir.if_then_else(((1 <= (threadIdx.x_1 + rx.outer.outer)) && ((threadIdx.x_1 + rx.outer.outer) < 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 76)], 0f32, dtype=float32)
+              attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              pad_temp.shared_1[(threadIdx.x_1 + 105)] = @tir.if_then_else(((1 <= (threadIdx.x_1 + rx.outer.outer)) && ((threadIdx.x_1 + rx.outer.outer) < 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 83)], 0f32, dtype=float32)
+              attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              pad_temp.shared_1[(threadIdx.x_1 + 112)] = @tir.if_then_else(((1 <= (threadIdx.x_1 + rx.outer.outer)) && ((threadIdx.x_1 + rx.outer.outer) < 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 90)], 0f32, dtype=float32)
+              attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              pad_temp.shared_1[(threadIdx.x_1 + 119)] = 0f32
+              attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              pad_temp.shared_1[(threadIdx.x_1 + 126)] = 0f32
+              attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              pad_temp.shared_1[(threadIdx.x_1 + 133)] = @tir.if_then_else(((1 <= (threadIdx.x_1 + rx.outer.outer)) && ((threadIdx.x_1 + rx.outer.outer) < 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 97)], 0f32, dtype=float32)
+              attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              pad_temp.shared_1[(threadIdx.x_1 + 140)] = @tir.if_then_else(((1 <= (threadIdx.x_1 + rx.outer.outer)) && ((threadIdx.x_1 + rx.outer.outer) < 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 104)], 0f32, dtype=float32)
+              attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              pad_temp.shared_1[(threadIdx.x_1 + 147)] = @tir.if_then_else(((1 <= (threadIdx.x_1 + rx.outer.outer)) && ((threadIdx.x_1 + rx.outer.outer) < 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 111)], 0f32, dtype=float32)
+              attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              pad_temp.shared_1[(threadIdx.x_1 + 154)] = @tir.if_then_else(((1 <= (threadIdx.x_1 + rx.outer.outer)) && ((threadIdx.x_1 + rx.outer.outer) < 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 118)], 0f32, dtype=float32)
+              attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              pad_temp.shared_1[(threadIdx.x_1 + 161)] = @tir.if_then_else(((1 <= (threadIdx.x_1 + rx.outer.outer)) && ((threadIdx.x_1 + rx.outer.outer) < 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 125)], 0f32, dtype=float32)
+              attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              pad_temp.shared_1[(threadIdx.x_1 + 168)] = @tir.if_then_else(((1 <= (threadIdx.x_1 + rx.outer.outer)) && ((threadIdx.x_1 + rx.outer.outer) < 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 132)], 0f32, dtype=float32)
+              attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              pad_temp.shared_1[(threadIdx.x_1 + 175)] = @tir.if_then_else(((1 <= (threadIdx.x_1 + rx.outer.outer)) && ((threadIdx.x_1 + rx.outer.outer) < 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 139)], 0f32, dtype=float32)
+              attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              pad_temp.shared_1[(threadIdx.x_1 + 182)] = 0f32
+              attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              pad_temp.shared_1[(threadIdx.x_1 + 189)] = 0f32
+              attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              pad_temp.shared_1[(threadIdx.x_1 + 196)] = @tir.if_then_else(((1 <= (threadIdx.x_1 + rx.outer.outer)) && ((threadIdx.x_1 + rx.outer.outer) < 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 146)], 0f32, dtype=float32)
+              attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              pad_temp.shared_1[(threadIdx.x_1 + 203)] = @tir.if_then_else(((1 <= (threadIdx.x_1 + rx.outer.outer)) && ((threadIdx.x_1 + rx.outer.outer) < 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 153)], 0f32, dtype=float32)
+              attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              pad_temp.shared_1[(threadIdx.x_1 + 210)] = @tir.if_then_else(((1 <= (threadIdx.x_1 + rx.outer.outer)) && ((threadIdx.x_1 + rx.outer.outer) < 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 160)], 0f32, dtype=float32)
+              attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              pad_temp.shared_1[(threadIdx.x_1 + 217)] = @tir.if_then_else(((1 <= (threadIdx.x_1 + rx.outer.outer)) && ((threadIdx.x_1 + rx.outer.outer) < 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 167)], 0f32, dtype=float32)
+              attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              pad_temp.shared_1[(threadIdx.x_1 + 224)] = @tir.if_then_else(((1 <= (threadIdx.x_1 + rx.outer.outer)) && ((threadIdx.x_1 + rx.outer.outer) < 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 174)], 0f32, dtype=float32)
+              attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              pad_temp.shared_1[(threadIdx.x_1 + 231)] = @tir.if_then_else(((1 <= (threadIdx.x_1 + rx.outer.outer)) && ((threadIdx.x_1 + rx.outer.outer) < 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 181)], 0f32, dtype=float32)
+              attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              pad_temp.shared_1[(threadIdx.x_1 + 238)] = @tir.if_then_else(((1 <= (threadIdx.x_1 + rx.outer.outer)) && ((threadIdx.x_1 + rx.outer.outer) < 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 188)], 0f32, dtype=float32)
+              attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              pad_temp.shared_1[(threadIdx.x_1 + 245)] = 0f32
+              attr [IterVar(threadIdx.x_2: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              kernel.shared_1: Buffer(kernel.shared, float32, [48], [], scope="shared")[threadIdx.x_2] = kernel[((((blockIdx.x*18432) + cse_var_1) + (threadIdx.x_2*3)) + rx.outer.outer)]
+              attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              kernel.shared_1[(threadIdx.x_2 + 7)] = kernel[((((((blockIdx.x*18432) + (floordiv((threadIdx.x_2 + 7), 12)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 7), 12), 3)*9)) + (floormod((threadIdx.x_2 + 1), 3)*3)) + rx.outer.outer)]
+              attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              kernel.shared_1[(threadIdx.x_2 + 14)] = kernel[(((((blockIdx.x*18432) + (floordiv((threadIdx.x_2 + 14), 12)*4608)) + cse_var_1) + (floormod((threadIdx.x_2 + 2), 12)*3)) + rx.outer.outer)]
+              attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              kernel.shared_1[(threadIdx.x_2 + 21)] = kernel[((((((blockIdx.x*18432) + (floordiv((threadIdx.x_2 + 21), 12)*4608)) + cse_var_1) + (floormod((floordiv(threadIdx.x_2, 3) + 3), 4)*9)) + (floormod(threadIdx.x_2, 3)*3)) + rx.outer.outer)]
+              attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              kernel.shared_1[(threadIdx.x_2 + 28)] = kernel[((((((blockIdx.x*18432) + (floordiv((threadIdx.x_2 + 28), 12)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 4), 12), 3)*9)) + (floormod((threadIdx.x_2 + 1), 3)*3)) + rx.outer.outer)]
+              attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              kernel.shared_1[(threadIdx.x_2 + 35)] = kernel[((((((blockIdx.x*18432) + (floordiv((threadIdx.x_2 + 35), 12)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 11), 12), 3)*9)) + (floormod((threadIdx.x_2 + 2), 3)*3)) + rx.outer.outer)]
+              attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 7;
+              if @tir.likely((threadIdx.x_2 < 6), dtype=bool) {
+                kernel.shared_1[(threadIdx.x_2 + 42)] = kernel[((((((blockIdx.x*18432) + (floordiv((threadIdx.x_2 + 42), 12)*4608)) + cse_var_1) + ((floordiv(threadIdx.x_2, 3) + 2)*9)) + (floormod(threadIdx.x_2, 3)*3)) + rx.outer.outer)]
+              }
+              conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[threadIdx.x]*kernel.shared_1[0]))
+              conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(threadIdx.x + 7)]*kernel.shared_1[0]))
+              conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(threadIdx.x + 14)]*kernel.shared_1[0]))
+              conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(threadIdx.x + 21)]*kernel.shared_1[0]))
+              conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(threadIdx.x + 28)]*kernel.shared_1[0]))
+              conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(threadIdx.x + 35)]*kernel.shared_1[0]))
+              conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(threadIdx.x + 42)]*kernel.shared_1[0]))
+              conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[threadIdx.x]*kernel.shared_1[24]))
+              conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(threadIdx.x + 7)]*kernel.shared_1[24]))
+              conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(threadIdx.x + 14)]*kernel.shared_1[24]))
+              conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(threadIdx.x + 21)]*kernel.shared_1[24]))
+              conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(threadIdx.x + 28)]*kernel.shared_1[24]))
+              conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(threadIdx.x + 35)]*kernel.shared_1[24]))
+              conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(threadIdx.x + 42)]*kernel.shared_1[24]))
+              conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[threadIdx.x]*kernel.shared_1[12]))
+              conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(threadIdx.x + 7)]*kernel.shared_1[12]))
+              conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(threadIdx.x + 14)]*kernel.shared_1[12]))
+              conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(threadIdx.x + 21)]*kernel.shared_1[12]))
+              conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(threadIdx.x + 28)]*kernel.shared_1[12]))
+              conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(threadIdx.x + 35)]*kernel.shared_1[12]))
+              conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(threadIdx.x + 42)]*kernel.shared_1[12]))
+              conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[threadIdx.x]*kernel.shared_1[36]))
+              conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(threadIdx.x + 7)]*kernel.shared_1[36]))
+              conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(threadIdx.x + 14)]*kernel.shared_1[36]))
+              conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(threadIdx.x + 21)]*kernel.shared_1[36]))
+              conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(threadIdx.x + 28)]*kernel.shared_1[36]))
+              conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(threadIdx.x + 35)]*kernel.shared_1[36]))
+              conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(threadIdx.x + 42)]*kernel.shared_1[36]))
+              conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(threadIdx.x + 7)]*kernel.shared_1[1]))
+              conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(threadIdx.x + 14)]*kernel.shared_1[1]))
+              conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(threadIdx.x + 21)]*kernel.shared_1[1]))
+              conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(threadIdx.x + 28)]*kernel.shared_1[1]))
+              conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(threadIdx.x + 35)]*kernel.shared_1[1]))
+              conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(threadIdx.x + 42)]*kernel.shared_1[1]))
+              conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(threadIdx.x + 49)]*kernel.shared_1[1]))
+              conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(threadIdx.x + 7)]*kernel.shared_1[25]))
+              conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(threadIdx.x + 14)]*kernel.shared_1[25]))
+              conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(threadIdx.x + 21)]*kernel.shared_1[25]))
+              conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(threadIdx.x + 28)]*kernel.shared_1[25]))
+              conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(threadIdx.x + 35)]*kernel.shared_1[25]))
+              conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(threadIdx.x + 42)]*kernel.shared_1[25]))
+              conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(threadIdx.x + 49)]*kernel.shared_1[25]))
+              conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(threadIdx.x + 7)]*kernel.shared_1[13]))
+              conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(threadIdx.x + 14)]*kernel.shared_1[13]))
+              conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(threadIdx.x + 21)]*kernel.shared_1[13]))
+              conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(threadIdx.x + 28)]*kernel.shared_1[13]))
+              conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(threadIdx.x + 35)]*kernel.shared_1[13]))
+              conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(threadIdx.x + 42)]*kernel.shared_1[13]))
+              conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(threadIdx.x + 49)]*kernel.shared_1[13]))
+              conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(threadIdx.x + 7)]*kernel.shared_1[37]))
+              conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(threadIdx.x + 14)]*kernel.shared_1[37]))
+              conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(threadIdx.x + 21)]*kernel.shared_1[37]))
+              conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(threadIdx.x + 28)]*kernel.shared_1[37]))
+              conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(threadIdx.x + 35)]*kernel.shared_1[37]))
+              conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(threadIdx.x + 42)]*kernel.shared_1[37]))
+              conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(threadIdx.x + 49)]*kernel.shared_1[37]))
+              conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(threadIdx.x + 14)]*kernel.shared_1[2]))
+              conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(threadIdx.x + 21)]*kernel.shared_1[2]))
+              conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(threadIdx.x + 28)]*kernel.shared_1[2]))
+              conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(threadIdx.x + 35)]*kernel.shared_1[2]))
+              conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(threadIdx.x + 42)]*kernel.shared_1[2]))
+              conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(threadIdx.x + 49)]*kernel.shared_1[2]))
+              conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(threadIdx.x + 56)]*kernel.shared_1[2]))
+              conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(threadIdx.x + 14)]*kernel.shared_1[26]))
+              conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(threadIdx.x + 21)]*kernel.shared_1[26]))
+              conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(threadIdx.x + 28)]*kernel.shared_1[26]))
+              conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(threadIdx.x + 35)]*kernel.shared_1[26]))
+              conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(threadIdx.x + 42)]*kernel.shared_1[26]))
+              conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(threadIdx.x + 49)]*kernel.shared_1[26]))
+              conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(threadIdx.x + 56)]*kernel.shared_1[26]))
+              conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(threadIdx.x + 14)]*kernel.shared_1[14]))
+              conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(threadIdx.x + 21)]*kernel.shared_1[14]))
+              conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(threadIdx.x + 28)]*kernel.shared_1[14]))
+              conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(threadIdx.x + 35)]*kernel.shared_1[14]))
+              conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(threadIdx.x + 42)]*kernel.shared_1[14]))
+              conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(threadIdx.x + 49)]*kernel.shared_1[14]))
+              conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(threadIdx.x + 56)]*kernel.shared_1[14]))
+              conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(threadIdx.x + 14)]*kernel.shared_1[38]))
+              conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(threadIdx.x + 21)]*kernel.shared_1[38]))
+              conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(threadIdx.x + 28)]*kernel.shared_1[38]))
+              conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(threadIdx.x + 35)]*kernel.shared_1[38]))
+              conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(threadIdx.x + 42)]*kernel.shared_1[38]))
+              conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(threadIdx.x + 49)]*kernel.shared_1[38]))
+              conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(threadIdx.x + 56)]*kernel.shared_1[38]))
+              conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(threadIdx.x + 63)]*kernel.shared_1[3]))
+              conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(threadIdx.x + 70)]*kernel.shared_1[3]))
+              conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(threadIdx.x + 77)]*kernel.shared_1[3]))
+              conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(threadIdx.x + 84)]*kernel.shared_1[3]))
+              conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(threadIdx.x + 91)]*kernel.shared_1[3]))
+              conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(threadIdx.x + 98)]*kernel.shared_1[3]))
+              conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(threadIdx.x + 105)]*kernel.shared_1[3]))
+              conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(threadIdx.x + 63)]*kernel.shared_1[27]))
+              conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(threadIdx.x + 70)]*kernel.shared_1[27]))
+              conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(threadIdx.x + 77)]*kernel.shared_1[27]))
+              conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(threadIdx.x + 84)]*kernel.shared_1[27]))
+              conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(threadIdx.x + 91)]*kernel.shared_1[27]))
+              conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(threadIdx.x + 98)]*kernel.shared_1[27]))
+              conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(threadIdx.x + 105)]*kernel.shared_1[27]))
+              conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(threadIdx.x + 63)]*kernel.shared_1[15]))
+              conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(threadIdx.x + 70)]*kernel.shared_1[15]))
+              conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(threadIdx.x + 77)]*kernel.shared_1[15]))
+              conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(threadIdx.x + 84)]*kernel.shared_1[15]))
+              conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(threadIdx.x + 91)]*kernel.shared_1[15]))
+              conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(threadIdx.x + 98)]*kernel.shared_1[15]))
+              conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(threadIdx.x + 105)]*kernel.shared_1[15]))
+              conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(threadIdx.x + 63)]*kernel.shared_1[39]))
+              conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(threadIdx.x + 70)]*kernel.shared_1[39]))
+              conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(threadIdx.x + 77)]*kernel.shared_1[39]))
+              conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(threadIdx.x + 84)]*kernel.shared_1[39]))
+              conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(threadIdx.x + 91)]*kernel.shared_1[39]))
+              conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(threadIdx.x + 98)]*kernel.shared_1[39]))
+              conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(threadIdx.x + 105)]*kernel.shared_1[39]))
+              conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(threadIdx.x + 70)]*kernel.shared_1[4]))
+              conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(threadIdx.x + 77)]*kernel.shared_1[4]))
+              conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(threadIdx.x + 84)]*kernel.shared_1[4]))
+              conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(threadIdx.x + 91)]*kernel.shared_1[4]))
+              conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(threadIdx.x + 98)]*kernel.shared_1[4]))
+              conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(threadIdx.x + 105)]*kernel.shared_1[4]))
+              conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(threadIdx.x + 112)]*kernel.shared_1[4]))
+              conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(threadIdx.x + 70)]*kernel.shared_1[28]))
+              conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(threadIdx.x + 77)]*kernel.shared_1[28]))
+              conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(threadIdx.x + 84)]*kernel.shared_1[28]))
+              conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(threadIdx.x + 91)]*kernel.shared_1[28]))
+              conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(threadIdx.x + 98)]*kernel.shared_1[28]))
+              conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(threadIdx.x + 105)]*kernel.shared_1[28]))
+              conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(threadIdx.x + 112)]*kernel.shared_1[28]))
+              conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(threadIdx.x + 70)]*kernel.shared_1[16]))
+              conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(threadIdx.x + 77)]*kernel.shared_1[16]))
+              conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(threadIdx.x + 84)]*kernel.shared_1[16]))
+              conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(threadIdx.x + 91)]*kernel.shared_1[16]))
+              conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(threadIdx.x + 98)]*kernel.shared_1[16]))
+              conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(threadIdx.x + 105)]*kernel.shared_1[16]))
+              conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(threadIdx.x + 112)]*kernel.shared_1[16]))
+              conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(threadIdx.x + 70)]*kernel.shared_1[40]))
+              conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(threadIdx.x + 77)]*kernel.shared_1[40]))
+              conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(threadIdx.x + 84)]*kernel.shared_1[40]))
+              conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(threadIdx.x + 91)]*kernel.shared_1[40]))
+              conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(threadIdx.x + 98)]*kernel.shared_1[40]))
+              conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(threadIdx.x + 105)]*kernel.shared_1[40]))
+              conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(threadIdx.x + 112)]*kernel.shared_1[40]))
+              conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(threadIdx.x + 77)]*kernel.shared_1[5]))
+              conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(threadIdx.x + 84)]*kernel.shared_1[5]))
+              conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(threadIdx.x + 91)]*kernel.shared_1[5]))
+              conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(threadIdx.x + 98)]*kernel.shared_1[5]))
+              conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(threadIdx.x + 105)]*kernel.shared_1[5]))
+              conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(threadIdx.x + 112)]*kernel.shared_1[5]))
+              conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(threadIdx.x + 119)]*kernel.shared_1[5]))
+              conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(threadIdx.x + 77)]*kernel.shared_1[29]))
+              conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(threadIdx.x + 84)]*kernel.shared_1[29]))
+              conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(threadIdx.x + 91)]*kernel.shared_1[29]))
+              conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(threadIdx.x + 98)]*kernel.shared_1[29]))
+              conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(threadIdx.x + 105)]*kernel.shared_1[29]))
+              conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(threadIdx.x + 112)]*kernel.shared_1[29]))
+              conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(threadIdx.x + 119)]*kernel.shared_1[29]))
+              conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(threadIdx.x + 77)]*kernel.shared_1[17]))
+              conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(threadIdx.x + 84)]*kernel.shared_1[17]))
+              conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(threadIdx.x + 91)]*kernel.shared_1[17]))
+              conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(threadIdx.x + 98)]*kernel.shared_1[17]))
+              conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(threadIdx.x + 105)]*kernel.shared_1[17]))
+              conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(threadIdx.x + 112)]*kernel.shared_1[17]))
+              conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(threadIdx.x + 119)]*kernel.shared_1[17]))
+              conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(threadIdx.x + 77)]*kernel.shared_1[41]))
+              conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(threadIdx.x + 84)]*kernel.shared_1[41]))
+              conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(threadIdx.x + 91)]*kernel.shared_1[41]))
+              conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(threadIdx.x + 98)]*kernel.shared_1[41]))
+              conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(threadIdx.x + 105)]*kernel.shared_1[41]))
+              conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(threadIdx.x + 112)]*kernel.shared_1[41]))
+              conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(threadIdx.x + 119)]*kernel.shared_1[41]))
+              conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(threadIdx.x + 126)]*kernel.shared_1[6]))
+              conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(threadIdx.x + 133)]*kernel.shared_1[6]))
+              conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(threadIdx.x + 140)]*kernel.shared_1[6]))
+              conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(threadIdx.x + 147)]*kernel.shared_1[6]))
+              conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(threadIdx.x + 154)]*kernel.shared_1[6]))
+              conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(threadIdx.x + 161)]*kernel.shared_1[6]))
+              conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(threadIdx.x + 168)]*kernel.shared_1[6]))
+              conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(threadIdx.x + 126)]*kernel.shared_1[30]))
+              conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(threadIdx.x + 133)]*kernel.shared_1[30]))
+              conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(threadIdx.x + 140)]*kernel.shared_1[30]))
+              conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(threadIdx.x + 147)]*kernel.shared_1[30]))
+              conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(threadIdx.x + 154)]*kernel.shared_1[30]))
+              conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(threadIdx.x + 161)]*kernel.shared_1[30]))
+              conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(threadIdx.x + 168)]*kernel.shared_1[30]))
+              conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(threadIdx.x + 126)]*kernel.shared_1[18]))
+              conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(threadIdx.x + 133)]*kernel.shared_1[18]))
+              conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(threadIdx.x + 140)]*kernel.shared_1[18]))
+              conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(threadIdx.x + 147)]*kernel.shared_1[18]))
+              conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(threadIdx.x + 154)]*kernel.shared_1[18]))
+              conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(threadIdx.x + 161)]*kernel.shared_1[18]))
+              conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(threadIdx.x + 168)]*kernel.shared_1[18]))
+              conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(threadIdx.x + 126)]*kernel.shared_1[42]))
+              conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(threadIdx.x + 133)]*kernel.shared_1[42]))
+              conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(threadIdx.x + 140)]*kernel.shared_1[42]))
+              conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(threadIdx.x + 147)]*kernel.shared_1[42]))
+              conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(threadIdx.x + 154)]*kernel.shared_1[42]))
+              conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(threadIdx.x + 161)]*kernel.shared_1[42]))
+              conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(threadIdx.x + 168)]*kernel.shared_1[42]))
+              conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(threadIdx.x + 133)]*kernel.shared_1[7]))
+              conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(threadIdx.x + 140)]*kernel.shared_1[7]))
+              conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(threadIdx.x + 147)]*kernel.shared_1[7]))
+              conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(threadIdx.x + 154)]*kernel.shared_1[7]))
+              conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(threadIdx.x + 161)]*kernel.shared_1[7]))
+              conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(threadIdx.x + 168)]*kernel.shared_1[7]))
+              conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(threadIdx.x + 175)]*kernel.shared_1[7]))
+              conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(threadIdx.x + 133)]*kernel.shared_1[31]))
+              conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(threadIdx.x + 140)]*kernel.shared_1[31]))
+              conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(threadIdx.x + 147)]*kernel.shared_1[31]))
+              conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(threadIdx.x + 154)]*kernel.shared_1[31]))
+              conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(threadIdx.x + 161)]*kernel.shared_1[31]))
+              conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(threadIdx.x + 168)]*kernel.shared_1[31]))
+              conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(threadIdx.x + 175)]*kernel.shared_1[31]))
+              conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(threadIdx.x + 133)]*kernel.shared_1[19]))
+              conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(threadIdx.x + 140)]*kernel.shared_1[19]))
+              conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(threadIdx.x + 147)]*kernel.shared_1[19]))
+              conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(threadIdx.x + 154)]*kernel.shared_1[19]))
+              conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(threadIdx.x + 161)]*kernel.shared_1[19]))
+              conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(threadIdx.x + 168)]*kernel.shared_1[19]))
+              conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(threadIdx.x + 175)]*kernel.shared_1[19]))
+              conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(threadIdx.x + 133)]*kernel.shared_1[43]))
+              conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(threadIdx.x + 140)]*kernel.shared_1[43]))
+              conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(threadIdx.x + 147)]*kernel.shared_1[43]))
+              conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(threadIdx.x + 154)]*kernel.shared_1[43]))
+              conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(threadIdx.x + 161)]*kernel.shared_1[43]))
+              conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(threadIdx.x + 168)]*kernel.shared_1[43]))
+              conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(threadIdx.x + 175)]*kernel.shared_1[43]))
+              conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(threadIdx.x + 140)]*kernel.shared_1[8]))
+              conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(threadIdx.x + 147)]*kernel.shared_1[8]))
+              conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(threadIdx.x + 154)]*kernel.shared_1[8]))
+              conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(threadIdx.x + 161)]*kernel.shared_1[8]))
+              conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(threadIdx.x + 168)]*kernel.shared_1[8]))
+              conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(threadIdx.x + 175)]*kernel.shared_1[8]))
+              conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(threadIdx.x + 182)]*kernel.shared_1[8]))
+              conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(threadIdx.x + 140)]*kernel.shared_1[32]))
+              conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(threadIdx.x + 147)]*kernel.shared_1[32]))
+              conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(threadIdx.x + 154)]*kernel.shared_1[32]))
+              conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(threadIdx.x + 161)]*kernel.shared_1[32]))
+              conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(threadIdx.x + 168)]*kernel.shared_1[32]))
+              conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(threadIdx.x + 175)]*kernel.shared_1[32]))
+              conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(threadIdx.x + 182)]*kernel.shared_1[32]))
+              conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(threadIdx.x + 140)]*kernel.shared_1[20]))
+              conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(threadIdx.x + 147)]*kernel.shared_1[20]))
+              conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(threadIdx.x + 154)]*kernel.shared_1[20]))
+              conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(threadIdx.x + 161)]*kernel.shared_1[20]))
+              conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(threadIdx.x + 168)]*kernel.shared_1[20]))
+              conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(threadIdx.x + 175)]*kernel.shared_1[20]))
+              conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(threadIdx.x + 182)]*kernel.shared_1[20]))
+              conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(threadIdx.x + 140)]*kernel.shared_1[44]))
+              conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(threadIdx.x + 147)]*kernel.shared_1[44]))
+              conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(threadIdx.x + 154)]*kernel.shared_1[44]))
+              conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(threadIdx.x + 161)]*kernel.shared_1[44]))
+              conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(threadIdx.x + 168)]*kernel.shared_1[44]))
+              conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(threadIdx.x + 175)]*kernel.shared_1[44]))
+              conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(threadIdx.x + 182)]*kernel.shared_1[44]))
+              conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(threadIdx.x + 189)]*kernel.shared_1[9]))
+              conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(threadIdx.x + 196)]*kernel.shared_1[9]))
+              conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(threadIdx.x + 203)]*kernel.shared_1[9]))
+              conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(threadIdx.x + 210)]*kernel.shared_1[9]))
+              conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(threadIdx.x + 217)]*kernel.shared_1[9]))
+              conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(threadIdx.x + 224)]*kernel.shared_1[9]))
+              conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(threadIdx.x + 231)]*kernel.shared_1[9]))
+              conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(threadIdx.x + 189)]*kernel.shared_1[33]))
+              conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(threadIdx.x + 196)]*kernel.shared_1[33]))
+              conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(threadIdx.x + 203)]*kernel.shared_1[33]))
+              conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(threadIdx.x + 210)]*kernel.shared_1[33]))
+              conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(threadIdx.x + 217)]*kernel.shared_1[33]))
+              conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(threadIdx.x + 224)]*kernel.shared_1[33]))
+              conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(threadIdx.x + 231)]*kernel.shared_1[33]))
+              conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(threadIdx.x + 189)]*kernel.shared_1[21]))
+              conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(threadIdx.x + 196)]*kernel.shared_1[21]))
+              conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(threadIdx.x + 203)]*kernel.shared_1[21]))
+              conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(threadIdx.x + 210)]*kernel.shared_1[21]))
+              conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(threadIdx.x + 217)]*kernel.shared_1[21]))
+              conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(threadIdx.x + 224)]*kernel.shared_1[21]))
+              conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(threadIdx.x + 231)]*kernel.shared_1[21]))
+              conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(threadIdx.x + 189)]*kernel.shared_1[45]))
+              conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(threadIdx.x + 196)]*kernel.shared_1[45]))
+              conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(threadIdx.x + 203)]*kernel.shared_1[45]))
+              conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(threadIdx.x + 210)]*kernel.shared_1[45]))
+              conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(threadIdx.x + 217)]*kernel.shared_1[45]))
+              conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(threadIdx.x + 224)]*kernel.shared_1[45]))
+              conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(threadIdx.x + 231)]*kernel.shared_1[45]))
+              conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(threadIdx.x + 196)]*kernel.shared_1[10]))
+              conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(threadIdx.x + 203)]*kernel.shared_1[10]))
+              conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(threadIdx.x + 210)]*kernel.shared_1[10]))
+              conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(threadIdx.x + 217)]*kernel.shared_1[10]))
+              conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(threadIdx.x + 224)]*kernel.shared_1[10]))
+              conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(threadIdx.x + 231)]*kernel.shared_1[10]))
+              conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(threadIdx.x + 238)]*kernel.shared_1[10]))
+              conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(threadIdx.x + 196)]*kernel.shared_1[34]))
+              conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(threadIdx.x + 203)]*kernel.shared_1[34]))
+              conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(threadIdx.x + 210)]*kernel.shared_1[34]))
+              conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(threadIdx.x + 217)]*kernel.shared_1[34]))
+              conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(threadIdx.x + 224)]*kernel.shared_1[34]))
+              conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(threadIdx.x + 231)]*kernel.shared_1[34]))
+              conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(threadIdx.x + 238)]*kernel.shared_1[34]))
+              conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(threadIdx.x + 196)]*kernel.shared_1[22]))
+              conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(threadIdx.x + 203)]*kernel.shared_1[22]))
+              conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(threadIdx.x + 210)]*kernel.shared_1[22]))
+              conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(threadIdx.x + 217)]*kernel.shared_1[22]))
+              conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(threadIdx.x + 224)]*kernel.shared_1[22]))
+              conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(threadIdx.x + 231)]*kernel.shared_1[22]))
+              conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(threadIdx.x + 238)]*kernel.shared_1[22]))
+              conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(threadIdx.x + 196)]*kernel.shared_1[46]))
+              conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(threadIdx.x + 203)]*kernel.shared_1[46]))
+              conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(threadIdx.x + 210)]*kernel.shared_1[46]))
+              conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(threadIdx.x + 217)]*kernel.shared_1[46]))
+              conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(threadIdx.x + 224)]*kernel.shared_1[46]))
+              conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(threadIdx.x + 231)]*kernel.shared_1[46]))
+              conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(threadIdx.x + 238)]*kernel.shared_1[46]))
+              conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(threadIdx.x + 203)]*kernel.shared_1[11]))
+              conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(threadIdx.x + 210)]*kernel.shared_1[11]))
+              conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(threadIdx.x + 217)]*kernel.shared_1[11]))
+              conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(threadIdx.x + 224)]*kernel.shared_1[11]))
+              conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(threadIdx.x + 231)]*kernel.shared_1[11]))
+              conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(threadIdx.x + 238)]*kernel.shared_1[11]))
+              conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(threadIdx.x + 245)]*kernel.shared_1[11]))
+              conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(threadIdx.x + 203)]*kernel.shared_1[35]))
+              conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(threadIdx.x + 210)]*kernel.shared_1[35]))
+              conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(threadIdx.x + 217)]*kernel.shared_1[35]))
+              conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(threadIdx.x + 224)]*kernel.shared_1[35]))
+              conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(threadIdx.x + 231)]*kernel.shared_1[35]))
+              conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(threadIdx.x + 238)]*kernel.shared_1[35]))
+              conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(threadIdx.x + 245)]*kernel.shared_1[35]))
+              conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(threadIdx.x + 203)]*kernel.shared_1[23]))
+              conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(threadIdx.x + 210)]*kernel.shared_1[23]))
+              conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(threadIdx.x + 217)]*kernel.shared_1[23]))
+              conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(threadIdx.x + 224)]*kernel.shared_1[23]))
+              conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(threadIdx.x + 231)]*kernel.shared_1[23]))
+              conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(threadIdx.x + 238)]*kernel.shared_1[23]))
+              conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(threadIdx.x + 245)]*kernel.shared_1[23]))
+              conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(threadIdx.x + 203)]*kernel.shared_1[47]))
+              conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(threadIdx.x + 210)]*kernel.shared_1[47]))
+              conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(threadIdx.x + 217)]*kernel.shared_1[47]))
+              conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(threadIdx.x + 224)]*kernel.shared_1[47]))
+              conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(threadIdx.x + 231)]*kernel.shared_1[47]))
+              conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(threadIdx.x + 238)]*kernel.shared_1[47]))
+              conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(threadIdx.x + 245)]*kernel.shared_1[47]))
             }
-            attr [IterVar(threadIdx.x_2: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1: Buffer(kernel.shared, float32, [1152], [], scope="shared")[threadIdx.x_2] = kernel[(((floordiv(blockIdx.x, 7)*147456) + cse_var_1) + threadIdx.x_2)]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 16)] = kernel[((((floordiv(blockIdx.x, 7)*147456) + cse_var_1) + (floordiv((threadIdx.x_2 + 16), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 32)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 32), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 32), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 48)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 48), 36)*4608)) + cse_var_1) + ((floordiv(threadIdx.x_2, 3) + 4)*3)) + floormod(threadIdx.x_2, 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 64)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 64), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 28), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 80)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 80), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 8), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 96)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 96), 36)*4608)) + cse_var_1) + (floormod((floordiv(threadIdx.x_2, 3) + 8), 12)*3)) + floormod(threadIdx.x_2, 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 112)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 112), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 4), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 128)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 128), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 20), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 144)] = kernel[((((floordiv(blockIdx.x, 7)*147456) + cse_var_1) + threadIdx.x_2) + 18432)]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 160)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 160), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 16), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 176)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 176), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 32), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 192)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 192), 36)*4608)) + cse_var_1) + ((floordiv(threadIdx.x_2, 3) + 4)*3)) + floormod(threadIdx.x_2, 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 208)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 208), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 28), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 224)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 224), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 8), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 240)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 240), 36)*4608)) + cse_var_1) + (floormod((floordiv(threadIdx.x_2, 3) + 8), 12)*3)) + floormod(threadIdx.x_2, 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 256)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 256), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 4), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 272)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 272), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 20), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 288)] = kernel[((((floordiv(blockIdx.x, 7)*147456) + cse_var_1) + threadIdx.x_2) + 36864)]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 304)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 304), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 16), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 320)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 320), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 32), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 336)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 336), 36)*4608)) + cse_var_1) + ((floordiv(threadIdx.x_2, 3) + 4)*3)) + floormod(threadIdx.x_2, 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 352)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 352), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 28), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 368)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 368), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 8), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 384)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 384), 36)*4608)) + cse_var_1) + (floormod((floordiv(threadIdx.x_2, 3) + 8), 12)*3)) + floormod(threadIdx.x_2, 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 400)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 400), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 4), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 416)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 416), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 20), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 432)] = kernel[((((floordiv(blockIdx.x, 7)*147456) + cse_var_1) + threadIdx.x_2) + 55296)]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 448)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 448), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 16), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 464)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 464), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 32), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 480)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 480), 36)*4608)) + cse_var_1) + ((floordiv(threadIdx.x_2, 3) + 4)*3)) + floormod(threadIdx.x_2, 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 496)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 496), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 28), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 512)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 512), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 8), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 528)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 528), 36)*4608)) + cse_var_1) + (floormod((floordiv(threadIdx.x_2, 3) + 8), 12)*3)) + floormod(threadIdx.x_2, 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 544)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 544), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 4), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 560)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 560), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 20), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 576)] = kernel[((((floordiv(blockIdx.x, 7)*147456) + cse_var_1) + threadIdx.x_2) + 73728)]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 592)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 592), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 16), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 608)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 608), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 32), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 624)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 624), 36)*4608)) + cse_var_1) + ((floordiv(threadIdx.x_2, 3) + 4)*3)) + floormod(threadIdx.x_2, 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 640)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 640), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 28), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 656)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 656), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 8), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 672)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 672), 36)*4608)) + cse_var_1) + (floormod((floordiv(threadIdx.x_2, 3) + 8), 12)*3)) + floormod(threadIdx.x_2, 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 688)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 688), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 4), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 704)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 704), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 20), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 720)] = kernel[((((floordiv(blockIdx.x, 7)*147456) + cse_var_1) + threadIdx.x_2) + 92160)]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 736)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 736), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 16), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 752)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 752), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 32), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 768)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 768), 36)*4608)) + cse_var_1) + ((floordiv(threadIdx.x_2, 3) + 4)*3)) + floormod(threadIdx.x_2, 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 784)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 784), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 28), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 800)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 800), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 8), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 816)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 816), 36)*4608)) + cse_var_1) + (floormod((floordiv(threadIdx.x_2, 3) + 8), 12)*3)) + floormod(threadIdx.x_2, 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 832)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 832), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 4), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 848)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 848), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 20), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 864)] = kernel[((((floordiv(blockIdx.x, 7)*147456) + cse_var_1) + threadIdx.x_2) + 110592)]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 880)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 880), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 16), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 896)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 896), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 32), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 912)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 912), 36)*4608)) + cse_var_1) + ((floordiv(threadIdx.x_2, 3) + 4)*3)) + floormod(threadIdx.x_2, 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 928)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 928), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 28), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 944)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 944), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 8), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 960)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 960), 36)*4608)) + cse_var_1) + (floormod((floordiv(threadIdx.x_2, 3) + 8), 12)*3)) + floormod(threadIdx.x_2, 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 976)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 976), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 4), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 992)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 992), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 20), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 1008)] = kernel[((((floordiv(blockIdx.x, 7)*147456) + cse_var_1) + threadIdx.x_2) + 129024)]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 1024)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 1024), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 16), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 1040)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 1040), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 32), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 1056)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 1056), 36)*4608)) + cse_var_1) + ((floordiv(threadIdx.x_2, 3) + 4)*3)) + floormod(threadIdx.x_2, 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 1072)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 1072), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 28), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 1088)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 1088), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 8), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 1104)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 1104), 36)*4608)) + cse_var_1) + (floormod((floordiv(threadIdx.x_2, 3) + 8), 12)*3)) + floormod(threadIdx.x_2, 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 1120)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 1120), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 4), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-            attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 16;
-            kernel.shared_1[(threadIdx.x_2 + 1136)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 1136), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 20), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-            conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[0]*kernel.shared_1[(threadIdx.x*72)]))
-            conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[1]*kernel.shared_1[((threadIdx.x*72) + 1)]))
-            conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[2]*kernel.shared_1[((threadIdx.x*72) + 2)]))
-            conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[3]*kernel.shared_1[((threadIdx.x*72) + 3)]))
-            conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[4]*kernel.shared_1[((threadIdx.x*72) + 4)]))
-            conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[5]*kernel.shared_1[((threadIdx.x*72) + 5)]))
-            conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[6]*kernel.shared_1[((threadIdx.x*72) + 6)]))
-            conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[7]*kernel.shared_1[((threadIdx.x*72) + 7)]))
-            conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[8]*kernel.shared_1[((threadIdx.x*72) + 8)]))
-            conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[3]*kernel.shared_1[(threadIdx.x*72)]))
-            conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[4]*kernel.shared_1[((threadIdx.x*72) + 1)]))
-            conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[5]*kernel.shared_1[((threadIdx.x*72) + 2)]))
-            conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[6]*kernel.shared_1[((threadIdx.x*72) + 3)]))
-            conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[7]*kernel.shared_1[((threadIdx.x*72) + 4)]))
-            conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[8]*kernel.shared_1[((threadIdx.x*72) + 5)]))
-            conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[9]*kernel.shared_1[((threadIdx.x*72) + 6)]))
-            conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[10]*kernel.shared_1[((threadIdx.x*72) + 7)]))
-            conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[11]*kernel.shared_1[((threadIdx.x*72) + 8)]))
-            conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[6]*kernel.shared_1[(threadIdx.x*72)]))
-            conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[7]*kernel.shared_1[((threadIdx.x*72) + 1)]))
-            conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[8]*kernel.shared_1[((threadIdx.x*72) + 2)]))
-            conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[9]*kernel.shared_1[((threadIdx.x*72) + 3)]))
-            conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[10]*kernel.shared_1[((threadIdx.x*72) + 4)]))
-            conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[11]*kernel.shared_1[((threadIdx.x*72) + 5)]))
-            conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[12]*kernel.shared_1[((threadIdx.x*72) + 6)]))
-            conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[13]*kernel.shared_1[((threadIdx.x*72) + 7)]))
-            conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[14]*kernel.shared_1[((threadIdx.x*72) + 8)]))
-            conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[9]*kernel.shared_1[(threadIdx.x*72)]))
-            conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[10]*kernel.shared_1[((threadIdx.x*72) + 1)]))
-            conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[11]*kernel.shared_1[((threadIdx.x*72) + 2)]))
-            conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[12]*kernel.shared_1[((threadIdx.x*72) + 3)]))
-            conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[13]*kernel.shared_1[((threadIdx.x*72) + 4)]))
-            conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[14]*kernel.shared_1[((threadIdx.x*72) + 5)]))
-            conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[15]*kernel.shared_1[((threadIdx.x*72) + 6)]))
-            conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[16]*kernel.shared_1[((threadIdx.x*72) + 7)]))
-            conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[17]*kernel.shared_1[((threadIdx.x*72) + 8)]))
-            conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[12]*kernel.shared_1[(threadIdx.x*72)]))
-            conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[13]*kernel.shared_1[((threadIdx.x*72) + 1)]))
-            conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[14]*kernel.shared_1[((threadIdx.x*72) + 2)]))
-            conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[15]*kernel.shared_1[((threadIdx.x*72) + 3)]))
-            conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[16]*kernel.shared_1[((threadIdx.x*72) + 4)]))
-            conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[17]*kernel.shared_1[((threadIdx.x*72) + 5)]))
-            conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[18]*kernel.shared_1[((threadIdx.x*72) + 6)]))
-            conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[19]*kernel.shared_1[((threadIdx.x*72) + 7)]))
-            conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[20]*kernel.shared_1[((threadIdx.x*72) + 8)]))
-            conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[15]*kernel.shared_1[(threadIdx.x*72)]))
-            conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[16]*kernel.shared_1[((threadIdx.x*72) + 1)]))
-            conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[17]*kernel.shared_1[((threadIdx.x*72) + 2)]))
-            conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[18]*kernel.shared_1[((threadIdx.x*72) + 3)]))
-            conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[19]*kernel.shared_1[((threadIdx.x*72) + 4)]))
-            conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[20]*kernel.shared_1[((threadIdx.x*72) + 5)]))
-            conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[21]*kernel.shared_1[((threadIdx.x*72) + 6)]))
-            conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[22]*kernel.shared_1[((threadIdx.x*72) + 7)]))
-            conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[23]*kernel.shared_1[((threadIdx.x*72) + 8)]))
-            conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[18]*kernel.shared_1[(threadIdx.x*72)]))
-            conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[19]*kernel.shared_1[((threadIdx.x*72) + 1)]))
-            conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[20]*kernel.shared_1[((threadIdx.x*72) + 2)]))
-            conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[21]*kernel.shared_1[((threadIdx.x*72) + 3)]))
-            conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[22]*kernel.shared_1[((threadIdx.x*72) + 4)]))
-            conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[23]*kernel.shared_1[((threadIdx.x*72) + 5)]))
-            conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[24]*kernel.shared_1[((threadIdx.x*72) + 6)]))
-            conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[25]*kernel.shared_1[((threadIdx.x*72) + 7)]))
-            conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[26]*kernel.shared_1[((threadIdx.x*72) + 8)]))
-            conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[0]*kernel.shared_1[((threadIdx.x*72) + 36)]))
-            conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[1]*kernel.shared_1[((threadIdx.x*72) + 37)]))
-            conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[2]*kernel.shared_1[((threadIdx.x*72) + 38)]))
-            conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[3]*kernel.shared_1[((threadIdx.x*72) + 39)]))
-            conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[4]*kernel.shared_1[((threadIdx.x*72) + 40)]))
-            conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[5]*kernel.shared_1[((threadIdx.x*72) + 41)]))
-            conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[6]*kernel.shared_1[((threadIdx.x*72) + 42)]))
-            conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[7]*kernel.shared_1[((threadIdx.x*72) + 43)]))
-            conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[8]*kernel.shared_1[((threadIdx.x*72) + 44)]))
-            conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[3]*kernel.shared_1[((threadIdx.x*72) + 36)]))
-            conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[4]*kernel.shared_1[((threadIdx.x*72) + 37)]))
-            conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[5]*kernel.shared_1[((threadIdx.x*72) + 38)]))
-            conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[6]*kernel.shared_1[((threadIdx.x*72) + 39)]))
-            conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[7]*kernel.shared_1[((threadIdx.x*72) + 40)]))
-            conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[8]*kernel.shared_1[((threadIdx.x*72) + 41)]))
-            conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[9]*kernel.shared_1[((threadIdx.x*72) + 42)]))
-            conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[10]*kernel.shared_1[((threadIdx.x*72) + 43)]))
-            conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[11]*kernel.shared_1[((threadIdx.x*72) + 44)]))
-            conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[6]*kernel.shared_1[((threadIdx.x*72) + 36)]))
-            conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[7]*kernel.shared_1[((threadIdx.x*72) + 37)]))
-            conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[8]*kernel.shared_1[((threadIdx.x*72) + 38)]))
-            conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[9]*kernel.shared_1[((threadIdx.x*72) + 39)]))
-            conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[10]*kernel.shared_1[((threadIdx.x*72) + 40)]))
-            conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[11]*kernel.shared_1[((threadIdx.x*72) + 41)]))
-            conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[12]*kernel.shared_1[((threadIdx.x*72) + 42)]))
-            conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[13]*kernel.shared_1[((threadIdx.x*72) + 43)]))
-            conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[14]*kernel.shared_1[((threadIdx.x*72) + 44)]))
-            conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[9]*kernel.shared_1[((threadIdx.x*72) + 36)]))
-            conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[10]*kernel.shared_1[((threadIdx.x*72) + 37)]))
-            conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[11]*kernel.shared_1[((threadIdx.x*72) + 38)]))
-            conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[12]*kernel.shared_1[((threadIdx.x*72) + 39)]))
-            conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[13]*kernel.shared_1[((threadIdx.x*72) + 40)]))
-            conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[14]*kernel.shared_1[((threadIdx.x*72) + 41)]))
-            conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[15]*kernel.shared_1[((threadIdx.x*72) + 42)]))
-            conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[16]*kernel.shared_1[((threadIdx.x*72) + 43)]))
-            conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[17]*kernel.shared_1[((threadIdx.x*72) + 44)]))
-            conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[12]*kernel.shared_1[((threadIdx.x*72) + 36)]))
-            conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[13]*kernel.shared_1[((threadIdx.x*72) + 37)]))
-            conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[14]*kernel.shared_1[((threadIdx.x*72) + 38)]))
-            conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[15]*kernel.shared_1[((threadIdx.x*72) + 39)]))
-            conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[16]*kernel.shared_1[((threadIdx.x*72) + 40)]))
-            conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[17]*kernel.shared_1[((threadIdx.x*72) + 41)]))
-            conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[18]*kernel.shared_1[((threadIdx.x*72) + 42)]))
-            conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[19]*kernel.shared_1[((threadIdx.x*72) + 43)]))
-            conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[20]*kernel.shared_1[((threadIdx.x*72) + 44)]))
-            conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[15]*kernel.shared_1[((threadIdx.x*72) + 36)]))
-            conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[16]*kernel.shared_1[((threadIdx.x*72) + 37)]))
-            conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[17]*kernel.shared_1[((threadIdx.x*72) + 38)]))
-            conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[18]*kernel.shared_1[((threadIdx.x*72) + 39)]))
-            conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[19]*kernel.shared_1[((threadIdx.x*72) + 40)]))
-            conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[20]*kernel.shared_1[((threadIdx.x*72) + 41)]))
-            conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[21]*kernel.shared_1[((threadIdx.x*72) + 42)]))
-            conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[22]*kernel.shared_1[((threadIdx.x*72) + 43)]))
-            conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[23]*kernel.shared_1[((threadIdx.x*72) + 44)]))
-            conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[18]*kernel.shared_1[((threadIdx.x*72) + 36)]))
-            conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[19]*kernel.shared_1[((threadIdx.x*72) + 37)]))
-            conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[20]*kernel.shared_1[((threadIdx.x*72) + 38)]))
-            conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[21]*kernel.shared_1[((threadIdx.x*72) + 39)]))
-            conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[22]*kernel.shared_1[((threadIdx.x*72) + 40)]))
-            conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[23]*kernel.shared_1[((threadIdx.x*72) + 41)]))
-            conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[24]*kernel.shared_1[((threadIdx.x*72) + 42)]))
-            conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[25]*kernel.shared_1[((threadIdx.x*72) + 43)]))
-            conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[26]*kernel.shared_1[((threadIdx.x*72) + 44)]))
-            conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[27]*kernel.shared_1[((threadIdx.x*72) + 9)]))
-            conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[28]*kernel.shared_1[((threadIdx.x*72) + 10)]))
-            conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[29]*kernel.shared_1[((threadIdx.x*72) + 11)]))
-            conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[30]*kernel.shared_1[((threadIdx.x*72) + 12)]))
-            conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[31]*kernel.shared_1[((threadIdx.x*72) + 13)]))
-            conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[32]*kernel.shared_1[((threadIdx.x*72) + 14)]))
-            conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[33]*kernel.shared_1[((threadIdx.x*72) + 15)]))
-            conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[34]*kernel.shared_1[((threadIdx.x*72) + 16)]))
-            conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[35]*kernel.shared_1[((threadIdx.x*72) + 17)]))
-            conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[30]*kernel.shared_1[((threadIdx.x*72) + 9)]))
-            conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[31]*kernel.shared_1[((threadIdx.x*72) + 10)]))
-            conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[32]*kernel.shared_1[((threadIdx.x*72) + 11)]))
-            conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[33]*kernel.shared_1[((threadIdx.x*72) + 12)]))
-            conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[34]*kernel.shared_1[((threadIdx.x*72) + 13)]))
-            conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[35]*kernel.shared_1[((threadIdx.x*72) + 14)]))
-            conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[36]*kernel.shared_1[((threadIdx.x*72) + 15)]))
-            conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[37]*kernel.shared_1[((threadIdx.x*72) + 16)]))
-            conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[38]*kernel.shared_1[((threadIdx.x*72) + 17)]))
-            conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[33]*kernel.shared_1[((threadIdx.x*72) + 9)]))
-            conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[34]*kernel.shared_1[((threadIdx.x*72) + 10)]))
-            conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[35]*kernel.shared_1[((threadIdx.x*72) + 11)]))
-            conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[36]*kernel.shared_1[((threadIdx.x*72) + 12)]))
-            conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[37]*kernel.shared_1[((threadIdx.x*72) + 13)]))
-            conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[38]*kernel.shared_1[((threadIdx.x*72) + 14)]))
-            conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[39]*kernel.shared_1[((threadIdx.x*72) + 15)]))
-            conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[40]*kernel.shared_1[((threadIdx.x*72) + 16)]))
-            conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[41]*kernel.shared_1[((threadIdx.x*72) + 17)]))
-            conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[36]*kernel.shared_1[((threadIdx.x*72) + 9)]))
-            conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[37]*kernel.shared_1[((threadIdx.x*72) + 10)]))
-            conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[38]*kernel.shared_1[((threadIdx.x*72) + 11)]))
-            conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[39]*kernel.shared_1[((threadIdx.x*72) + 12)]))
-            conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[40]*kernel.shared_1[((threadIdx.x*72) + 13)]))
-            conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[41]*kernel.shared_1[((threadIdx.x*72) + 14)]))
-            conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[42]*kernel.shared_1[((threadIdx.x*72) + 15)]))
-            conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[43]*kernel.shared_1[((threadIdx.x*72) + 16)]))
-            conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[44]*kernel.shared_1[((threadIdx.x*72) + 17)]))
-            conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[39]*kernel.shared_1[((threadIdx.x*72) + 9)]))
-            conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[40]*kernel.shared_1[((threadIdx.x*72) + 10)]))
-            conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[41]*kernel.shared_1[((threadIdx.x*72) + 11)]))
-            conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[42]*kernel.shared_1[((threadIdx.x*72) + 12)]))
-            conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[43]*kernel.shared_1[((threadIdx.x*72) + 13)]))
-            conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[44]*kernel.shared_1[((threadIdx.x*72) + 14)]))
-            conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[45]*kernel.shared_1[((threadIdx.x*72) + 15)]))
-            conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[46]*kernel.shared_1[((threadIdx.x*72) + 16)]))
-            conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[47]*kernel.shared_1[((threadIdx.x*72) + 17)]))
-            conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[42]*kernel.shared_1[((threadIdx.x*72) + 9)]))
-            conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[43]*kernel.shared_1[((threadIdx.x*72) + 10)]))
-            conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[44]*kernel.shared_1[((threadIdx.x*72) + 11)]))
-            conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[45]*kernel.shared_1[((threadIdx.x*72) + 12)]))
-            conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[46]*kernel.shared_1[((threadIdx.x*72) + 13)]))
-            conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[47]*kernel.shared_1[((threadIdx.x*72) + 14)]))
-            conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[48]*kernel.shared_1[((threadIdx.x*72) + 15)]))
-            conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[49]*kernel.shared_1[((threadIdx.x*72) + 16)]))
-            conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[50]*kernel.shared_1[((threadIdx.x*72) + 17)]))
-            conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[45]*kernel.shared_1[((threadIdx.x*72) + 9)]))
-            conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[46]*kernel.shared_1[((threadIdx.x*72) + 10)]))
-            conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[47]*kernel.shared_1[((threadIdx.x*72) + 11)]))
-            conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[48]*kernel.shared_1[((threadIdx.x*72) + 12)]))
-            conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[49]*kernel.shared_1[((threadIdx.x*72) + 13)]))
-            conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[50]*kernel.shared_1[((threadIdx.x*72) + 14)]))
-            conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[51]*kernel.shared_1[((threadIdx.x*72) + 15)]))
-            conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[52]*kernel.shared_1[((threadIdx.x*72) + 16)]))
-            conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[53]*kernel.shared_1[((threadIdx.x*72) + 17)]))
-            conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[27]*kernel.shared_1[((threadIdx.x*72) + 45)]))
-            conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[28]*kernel.shared_1[((threadIdx.x*72) + 46)]))
-            conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[29]*kernel.shared_1[((threadIdx.x*72) + 47)]))
-            conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[30]*kernel.shared_1[((threadIdx.x*72) + 48)]))
-            conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[31]*kernel.shared_1[((threadIdx.x*72) + 49)]))
-            conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[32]*kernel.shared_1[((threadIdx.x*72) + 50)]))
-            conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[33]*kernel.shared_1[((threadIdx.x*72) + 51)]))
-            conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[34]*kernel.shared_1[((threadIdx.x*72) + 52)]))
-            conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[35]*kernel.shared_1[((threadIdx.x*72) + 53)]))
-            conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[30]*kernel.shared_1[((threadIdx.x*72) + 45)]))
-            conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[31]*kernel.shared_1[((threadIdx.x*72) + 46)]))
-            conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[32]*kernel.shared_1[((threadIdx.x*72) + 47)]))
-            conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[33]*kernel.shared_1[((threadIdx.x*72) + 48)]))
-            conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[34]*kernel.shared_1[((threadIdx.x*72) + 49)]))
-            conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[35]*kernel.shared_1[((threadIdx.x*72) + 50)]))
-            conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[36]*kernel.shared_1[((threadIdx.x*72) + 51)]))
-            conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[37]*kernel.shared_1[((threadIdx.x*72) + 52)]))
-            conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[38]*kernel.shared_1[((threadIdx.x*72) + 53)]))
-            conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[33]*kernel.shared_1[((threadIdx.x*72) + 45)]))
-            conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[34]*kernel.shared_1[((threadIdx.x*72) + 46)]))
-            conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[35]*kernel.shared_1[((threadIdx.x*72) + 47)]))
-            conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[36]*kernel.shared_1[((threadIdx.x*72) + 48)]))
-            conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[37]*kernel.shared_1[((threadIdx.x*72) + 49)]))
-            conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[38]*kernel.shared_1[((threadIdx.x*72) + 50)]))
-            conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[39]*kernel.shared_1[((threadIdx.x*72) + 51)]))
-            conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[40]*kernel.shared_1[((threadIdx.x*72) + 52)]))
-            conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[41]*kernel.shared_1[((threadIdx.x*72) + 53)]))
-            conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[36]*kernel.shared_1[((threadIdx.x*72) + 45)]))
-            conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[37]*kernel.shared_1[((threadIdx.x*72) + 46)]))
-            conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[38]*kernel.shared_1[((threadIdx.x*72) + 47)]))
-            conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[39]*kernel.shared_1[((threadIdx.x*72) + 48)]))
-            conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[40]*kernel.shared_1[((threadIdx.x*72) + 49)]))
-            conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[41]*kernel.shared_1[((threadIdx.x*72) + 50)]))
-            conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[42]*kernel.shared_1[((threadIdx.x*72) + 51)]))
-            conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[43]*kernel.shared_1[((threadIdx.x*72) + 52)]))
-            conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[44]*kernel.shared_1[((threadIdx.x*72) + 53)]))
-            conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[39]*kernel.shared_1[((threadIdx.x*72) + 45)]))
-            conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[40]*kernel.shared_1[((threadIdx.x*72) + 46)]))
-            conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[41]*kernel.shared_1[((threadIdx.x*72) + 47)]))
-            conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[42]*kernel.shared_1[((threadIdx.x*72) + 48)]))
-            conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[43]*kernel.shared_1[((threadIdx.x*72) + 49)]))
-            conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[44]*kernel.shared_1[((threadIdx.x*72) + 50)]))
-            conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[45]*kernel.shared_1[((threadIdx.x*72) + 51)]))
-            conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[46]*kernel.shared_1[((threadIdx.x*72) + 52)]))
-            conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[47]*kernel.shared_1[((threadIdx.x*72) + 53)]))
-            conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[42]*kernel.shared_1[((threadIdx.x*72) + 45)]))
-            conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[43]*kernel.shared_1[((threadIdx.x*72) + 46)]))
-            conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[44]*kernel.shared_1[((threadIdx.x*72) + 47)]))
-            conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[45]*kernel.shared_1[((threadIdx.x*72) + 48)]))
-            conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[46]*kernel.shared_1[((threadIdx.x*72) + 49)]))
-            conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[47]*kernel.shared_1[((threadIdx.x*72) + 50)]))
-            conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[48]*kernel.shared_1[((threadIdx.x*72) + 51)]))
-            conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[49]*kernel.shared_1[((threadIdx.x*72) + 52)]))
-            conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[50]*kernel.shared_1[((threadIdx.x*72) + 53)]))
-            conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[45]*kernel.shared_1[((threadIdx.x*72) + 45)]))
-            conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[46]*kernel.shared_1[((threadIdx.x*72) + 46)]))
-            conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[47]*kernel.shared_1[((threadIdx.x*72) + 47)]))
-            conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[48]*kernel.shared_1[((threadIdx.x*72) + 48)]))
-            conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[49]*kernel.shared_1[((threadIdx.x*72) + 49)]))
-            conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[50]*kernel.shared_1[((threadIdx.x*72) + 50)]))
-            conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[51]*kernel.shared_1[((threadIdx.x*72) + 51)]))
-            conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[52]*kernel.shared_1[((threadIdx.x*72) + 52)]))
-            conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[53]*kernel.shared_1[((threadIdx.x*72) + 53)]))
-            conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[54]*kernel.shared_1[((threadIdx.x*72) + 18)]))
-            conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[55]*kernel.shared_1[((threadIdx.x*72) + 19)]))
-            conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[56]*kernel.shared_1[((threadIdx.x*72) + 20)]))
-            conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[57]*kernel.shared_1[((threadIdx.x*72) + 21)]))
-            conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[58]*kernel.shared_1[((threadIdx.x*72) + 22)]))
-            conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[59]*kernel.shared_1[((threadIdx.x*72) + 23)]))
-            conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[60]*kernel.shared_1[((threadIdx.x*72) + 24)]))
-            conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[61]*kernel.shared_1[((threadIdx.x*72) + 25)]))
-            conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[62]*kernel.shared_1[((threadIdx.x*72) + 26)]))
-            conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[57]*kernel.shared_1[((threadIdx.x*72) + 18)]))
-            conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[58]*kernel.shared_1[((threadIdx.x*72) + 19)]))
-            conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[59]*kernel.shared_1[((threadIdx.x*72) + 20)]))
-            conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[60]*kernel.shared_1[((threadIdx.x*72) + 21)]))
-            conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[61]*kernel.shared_1[((threadIdx.x*72) + 22)]))
-            conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[62]*kernel.shared_1[((threadIdx.x*72) + 23)]))
-            conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[63]*kernel.shared_1[((threadIdx.x*72) + 24)]))
-            conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[64]*kernel.shared_1[((threadIdx.x*72) + 25)]))
-            conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[65]*kernel.shared_1[((threadIdx.x*72) + 26)]))
-            conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[60]*kernel.shared_1[((threadIdx.x*72) + 18)]))
-            conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[61]*kernel.shared_1[((threadIdx.x*72) + 19)]))
-            conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[62]*kernel.shared_1[((threadIdx.x*72) + 20)]))
-            conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[63]*kernel.shared_1[((threadIdx.x*72) + 21)]))
-            conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[64]*kernel.shared_1[((threadIdx.x*72) + 22)]))
-            conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[65]*kernel.shared_1[((threadIdx.x*72) + 23)]))
-            conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[66]*kernel.shared_1[((threadIdx.x*72) + 24)]))
-            conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[67]*kernel.shared_1[((threadIdx.x*72) + 25)]))
-            conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[68]*kernel.shared_1[((threadIdx.x*72) + 26)]))
-            conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[63]*kernel.shared_1[((threadIdx.x*72) + 18)]))
-            conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[64]*kernel.shared_1[((threadIdx.x*72) + 19)]))
-            conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[65]*kernel.shared_1[((threadIdx.x*72) + 20)]))
-            conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[66]*kernel.shared_1[((threadIdx.x*72) + 21)]))
-            conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[67]*kernel.shared_1[((threadIdx.x*72) + 22)]))
-            conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[68]*kernel.shared_1[((threadIdx.x*72) + 23)]))
-            conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[69]*kernel.shared_1[((threadIdx.x*72) + 24)]))
-            conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[70]*kernel.shared_1[((threadIdx.x*72) + 25)]))
-            conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[71]*kernel.shared_1[((threadIdx.x*72) + 26)]))
-            conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[66]*kernel.shared_1[((threadIdx.x*72) + 18)]))
-            conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[67]*kernel.shared_1[((threadIdx.x*72) + 19)]))
-            conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[68]*kernel.shared_1[((threadIdx.x*72) + 20)]))
-            conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[69]*kernel.shared_1[((threadIdx.x*72) + 21)]))
-            conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[70]*kernel.shared_1[((threadIdx.x*72) + 22)]))
-            conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[71]*kernel.shared_1[((threadIdx.x*72) + 23)]))
-            conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[72]*kernel.shared_1[((threadIdx.x*72) + 24)]))
-            conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[73]*kernel.shared_1[((threadIdx.x*72) + 25)]))
-            conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[74]*kernel.shared_1[((threadIdx.x*72) + 26)]))
-            conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[69]*kernel.shared_1[((threadIdx.x*72) + 18)]))
-            conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[70]*kernel.shared_1[((threadIdx.x*72) + 19)]))
-            conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[71]*kernel.shared_1[((threadIdx.x*72) + 20)]))
-            conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[72]*kernel.shared_1[((threadIdx.x*72) + 21)]))
-            conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[73]*kernel.shared_1[((threadIdx.x*72) + 22)]))
-            conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[74]*kernel.shared_1[((threadIdx.x*72) + 23)]))
-            conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[75]*kernel.shared_1[((threadIdx.x*72) + 24)]))
-            conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[76]*kernel.shared_1[((threadIdx.x*72) + 25)]))
-            conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[77]*kernel.shared_1[((threadIdx.x*72) + 26)]))
-            conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[72]*kernel.shared_1[((threadIdx.x*72) + 18)]))
-            conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[73]*kernel.shared_1[((threadIdx.x*72) + 19)]))
-            conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[74]*kernel.shared_1[((threadIdx.x*72) + 20)]))
-            conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[75]*kernel.shared_1[((threadIdx.x*72) + 21)]))
-            conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[76]*kernel.shared_1[((threadIdx.x*72) + 22)]))
-            conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[77]*kernel.shared_1[((threadIdx.x*72) + 23)]))
-            conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[78]*kernel.shared_1[((threadIdx.x*72) + 24)]))
-            conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[79]*kernel.shared_1[((threadIdx.x*72) + 25)]))
-            conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[80]*kernel.shared_1[((threadIdx.x*72) + 26)]))
-            conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[54]*kernel.shared_1[((threadIdx.x*72) + 54)]))
-            conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[55]*kernel.shared_1[((threadIdx.x*72) + 55)]))
-            conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[56]*kernel.shared_1[((threadIdx.x*72) + 56)]))
-            conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[57]*kernel.shared_1[((threadIdx.x*72) + 57)]))
-            conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[58]*kernel.shared_1[((threadIdx.x*72) + 58)]))
-            conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[59]*kernel.shared_1[((threadIdx.x*72) + 59)]))
-            conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[60]*kernel.shared_1[((threadIdx.x*72) + 60)]))
-            conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[61]*kernel.shared_1[((threadIdx.x*72) + 61)]))
-            conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[62]*kernel.shared_1[((threadIdx.x*72) + 62)]))
-            conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[57]*kernel.shared_1[((threadIdx.x*72) + 54)]))
-            conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[58]*kernel.shared_1[((threadIdx.x*72) + 55)]))
-            conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[59]*kernel.shared_1[((threadIdx.x*72) + 56)]))
-            conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[60]*kernel.shared_1[((threadIdx.x*72) + 57)]))
-            conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[61]*kernel.shared_1[((threadIdx.x*72) + 58)]))
-            conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[62]*kernel.shared_1[((threadIdx.x*72) + 59)]))
-            conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[63]*kernel.shared_1[((threadIdx.x*72) + 60)]))
-            conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[64]*kernel.shared_1[((threadIdx.x*72) + 61)]))
-            conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[65]*kernel.shared_1[((threadIdx.x*72) + 62)]))
-            conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[60]*kernel.shared_1[((threadIdx.x*72) + 54)]))
-            conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[61]*kernel.shared_1[((threadIdx.x*72) + 55)]))
-            conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[62]*kernel.shared_1[((threadIdx.x*72) + 56)]))
-            conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[63]*kernel.shared_1[((threadIdx.x*72) + 57)]))
-            conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[64]*kernel.shared_1[((threadIdx.x*72) + 58)]))
-            conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[65]*kernel.shared_1[((threadIdx.x*72) + 59)]))
-            conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[66]*kernel.shared_1[((threadIdx.x*72) + 60)]))
-            conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[67]*kernel.shared_1[((threadIdx.x*72) + 61)]))
-            conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[68]*kernel.shared_1[((threadIdx.x*72) + 62)]))
-            conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[63]*kernel.shared_1[((threadIdx.x*72) + 54)]))
-            conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[64]*kernel.shared_1[((threadIdx.x*72) + 55)]))
-            conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[65]*kernel.shared_1[((threadIdx.x*72) + 56)]))
-            conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[66]*kernel.shared_1[((threadIdx.x*72) + 57)]))
-            conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[67]*kernel.shared_1[((threadIdx.x*72) + 58)]))
-            conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[68]*kernel.shared_1[((threadIdx.x*72) + 59)]))
-            conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[69]*kernel.shared_1[((threadIdx.x*72) + 60)]))
-            conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[70]*kernel.shared_1[((threadIdx.x*72) + 61)]))
-            conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[71]*kernel.shared_1[((threadIdx.x*72) + 62)]))
-            conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[66]*kernel.shared_1[((threadIdx.x*72) + 54)]))
-            conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[67]*kernel.shared_1[((threadIdx.x*72) + 55)]))
-            conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[68]*kernel.shared_1[((threadIdx.x*72) + 56)]))
-            conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[69]*kernel.shared_1[((threadIdx.x*72) + 57)]))
-            conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[70]*kernel.shared_1[((threadIdx.x*72) + 58)]))
-            conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[71]*kernel.shared_1[((threadIdx.x*72) + 59)]))
-            conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[72]*kernel.shared_1[((threadIdx.x*72) + 60)]))
-            conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[73]*kernel.shared_1[((threadIdx.x*72) + 61)]))
-            conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[74]*kernel.shared_1[((threadIdx.x*72) + 62)]))
-            conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[69]*kernel.shared_1[((threadIdx.x*72) + 54)]))
-            conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[70]*kernel.shared_1[((threadIdx.x*72) + 55)]))
-            conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[71]*kernel.shared_1[((threadIdx.x*72) + 56)]))
-            conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[72]*kernel.shared_1[((threadIdx.x*72) + 57)]))
-            conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[73]*kernel.shared_1[((threadIdx.x*72) + 58)]))
-            conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[74]*kernel.shared_1[((threadIdx.x*72) + 59)]))
-            conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[75]*kernel.shared_1[((threadIdx.x*72) + 60)]))
-            conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[76]*kernel.shared_1[((threadIdx.x*72) + 61)]))
-            conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[77]*kernel.shared_1[((threadIdx.x*72) + 62)]))
-            conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[72]*kernel.shared_1[((threadIdx.x*72) + 54)]))
-            conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[73]*kernel.shared_1[((threadIdx.x*72) + 55)]))
-            conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[74]*kernel.shared_1[((threadIdx.x*72) + 56)]))
-            conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[75]*kernel.shared_1[((threadIdx.x*72) + 57)]))
-            conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[76]*kernel.shared_1[((threadIdx.x*72) + 58)]))
-            conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[77]*kernel.shared_1[((threadIdx.x*72) + 59)]))
-            conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[78]*kernel.shared_1[((threadIdx.x*72) + 60)]))
-            conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[79]*kernel.shared_1[((threadIdx.x*72) + 61)]))
-            conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[80]*kernel.shared_1[((threadIdx.x*72) + 62)]))
-            conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[81]*kernel.shared_1[((threadIdx.x*72) + 27)]))
-            conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[82]*kernel.shared_1[((threadIdx.x*72) + 28)]))
-            conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[83]*kernel.shared_1[((threadIdx.x*72) + 29)]))
-            conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[84]*kernel.shared_1[((threadIdx.x*72) + 30)]))
-            conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[85]*kernel.shared_1[((threadIdx.x*72) + 31)]))
-            conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[86]*kernel.shared_1[((threadIdx.x*72) + 32)]))
-            conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[87]*kernel.shared_1[((threadIdx.x*72) + 33)]))
-            conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[88]*kernel.shared_1[((threadIdx.x*72) + 34)]))
-            conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[89]*kernel.shared_1[((threadIdx.x*72) + 35)]))
-            conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[84]*kernel.shared_1[((threadIdx.x*72) + 27)]))
-            conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[85]*kernel.shared_1[((threadIdx.x*72) + 28)]))
-            conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[86]*kernel.shared_1[((threadIdx.x*72) + 29)]))
-            conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[87]*kernel.shared_1[((threadIdx.x*72) + 30)]))
-            conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[88]*kernel.shared_1[((threadIdx.x*72) + 31)]))
-            conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[89]*kernel.shared_1[((threadIdx.x*72) + 32)]))
-            conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[90]*kernel.shared_1[((threadIdx.x*72) + 33)]))
-            conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[91]*kernel.shared_1[((threadIdx.x*72) + 34)]))
-            conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[92]*kernel.shared_1[((threadIdx.x*72) + 35)]))
-            conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[87]*kernel.shared_1[((threadIdx.x*72) + 27)]))
-            conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[88]*kernel.shared_1[((threadIdx.x*72) + 28)]))
-            conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[89]*kernel.shared_1[((threadIdx.x*72) + 29)]))
-            conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[90]*kernel.shared_1[((threadIdx.x*72) + 30)]))
-            conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[91]*kernel.shared_1[((threadIdx.x*72) + 31)]))
-            conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[92]*kernel.shared_1[((threadIdx.x*72) + 32)]))
-            conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[93]*kernel.shared_1[((threadIdx.x*72) + 33)]))
-            conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[94]*kernel.shared_1[((threadIdx.x*72) + 34)]))
-            conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[95]*kernel.shared_1[((threadIdx.x*72) + 35)]))
-            conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[90]*kernel.shared_1[((threadIdx.x*72) + 27)]))
-            conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[91]*kernel.shared_1[((threadIdx.x*72) + 28)]))
-            conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[92]*kernel.shared_1[((threadIdx.x*72) + 29)]))
-            conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[93]*kernel.shared_1[((threadIdx.x*72) + 30)]))
-            conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[94]*kernel.shared_1[((threadIdx.x*72) + 31)]))
-            conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[95]*kernel.shared_1[((threadIdx.x*72) + 32)]))
-            conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[96]*kernel.shared_1[((threadIdx.x*72) + 33)]))
-            conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[97]*kernel.shared_1[((threadIdx.x*72) + 34)]))
-            conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[98]*kernel.shared_1[((threadIdx.x*72) + 35)]))
-            conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[93]*kernel.shared_1[((threadIdx.x*72) + 27)]))
-            conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[94]*kernel.shared_1[((threadIdx.x*72) + 28)]))
-            conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[95]*kernel.shared_1[((threadIdx.x*72) + 29)]))
-            conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[96]*kernel.shared_1[((threadIdx.x*72) + 30)]))
-            conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[97]*kernel.shared_1[((threadIdx.x*72) + 31)]))
-            conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[98]*kernel.shared_1[((threadIdx.x*72) + 32)]))
-            conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[99]*kernel.shared_1[((threadIdx.x*72) + 33)]))
-            conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[100]*kernel.shared_1[((threadIdx.x*72) + 34)]))
-            conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[101]*kernel.shared_1[((threadIdx.x*72) + 35)]))
-            conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[96]*kernel.shared_1[((threadIdx.x*72) + 27)]))
-            conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[97]*kernel.shared_1[((threadIdx.x*72) + 28)]))
-            conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[98]*kernel.shared_1[((threadIdx.x*72) + 29)]))
-            conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[99]*kernel.shared_1[((threadIdx.x*72) + 30)]))
-            conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[100]*kernel.shared_1[((threadIdx.x*72) + 31)]))
-            conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[101]*kernel.shared_1[((threadIdx.x*72) + 32)]))
-            conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[102]*kernel.shared_1[((threadIdx.x*72) + 33)]))
-            conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[103]*kernel.shared_1[((threadIdx.x*72) + 34)]))
-            conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[104]*kernel.shared_1[((threadIdx.x*72) + 35)]))
-            conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[99]*kernel.shared_1[((threadIdx.x*72) + 27)]))
-            conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[100]*kernel.shared_1[((threadIdx.x*72) + 28)]))
-            conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[101]*kernel.shared_1[((threadIdx.x*72) + 29)]))
-            conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[102]*kernel.shared_1[((threadIdx.x*72) + 30)]))
-            conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[103]*kernel.shared_1[((threadIdx.x*72) + 31)]))
-            conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[104]*kernel.shared_1[((threadIdx.x*72) + 32)]))
-            conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[105]*kernel.shared_1[((threadIdx.x*72) + 33)]))
-            conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[106]*kernel.shared_1[((threadIdx.x*72) + 34)]))
-            conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[107]*kernel.shared_1[((threadIdx.x*72) + 35)]))
-            conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[81]*kernel.shared_1[((threadIdx.x*72) + 63)]))
-            conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[82]*kernel.shared_1[((threadIdx.x*72) + 64)]))
-            conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[83]*kernel.shared_1[((threadIdx.x*72) + 65)]))
-            conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[84]*kernel.shared_1[((threadIdx.x*72) + 66)]))
-            conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[85]*kernel.shared_1[((threadIdx.x*72) + 67)]))
-            conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[86]*kernel.shared_1[((threadIdx.x*72) + 68)]))
-            conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[87]*kernel.shared_1[((threadIdx.x*72) + 69)]))
-            conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[88]*kernel.shared_1[((threadIdx.x*72) + 70)]))
-            conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[89]*kernel.shared_1[((threadIdx.x*72) + 71)]))
-            conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[84]*kernel.shared_1[((threadIdx.x*72) + 63)]))
-            conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[85]*kernel.shared_1[((threadIdx.x*72) + 64)]))
-            conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[86]*kernel.shared_1[((threadIdx.x*72) + 65)]))
-            conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[87]*kernel.shared_1[((threadIdx.x*72) + 66)]))
-            conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[88]*kernel.shared_1[((threadIdx.x*72) + 67)]))
-            conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[89]*kernel.shared_1[((threadIdx.x*72) + 68)]))
-            conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[90]*kernel.shared_1[((threadIdx.x*72) + 69)]))
-            conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[91]*kernel.shared_1[((threadIdx.x*72) + 70)]))
-            conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[92]*kernel.shared_1[((threadIdx.x*72) + 71)]))
-            conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[87]*kernel.shared_1[((threadIdx.x*72) + 63)]))
-            conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[88]*kernel.shared_1[((threadIdx.x*72) + 64)]))
-            conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[89]*kernel.shared_1[((threadIdx.x*72) + 65)]))
-            conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[90]*kernel.shared_1[((threadIdx.x*72) + 66)]))
-            conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[91]*kernel.shared_1[((threadIdx.x*72) + 67)]))
-            conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[92]*kernel.shared_1[((threadIdx.x*72) + 68)]))
-            conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[93]*kernel.shared_1[((threadIdx.x*72) + 69)]))
-            conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[94]*kernel.shared_1[((threadIdx.x*72) + 70)]))
-            conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[95]*kernel.shared_1[((threadIdx.x*72) + 71)]))
-            conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[90]*kernel.shared_1[((threadIdx.x*72) + 63)]))
-            conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[91]*kernel.shared_1[((threadIdx.x*72) + 64)]))
-            conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[92]*kernel.shared_1[((threadIdx.x*72) + 65)]))
-            conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[93]*kernel.shared_1[((threadIdx.x*72) + 66)]))
-            conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[94]*kernel.shared_1[((threadIdx.x*72) + 67)]))
-            conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[95]*kernel.shared_1[((threadIdx.x*72) + 68)]))
-            conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[96]*kernel.shared_1[((threadIdx.x*72) + 69)]))
-            conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[97]*kernel.shared_1[((threadIdx.x*72) + 70)]))
-            conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[98]*kernel.shared_1[((threadIdx.x*72) + 71)]))
-            conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[93]*kernel.shared_1[((threadIdx.x*72) + 63)]))
-            conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[94]*kernel.shared_1[((threadIdx.x*72) + 64)]))
-            conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[95]*kernel.shared_1[((threadIdx.x*72) + 65)]))
-            conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[96]*kernel.shared_1[((threadIdx.x*72) + 66)]))
-            conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[97]*kernel.shared_1[((threadIdx.x*72) + 67)]))
-            conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[98]*kernel.shared_1[((threadIdx.x*72) + 68)]))
-            conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[99]*kernel.shared_1[((threadIdx.x*72) + 69)]))
-            conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[100]*kernel.shared_1[((threadIdx.x*72) + 70)]))
-            conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[101]*kernel.shared_1[((threadIdx.x*72) + 71)]))
-            conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[96]*kernel.shared_1[((threadIdx.x*72) + 63)]))
-            conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[97]*kernel.shared_1[((threadIdx.x*72) + 64)]))
-            conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[98]*kernel.shared_1[((threadIdx.x*72) + 65)]))
-            conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[99]*kernel.shared_1[((threadIdx.x*72) + 66)]))
-            conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[100]*kernel.shared_1[((threadIdx.x*72) + 67)]))
-            conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[101]*kernel.shared_1[((threadIdx.x*72) + 68)]))
-            conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[102]*kernel.shared_1[((threadIdx.x*72) + 69)]))
-            conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[103]*kernel.shared_1[((threadIdx.x*72) + 70)]))
-            conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[104]*kernel.shared_1[((threadIdx.x*72) + 71)]))
-            conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[99]*kernel.shared_1[((threadIdx.x*72) + 63)]))
-            conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[100]*kernel.shared_1[((threadIdx.x*72) + 64)]))
-            conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[101]*kernel.shared_1[((threadIdx.x*72) + 65)]))
-            conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[102]*kernel.shared_1[((threadIdx.x*72) + 66)]))
-            conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[103]*kernel.shared_1[((threadIdx.x*72) + 67)]))
-            conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[104]*kernel.shared_1[((threadIdx.x*72) + 68)]))
-            conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[105]*kernel.shared_1[((threadIdx.x*72) + 69)]))
-            conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[106]*kernel.shared_1[((threadIdx.x*72) + 70)]))
-            conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[107]*kernel.shared_1[((threadIdx.x*72) + 71)]))
           }
         }
         for (i1.inner: int32, 0, 2) {
-          for (i2.inner: int32, 0, 7) {
-            compute[(((((floordiv(blockIdx.x, 7)*1568) + (threadIdx.x*98)) + (i1.inner*49)) + (i2.inner*7)) + floormod(blockIdx.x, 7))] = max((conv2d_nchw_1[((i1.inner*7) + i2.inner)] + bias[(((floordiv(blockIdx.x, 7)*32) + (threadIdx.x*2)) + i1.inner)]), 0f32)
-          }
+          compute[(((blockIdx.x*196) + (i1.inner*49)) + threadIdx.x)] = max((conv2d_nchw_1[i1.inner] + bias[((blockIdx.x*4) + i1.inner)]), 0f32)
+          compute[((((blockIdx.x*196) + (i1.inner*49)) + threadIdx.x) + 7)] = max((conv2d_nchw_1[(i1.inner + 2)] + bias[((blockIdx.x*4) + i1.inner)]), 0f32)
+          compute[((((blockIdx.x*196) + (i1.inner*49)) + threadIdx.x) + 14)] = max((conv2d_nchw_1[(i1.inner + 4)] + bias[((blockIdx.x*4) + i1.inner)]), 0f32)
+          compute[((((blockIdx.x*196) + (i1.inner*49)) + threadIdx.x) + 21)] = max((conv2d_nchw_1[(i1.inner + 6)] + bias[((blockIdx.x*4) + i1.inner)]), 0f32)
+          compute[((((blockIdx.x*196) + (i1.inner*49)) + threadIdx.x) + 28)] = max((conv2d_nchw_1[(i1.inner + 8)] + bias[((blockIdx.x*4) + i1.inner)]), 0f32)
+          compute[((((blockIdx.x*196) + (i1.inner*49)) + threadIdx.x) + 35)] = max((conv2d_nchw_1[(i1.inner + 10)] + bias[((blockIdx.x*4) + i1.inner)]), 0f32)
+          compute[((((blockIdx.x*196) + (i1.inner*49)) + threadIdx.x) + 42)] = max((conv2d_nchw_1[(i1.inner + 12)] + bias[((blockIdx.x*4) + i1.inner)]), 0f32)
+          compute[((((blockIdx.x*196) + (i1.inner*49)) + threadIdx.x) + 98)] = max((conv2d_nchw_1[(i1.inner + 14)] + bias[(((blockIdx.x*4) + i1.inner) + 2)]), 0f32)
+          compute[((((blockIdx.x*196) + (i1.inner*49)) + threadIdx.x) + 105)] = max((conv2d_nchw_1[(i1.inner + 16)] + bias[(((blockIdx.x*4) + i1.inner) + 2)]), 0f32)
+          compute[((((blockIdx.x*196) + (i1.inner*49)) + threadIdx.x) + 112)] = max((conv2d_nchw_1[(i1.inner + 18)] + bias[(((blockIdx.x*4) + i1.inner) + 2)]), 0f32)
+          compute[((((blockIdx.x*196) + (i1.inner*49)) + threadIdx.x) + 119)] = max((conv2d_nchw_1[(i1.inner + 20)] + bias[(((blockIdx.x*4) + i1.inner) + 2)]), 0f32)
+          compute[((((blockIdx.x*196) + (i1.inner*49)) + threadIdx.x) + 126)] = max((conv2d_nchw_1[(i1.inner + 22)] + bias[(((blockIdx.x*4) + i1.inner) + 2)]), 0f32)
+          compute[((((blockIdx.x*196) + (i1.inner*49)) + threadIdx.x) + 133)] = max((conv2d_nchw_1[(i1.inner + 24)] + bias[(((blockIdx.x*4) + i1.inner) + 2)]), 0f32)
+          compute[((((blockIdx.x*196) + (i1.inner*49)) + threadIdx.x) + 140)] = max((conv2d_nchw_1[(i1.inner + 26)] + bias[(((blockIdx.x*4) + i1.inner) + 2)]), 0f32)
         }
       }
     }
@@ -987,7 +773,7 @@ We build the binary and check its correctness and performance.
 
  .. code-block:: none
 
-    Execution time of this operator: 0.361 ms
+    Execution time of this operator: 0.369 ms
 
 
 
@@ -1037,34 +823,34 @@ They can be used for debugging and learning the behavior of the auto-scheduler.
     conv2d_nchw_nn_o_o_o_o, conv2d_nchw_nn_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_nn_o_o_o_i, factor=1)
     conv2d_nchw_ff_o_i, conv2d_nchw_ff_i = s[conv2d_nchw].split(conv2d_nchw_ff, factor=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=1)
+    conv2d_nchw_ff_o_o_o_i, conv2d_nchw_ff_o_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_o_i, factor=1)
+    conv2d_nchw_ff_o_o_o_o, conv2d_nchw_ff_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_o_o_i, factor=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=7)
+    conv2d_nchw_yy_o_o_i, conv2d_nchw_yy_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_i, factor=1)
     conv2d_nchw_yy_o_o_o_i, conv2d_nchw_yy_o_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_o_i, factor=1)
-    conv2d_nchw_yy_o_o_o_o, conv2d_nchw_yy_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_o_o_i, factor=1)
+    conv2d_nchw_yy_o_o_o_o, conv2d_nchw_yy_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_o_o_i, factor=7)
     conv2d_nchw_xx_o_i, conv2d_nchw_xx_i = s[conv2d_nchw].split(conv2d_nchw_xx, factor=1)
     conv2d_nchw_xx_o_o_i, conv2d_nchw_xx_o_i = s[conv2d_nchw].split(conv2d_nchw_xx_o_i, factor=1)
-    conv2d_nchw_xx_o_o_o_i, conv2d_nchw_xx_o_o_i = s[conv2d_nchw].split(conv2d_nchw_xx_o_o_i, factor=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=4)
-    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=3)
+    conv2d_nchw_ry_o_i, conv2d_nchw_ry_i = s[conv2d_nchw].split(conv2d_nchw_ry, factor=1)
+    conv2d_nchw_ry_o_o, conv2d_nchw_ry_o_i = s[conv2d_nchw].split(conv2d_nchw_ry_o_i, factor=3)
+    conv2d_nchw_rx_o_i, conv2d_nchw_rx_i = s[conv2d_nchw].split(conv2d_nchw_rx, factor=1)
     conv2d_nchw_rx_o_o, conv2d_nchw_rx_o_i = s[conv2d_nchw].split(conv2d_nchw_rx_o_i, factor=1)
     s[conv2d_nchw].reorder(conv2d_nchw_nn_o_o_o_o, conv2d_nchw_ff_o_o_o_o, conv2d_nchw_yy_o_o_o_o, conv2d_nchw_xx_o_o_o_o, conv2d_nchw_nn_o_o_o_i, conv2d_nchw_ff_o_o_o_i, conv2d_nchw_yy_o_o_o_i, conv2d_nchw_xx_o_o_o_i, conv2d_nchw_nn_o_o_i, conv2d_nchw_ff_o_o_i, conv2d_nchw_yy_o_o_i, conv2d_nchw_xx_o_o_i, conv2d_nchw_rc_o_o, conv2d_nchw_ry_o_o, conv2d_nchw_rx_o_o, conv2d_nchw_rc_o_i, conv2d_nchw_ry_o_i, conv2d_nchw_rx_o_i, conv2d_nchw_nn_o_i, conv2d_nchw_ff_o_i, conv2d_nchw_yy_o_i, conv2 [...]
     compute_i0_o_i, compute_i0_i = s[compute].split(compute_i0, factor=1)
     compute_i0_o_o_i, compute_i0_o_i = s[compute].split(compute_i0_o_i, factor=1)
     compute_i0_o_o_o, compute_i0_o_o_i = s[compute].split(compute_i0_o_o_i, factor=1)
     compute_i1_o_i, compute_i1_i = s[compute].split(compute_i1, factor=2)
-    compute_i1_o_o_i, compute_i1_o_i = s[compute].split(compute_i1_o_i, factor=16)
-    compute_i1_o_o_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=7)
+    compute_i1_o_o_i, compute_i1_o_i = s[compute].split(compute_i1_o_i, factor=1)
+    compute_i1_o_o_o, compute_i1_o_o_i = s[compute].split(compute_i1_o_o_i, factor=2)
+    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=1)
-    compute_i2_o_o_o, compute_i2_o_o_i = s[compute].split(compute_i2_o_o_i, factor=1)
+    compute_i2_o_o_o, compute_i2_o_o_i = s[compute].split(compute_i2_o_o_i, factor=7)
     compute_i3_o_i, compute_i3_i = s[compute].split(compute_i3, factor=1)
-    compute_i3_o_o_i, compute_i3_o_i = s[compute].split(compute_i3_o_i, factor=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)
@@ -1084,14 +870,14 @@ 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=16)
+    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=7)
     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=16)
+    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=7)
     s[pad_temp_shared].bind(pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i, te.thread_axis("threadIdx.x"))
-    s[conv2d_nchw].pragma(conv2d_nchw_nn_o_o_o_o, "auto_unroll_max_step", 512)
+    s[conv2d_nchw].pragma(conv2d_nchw_nn_o_o_o_o, "auto_unroll_max_step", 1024)
     s[conv2d_nchw].pragma(conv2d_nchw_nn_o_o_o_o, "unroll_explicit", True)
 
     CUDA source code:
@@ -1109,617 +895,440 @@ 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__(16) default_function_kernel0(float* __restrict__ data, float* __restrict__ kernel, float* __restrict__ compute, float* __restrict__ bias) {
-      float conv2d_nchw[14];
-      __shared__ float pad_temp_shared[108];
-      __shared__ float kernel_shared[1152];
+    extern "C" __global__ void __launch_bounds__(7) default_function_kernel0(float* __restrict__ data, float* __restrict__ kernel, float* __restrict__ compute, float* __restrict__ bias) {
+      float conv2d_nchw[28];
+      __shared__ float pad_temp_shared[252];
+      __shared__ float kernel_shared[48];
       conv2d_nchw[0] = 0.000000e+00f;
-      conv2d_nchw[1] = 0.000000e+00f;
       conv2d_nchw[2] = 0.000000e+00f;
-      conv2d_nchw[3] = 0.000000e+00f;
       conv2d_nchw[4] = 0.000000e+00f;
-      conv2d_nchw[5] = 0.000000e+00f;
       conv2d_nchw[6] = 0.000000e+00f;
-      conv2d_nchw[7] = 0.000000e+00f;
       conv2d_nchw[8] = 0.000000e+00f;
-      conv2d_nchw[9] = 0.000000e+00f;
       conv2d_nchw[10] = 0.000000e+00f;
-      conv2d_nchw[11] = 0.000000e+00f;
       conv2d_nchw[12] = 0.000000e+00f;
+      conv2d_nchw[14] = 0.000000e+00f;
+      conv2d_nchw[16] = 0.000000e+00f;
+      conv2d_nchw[18] = 0.000000e+00f;
+      conv2d_nchw[20] = 0.000000e+00f;
+      conv2d_nchw[22] = 0.000000e+00f;
+      conv2d_nchw[24] = 0.000000e+00f;
+      conv2d_nchw[26] = 0.000000e+00f;
+      conv2d_nchw[1] = 0.000000e+00f;
+      conv2d_nchw[3] = 0.000000e+00f;
+      conv2d_nchw[5] = 0.000000e+00f;
+      conv2d_nchw[7] = 0.000000e+00f;
+      conv2d_nchw[9] = 0.000000e+00f;
+      conv2d_nchw[11] = 0.000000e+00f;
       conv2d_nchw[13] = 0.000000e+00f;
+      conv2d_nchw[15] = 0.000000e+00f;
+      conv2d_nchw[17] = 0.000000e+00f;
+      conv2d_nchw[19] = 0.000000e+00f;
+      conv2d_nchw[21] = 0.000000e+00f;
+      conv2d_nchw[23] = 0.000000e+00f;
+      conv2d_nchw[25] = 0.000000e+00f;
+      conv2d_nchw[27] = 0.000000e+00f;
       for (int rc_outer_outer = 0; rc_outer_outer < 128; ++rc_outer_outer) {
-        __syncthreads();
-        pad_temp_shared[((int)threadIdx.x)] = ((((3 <= ((int)threadIdx.x)) && (1 <= ((((int)blockIdx.x) % 7) + (((int)threadIdx.x) % 3)))) && (((((int)blockIdx.x) % 7) + (((int)threadIdx.x) % 3)) < 8)) ? data[(((((rc_outer_outer * 196) + ((((int)threadIdx.x) / 3) * 7)) + (((int)blockIdx.x) % 7)) + (((int)threadIdx.x) % 3)) - 8)] : 0.000000e+00f);
-        pad_temp_shared[(((int)threadIdx.x) + 16)] = (((((3 <= ((((int)threadIdx.x) + 16) % 27)) && (((((int)threadIdx.x) + 16) % 27) < 24)) && (1 <= ((((int)blockIdx.x) % 7) + ((((int)threadIdx.x) + 1) % 3)))) && (((((int)blockIdx.x) % 7) + ((((int)threadIdx.x) + 1) % 3)) < 8)) ? data[((((((rc_outer_outer * 196) + (((((int)threadIdx.x) + 16) / 27) * 49)) + ((((((int)threadIdx.x) + 16) % 27) / 3) * 7)) + (((int)blockIdx.x) % 7)) + ((((int)threadIdx.x) + 1) % 3)) - 8)] : 0.000000e+00f);
-        pad_temp_shared[(((int)threadIdx.x) + 32)] = (((1 <= ((((int)blockIdx.x) % 7) + ((((int)threadIdx.x) + 2) % 3))) && (((((int)blockIdx.x) % 7) + ((((int)threadIdx.x) + 2) % 3)) < 8)) ? data[((((((rc_outer_outer * 196) + (((((int)threadIdx.x) + 32) / 27) * 49)) + (((((int)threadIdx.x) + 5) / 3) * 7)) + (((int)blockIdx.x) % 7)) + ((((int)threadIdx.x) + 2) % 3)) - 8)] : 0.000000e+00f);
-        pad_temp_shared[(((int)threadIdx.x) + 48)] = (((((1 <= (((((int)threadIdx.x) / 3) + 7) % 9)) && (((((int)threadIdx.x) + 21) % 27) < 24)) && (1 <= ((((int)blockIdx.x) % 7) + (((int)threadIdx.x) % 3)))) && (((((int)blockIdx.x) % 7) + (((int)threadIdx.x) % 3)) < 8)) ? data[((((((rc_outer_outer * 196) + (((((int)threadIdx.x) + 48) / 27) * 49)) + ((((((int)threadIdx.x) / 3) + 7) % 9) * 7)) + (((int)blockIdx.x) % 7)) + (((int)threadIdx.x) % 3)) - 8)] : 0.000000e+00f);
-        pad_temp_shared[(((int)threadIdx.x) + 64)] = ((((((int)threadIdx.x) < 14) && (1 <= ((((int)blockIdx.x) % 7) + ((((int)threadIdx.x) + 1) % 3)))) && (((((int)blockIdx.x) % 7) + ((((int)threadIdx.x) + 1) % 3)) < 8)) ? data[((((((rc_outer_outer * 196) + (((((int)threadIdx.x) + 64) / 27) * 49)) + (((((int)threadIdx.x) + 10) / 3) * 7)) + (((int)blockIdx.x) % 7)) + ((((int)threadIdx.x) + 1) % 3)) - 8)] : 0.000000e+00f);
-        pad_temp_shared[(((int)threadIdx.x) + 80)] = (((((3 <= ((((int)threadIdx.x) + 26) % 27)) && (((((int)threadIdx.x) + 26) % 27) < 24)) && (1 <= ((((int)blockIdx.x) % 7) + ((((int)threadIdx.x) + 2) % 3)))) && (((((int)blockIdx.x) % 7) + ((((int)threadIdx.x) + 2) % 3)) < 8)) ? data[((((((rc_outer_outer * 196) + (((((int)threadIdx.x) + 80) / 27) * 49)) + ((((((int)threadIdx.x) + 26) % 27) / 3) * 7)) + (((int)blockIdx.x) % 7)) + ((((int)threadIdx.x) + 2) % 3)) - 8)] : 0.000000e+00f);
-        if (((int)threadIdx.x) < 12) {
-          pad_temp_shared[(((int)threadIdx.x) + 96)] = ((((((int)threadIdx.x) < 9) && (1 <= ((((int)blockIdx.x) % 7) + (((int)threadIdx.x) % 3)))) && (((((int)blockIdx.x) % 7) + (((int)threadIdx.x) % 3)) < 8)) ? data[((((((rc_outer_outer * 196) + (((((int)threadIdx.x) + 96) / 27) * 49)) + ((((int)threadIdx.x) / 3) * 7)) + (((int)blockIdx.x) % 7)) + (((int)threadIdx.x) % 3)) + 27)] : 0.000000e+00f);
+        for (int rx_outer_outer = 0; rx_outer_outer < 3; ++rx_outer_outer) {
+          __syncthreads();
+          pad_temp_shared[((int)threadIdx.x)] = 0.000000e+00f;
+          pad_temp_shared[(((int)threadIdx.x) + 7)] = (((1 <= (((int)threadIdx.x) + rx_outer_outer)) && ((((int)threadIdx.x) + rx_outer_outer) < 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) - 1)] : 0.000000e+00f);
+          pad_temp_shared[(((int)threadIdx.x) + 14)] = (((1 <= (((int)threadIdx.x) + rx_outer_outer)) && ((((int)threadIdx.x) + rx_outer_outer) < 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 6)] : 0.000000e+00f);
+          pad_temp_shared[(((int)threadIdx.x) + 21)] = (((1 <= (((int)threadIdx.x) + rx_outer_outer)) && ((((int)threadIdx.x) + rx_outer_outer) < 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 13)] : 0.000000e+00f);
+          pad_temp_shared[(((int)threadIdx.x) + 28)] = (((1 <= (((int)threadIdx.x) + rx_outer_outer)) && ((((int)threadIdx.x) + rx_outer_outer) < 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 20)] : 0.000000e+00f);
+          pad_temp_shared[(((int)threadIdx.x) + 35)] = (((1 <= (((int)threadIdx.x) + rx_outer_outer)) && ((((int)threadIdx.x) + rx_outer_outer) < 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 27)] : 0.000000e+00f);
+          pad_temp_shared[(((int)threadIdx.x) + 42)] = (((1 <= (((int)threadIdx.x) + rx_outer_outer)) && ((((int)threadIdx.x) + rx_outer_outer) < 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 34)] : 0.000000e+00f);
+          pad_temp_shared[(((int)threadIdx.x) + 49)] = (((1 <= (((int)threadIdx.x) + rx_outer_outer)) && ((((int)threadIdx.x) + rx_outer_outer) < 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 41)] : 0.000000e+00f);
+          pad_temp_shared[(((int)threadIdx.x) + 56)] = 0.000000e+00f;
+          pad_temp_shared[(((int)threadIdx.x) + 63)] = 0.000000e+00f;
+          pad_temp_shared[(((int)threadIdx.x) + 70)] = (((1 <= (((int)threadIdx.x) + rx_outer_outer)) && ((((int)threadIdx.x) + rx_outer_outer) < 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 48)] : 0.000000e+00f);
+          pad_temp_shared[(((int)threadIdx.x) + 77)] = (((1 <= (((int)threadIdx.x) + rx_outer_outer)) && ((((int)threadIdx.x) + rx_outer_outer) < 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 55)] : 0.000000e+00f);
+          pad_temp_shared[(((int)threadIdx.x) + 84)] = (((1 <= (((int)threadIdx.x) + rx_outer_outer)) && ((((int)threadIdx.x) + rx_outer_outer) < 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 62)] : 0.000000e+00f);
+          pad_temp_shared[(((int)threadIdx.x) + 91)] = (((1 <= (((int)threadIdx.x) + rx_outer_outer)) && ((((int)threadIdx.x) + rx_outer_outer) < 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 69)] : 0.000000e+00f);
+          pad_temp_shared[(((int)threadIdx.x) + 98)] = (((1 <= (((int)threadIdx.x) + rx_outer_outer)) && ((((int)threadIdx.x) + rx_outer_outer) < 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 76)] : 0.000000e+00f);
+          pad_temp_shared[(((int)threadIdx.x) + 105)] = (((1 <= (((int)threadIdx.x) + rx_outer_outer)) && ((((int)threadIdx.x) + rx_outer_outer) < 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 83)] : 0.000000e+00f);
+          pad_temp_shared[(((int)threadIdx.x) + 112)] = (((1 <= (((int)threadIdx.x) + rx_outer_outer)) && ((((int)threadIdx.x) + rx_outer_outer) < 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 90)] : 0.000000e+00f);
+          pad_temp_shared[(((int)threadIdx.x) + 119)] = 0.000000e+00f;
+          pad_temp_shared[(((int)threadIdx.x) + 126)] = 0.000000e+00f;
+          pad_temp_shared[(((int)threadIdx.x) + 133)] = (((1 <= (((int)threadIdx.x) + rx_outer_outer)) && ((((int)threadIdx.x) + rx_outer_outer) < 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 97)] : 0.000000e+00f);
+          pad_temp_shared[(((int)threadIdx.x) + 140)] = (((1 <= (((int)threadIdx.x) + rx_outer_outer)) && ((((int)threadIdx.x) + rx_outer_outer) < 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 104)] : 0.000000e+00f);
+          pad_temp_shared[(((int)threadIdx.x) + 147)] = (((1 <= (((int)threadIdx.x) + rx_outer_outer)) && ((((int)threadIdx.x) + rx_outer_outer) < 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 111)] : 0.000000e+00f);
+          pad_temp_shared[(((int)threadIdx.x) + 154)] = (((1 <= (((int)threadIdx.x) + rx_outer_outer)) && ((((int)threadIdx.x) + rx_outer_outer) < 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 118)] : 0.000000e+00f);
+          pad_temp_shared[(((int)threadIdx.x) + 161)] = (((1 <= (((int)threadIdx.x) + rx_outer_outer)) && ((((int)threadIdx.x) + rx_outer_outer) < 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 125)] : 0.000000e+00f);
+          pad_temp_shared[(((int)threadIdx.x) + 168)] = (((1 <= (((int)threadIdx.x) + rx_outer_outer)) && ((((int)threadIdx.x) + rx_outer_outer) < 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 132)] : 0.000000e+00f);
+          pad_temp_shared[(((int)threadIdx.x) + 175)] = (((1 <= (((int)threadIdx.x) + rx_outer_outer)) && ((((int)threadIdx.x) + rx_outer_outer) < 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 139)] : 0.000000e+00f);
+          pad_temp_shared[(((int)threadIdx.x) + 182)] = 0.000000e+00f;
+          pad_temp_shared[(((int)threadIdx.x) + 189)] = 0.000000e+00f;
+          pad_temp_shared[(((int)threadIdx.x) + 196)] = (((1 <= (((int)threadIdx.x) + rx_outer_outer)) && ((((int)threadIdx.x) + rx_outer_outer) < 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 146)] : 0.000000e+00f);
+          pad_temp_shared[(((int)threadIdx.x) + 203)] = (((1 <= (((int)threadIdx.x) + rx_outer_outer)) && ((((int)threadIdx.x) + rx_outer_outer) < 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 153)] : 0.000000e+00f);
+          pad_temp_shared[(((int)threadIdx.x) + 210)] = (((1 <= (((int)threadIdx.x) + rx_outer_outer)) && ((((int)threadIdx.x) + rx_outer_outer) < 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 160)] : 0.000000e+00f);
+          pad_temp_shared[(((int)threadIdx.x) + 217)] = (((1 <= (((int)threadIdx.x) + rx_outer_outer)) && ((((int)threadIdx.x) + rx_outer_outer) < 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 167)] : 0.000000e+00f);
+          pad_temp_shared[(((int)threadIdx.x) + 224)] = (((1 <= (((int)threadIdx.x) + rx_outer_outer)) && ((((int)threadIdx.x) + rx_outer_outer) < 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 174)] : 0.000000e+00f);
+          pad_temp_shared[(((int)threadIdx.x) + 231)] = (((1 <= (((int)threadIdx.x) + rx_outer_outer)) && ((((int)threadIdx.x) + rx_outer_outer) < 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 181)] : 0.000000e+00f);
+          pad_temp_shared[(((int)threadIdx.x) + 238)] = (((1 <= (((int)threadIdx.x) + rx_outer_outer)) && ((((int)threadIdx.x) + rx_outer_outer) < 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 188)] : 0.000000e+00f);
+          pad_temp_shared[(((int)threadIdx.x) + 245)] = 0.000000e+00f;
+          kernel_shared[((int)threadIdx.x)] = kernel[((((((int)blockIdx.x) * 18432) + (rc_outer_outer * 36)) + (((int)threadIdx.x) * 3)) + rx_outer_outer)];
+          kernel_shared[(((int)threadIdx.x) + 7)] = kernel[((((((((int)blockIdx.x) * 18432) + (((((int)threadIdx.x) + 7) / 12) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 7) % 12) / 3) * 9)) + (((((int)threadIdx.x) + 1) % 3) * 3)) + rx_outer_outer)];
+          kernel_shared[(((int)threadIdx.x) + 14)] = kernel[(((((((int)blockIdx.x) * 18432) + (((((int)threadIdx.x) + 14) / 12) * 4608)) + (rc_outer_outer * 36)) + ((((int)threadIdx.x) + 2) * 3)) + rx_outer_outer)];
+          kernel_shared[(((int)threadIdx.x) + 21)] = kernel[((((((((int)blockIdx.x) * 18432) + (((((int)threadIdx.x) + 21) / 12) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) / 3) + 3) & 3) * 9)) + ((((int)threadIdx.x) % 3) * 3)) + rx_outer_outer)];
+          kernel_shared[(((int)threadIdx.x) + 28)] = kernel[((((((((int)blockIdx.x) * 18432) + (((((int)threadIdx.x) + 28) / 12) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 4) / 3) * 9)) + (((((int)threadIdx.x) + 1) % 3) * 3)) + rx_outer_outer)];
+          kernel_shared[(((int)threadIdx.x) + 35)] = kernel[((((((((int)blockIdx.x) * 18432) + (((((int)threadIdx.x) + 35) / 12) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 11) % 12) / 3) * 9)) + (((((int)threadIdx.x) + 2) % 3) * 3)) + rx_outer_outer)];
+          if (((int)threadIdx.x) < 6) {
+            kernel_shared[(((int)threadIdx.x) + 42)] = kernel[((((((((int)blockIdx.x) * 18432) + (((((int)threadIdx.x) + 42) / 12) * 4608)) + (rc_outer_outer * 36)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 18)];
+          }
+          __syncthreads();
+          conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((int)threadIdx.x)] * kernel_shared[0]));
+          conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((int)threadIdx.x) + 7)] * kernel_shared[0]));
+          conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((int)threadIdx.x) + 14)] * kernel_shared[0]));
+          conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((int)threadIdx.x) + 21)] * kernel_shared[0]));
+          conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((int)threadIdx.x) + 28)] * kernel_shared[0]));
+          conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((int)threadIdx.x) + 35)] * kernel_shared[0]));
+          conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((int)threadIdx.x) + 42)] * kernel_shared[0]));
+          conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[((int)threadIdx.x)] * kernel_shared[24]));
+          conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((int)threadIdx.x) + 7)] * kernel_shared[24]));
+          conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((int)threadIdx.x) + 14)] * kernel_shared[24]));
+          conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((int)threadIdx.x) + 21)] * kernel_shared[24]));
+          conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((int)threadIdx.x) + 28)] * kernel_shared[24]));
+          conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((int)threadIdx.x) + 35)] * kernel_shared[24]));
+          conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((int)threadIdx.x) + 42)] * kernel_shared[24]));
+          conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((int)threadIdx.x)] * kernel_shared[12]));
+          conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((int)threadIdx.x) + 7)] * kernel_shared[12]));
+          conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((int)threadIdx.x) + 14)] * kernel_shared[12]));
+          conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((int)threadIdx.x) + 21)] * kernel_shared[12]));
+          conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((int)threadIdx.x) + 28)] * kernel_shared[12]));
+          conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((int)threadIdx.x) + 35)] * kernel_shared[12]));
+          conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((int)threadIdx.x) + 42)] * kernel_shared[12]));
+          conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[((int)threadIdx.x)] * kernel_shared[36]));
+          conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((int)threadIdx.x) + 7)] * kernel_shared[36]));
+          conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((int)threadIdx.x) + 14)] * kernel_shared[36]));
+          conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((int)threadIdx.x) + 21)] * kernel_shared[36]));
+          conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((int)threadIdx.x) + 28)] * kernel_shared[36]));
+          conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((int)threadIdx.x) + 35)] * kernel_shared[36]));
+          conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((int)threadIdx.x) + 42)] * kernel_shared[36]));
+          conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((int)threadIdx.x) + 7)] * kernel_shared[1]));
+          conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((int)threadIdx.x) + 14)] * kernel_shared[1]));
+          conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((int)threadIdx.x) + 21)] * kernel_shared[1]));
+          conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((int)threadIdx.x) + 28)] * kernel_shared[1]));
+          conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((int)threadIdx.x) + 35)] * kernel_shared[1]));
+          conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((int)threadIdx.x) + 42)] * kernel_shared[1]));
+          conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((int)threadIdx.x) + 49)] * kernel_shared[1]));
+          conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((int)threadIdx.x) + 7)] * kernel_shared[25]));
+          conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((int)threadIdx.x) + 14)] * kernel_shared[25]));
+          conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((int)threadIdx.x) + 21)] * kernel_shared[25]));
+          conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((int)threadIdx.x) + 28)] * kernel_shared[25]));
+          conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((int)threadIdx.x) + 35)] * kernel_shared[25]));
+          conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((int)threadIdx.x) + 42)] * kernel_shared[25]));
+          conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((int)threadIdx.x) + 49)] * kernel_shared[25]));
+          conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((int)threadIdx.x) + 7)] * kernel_shared[13]));
+          conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((int)threadIdx.x) + 14)] * kernel_shared[13]));
+          conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((int)threadIdx.x) + 21)] * kernel_shared[13]));
+          conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((int)threadIdx.x) + 28)] * kernel_shared[13]));
+          conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((int)threadIdx.x) + 35)] * kernel_shared[13]));
+          conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((int)threadIdx.x) + 42)] * kernel_shared[13]));
+          conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((int)threadIdx.x) + 49)] * kernel_shared[13]));
+          conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((int)threadIdx.x) + 7)] * kernel_shared[37]));
+          conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((int)threadIdx.x) + 14)] * kernel_shared[37]));
+          conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((int)threadIdx.x) + 21)] * kernel_shared[37]));
+          conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((int)threadIdx.x) + 28)] * kernel_shared[37]));
+          conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((int)threadIdx.x) + 35)] * kernel_shared[37]));
+          conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((int)threadIdx.x) + 42)] * kernel_shared[37]));
+          conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((int)threadIdx.x) + 49)] * kernel_shared[37]));
+          conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((int)threadIdx.x) + 14)] * kernel_shared[2]));
+          conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((int)threadIdx.x) + 21)] * kernel_shared[2]));
+          conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((int)threadIdx.x) + 28)] * kernel_shared[2]));
+          conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((int)threadIdx.x) + 35)] * kernel_shared[2]));
+          conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((int)threadIdx.x) + 42)] * kernel_shared[2]));
+          conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((int)threadIdx.x) + 49)] * kernel_shared[2]));
+          conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((int)threadIdx.x) + 56)] * kernel_shared[2]));
+          conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((int)threadIdx.x) + 14)] * kernel_shared[26]));
+          conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((int)threadIdx.x) + 21)] * kernel_shared[26]));
+          conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((int)threadIdx.x) + 28)] * kernel_shared[26]));
+          conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((int)threadIdx.x) + 35)] * kernel_shared[26]));
+          conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((int)threadIdx.x) + 42)] * kernel_shared[26]));
+          conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((int)threadIdx.x) + 49)] * kernel_shared[26]));
+          conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((int)threadIdx.x) + 56)] * kernel_shared[26]));
+          conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((int)threadIdx.x) + 14)] * kernel_shared[14]));
+          conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((int)threadIdx.x) + 21)] * kernel_shared[14]));
+          conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((int)threadIdx.x) + 28)] * kernel_shared[14]));
+          conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((int)threadIdx.x) + 35)] * kernel_shared[14]));
+          conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((int)threadIdx.x) + 42)] * kernel_shared[14]));
+          conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((int)threadIdx.x) + 49)] * kernel_shared[14]));
+          conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((int)threadIdx.x) + 56)] * kernel_shared[14]));
+          conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((int)threadIdx.x) + 14)] * kernel_shared[38]));
+          conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((int)threadIdx.x) + 21)] * kernel_shared[38]));
+          conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((int)threadIdx.x) + 28)] * kernel_shared[38]));
+          conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((int)threadIdx.x) + 35)] * kernel_shared[38]));
+          conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((int)threadIdx.x) + 42)] * kernel_shared[38]));
+          conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((int)threadIdx.x) + 49)] * kernel_shared[38]));
+          conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((int)threadIdx.x) + 56)] * kernel_shared[38]));
+          conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((int)threadIdx.x) + 63)] * kernel_shared[3]));
+          conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((int)threadIdx.x) + 70)] * kernel_shared[3]));
+          conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((int)threadIdx.x) + 77)] * kernel_shared[3]));
+          conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((int)threadIdx.x) + 84)] * kernel_shared[3]));
+          conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((int)threadIdx.x) + 91)] * kernel_shared[3]));
+          conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((int)threadIdx.x) + 98)] * kernel_shared[3]));
+          conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((int)threadIdx.x) + 105)] * kernel_shared[3]));
+          conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((int)threadIdx.x) + 63)] * kernel_shared[27]));
+          conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((int)threadIdx.x) + 70)] * kernel_shared[27]));
+          conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((int)threadIdx.x) + 77)] * kernel_shared[27]));
+          conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((int)threadIdx.x) + 84)] * kernel_shared[27]));
+          conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((int)threadIdx.x) + 91)] * kernel_shared[27]));
+          conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((int)threadIdx.x) + 98)] * kernel_shared[27]));
+          conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((int)threadIdx.x) + 105)] * kernel_shared[27]));
+          conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((int)threadIdx.x) + 63)] * kernel_shared[15]));
+          conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((int)threadIdx.x) + 70)] * kernel_shared[15]));
+          conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((int)threadIdx.x) + 77)] * kernel_shared[15]));
+          conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((int)threadIdx.x) + 84)] * kernel_shared[15]));
+          conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((int)threadIdx.x) + 91)] * kernel_shared[15]));
+          conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((int)threadIdx.x) + 98)] * kernel_shared[15]));
+          conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((int)threadIdx.x) + 105)] * kernel_shared[15]));
+          conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((int)threadIdx.x) + 63)] * kernel_shared[39]));
+          conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((int)threadIdx.x) + 70)] * kernel_shared[39]));
+          conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((int)threadIdx.x) + 77)] * kernel_shared[39]));
+          conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((int)threadIdx.x) + 84)] * kernel_shared[39]));
+          conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((int)threadIdx.x) + 91)] * kernel_shared[39]));
+          conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((int)threadIdx.x) + 98)] * kernel_shared[39]));
+          conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((int)threadIdx.x) + 105)] * kernel_shared[39]));
+          conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((int)threadIdx.x) + 70)] * kernel_shared[4]));
+          conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((int)threadIdx.x) + 77)] * kernel_shared[4]));
+          conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((int)threadIdx.x) + 84)] * kernel_shared[4]));
+          conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((int)threadIdx.x) + 91)] * kernel_shared[4]));
+          conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((int)threadIdx.x) + 98)] * kernel_shared[4]));
+          conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((int)threadIdx.x) + 105)] * kernel_shared[4]));
+          conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((int)threadIdx.x) + 112)] * kernel_shared[4]));
+          conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((int)threadIdx.x) + 70)] * kernel_shared[28]));
+          conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((int)threadIdx.x) + 77)] * kernel_shared[28]));
+          conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((int)threadIdx.x) + 84)] * kernel_shared[28]));
+          conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((int)threadIdx.x) + 91)] * kernel_shared[28]));
+          conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((int)threadIdx.x) + 98)] * kernel_shared[28]));
+          conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((int)threadIdx.x) + 105)] * kernel_shared[28]));
+          conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((int)threadIdx.x) + 112)] * kernel_shared[28]));
+          conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((int)threadIdx.x) + 70)] * kernel_shared[16]));
+          conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((int)threadIdx.x) + 77)] * kernel_shared[16]));
+          conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((int)threadIdx.x) + 84)] * kernel_shared[16]));
+          conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((int)threadIdx.x) + 91)] * kernel_shared[16]));
+          conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((int)threadIdx.x) + 98)] * kernel_shared[16]));
+          conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((int)threadIdx.x) + 105)] * kernel_shared[16]));
+          conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((int)threadIdx.x) + 112)] * kernel_shared[16]));
+          conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((int)threadIdx.x) + 70)] * kernel_shared[40]));
+          conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((int)threadIdx.x) + 77)] * kernel_shared[40]));
+          conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((int)threadIdx.x) + 84)] * kernel_shared[40]));
+          conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((int)threadIdx.x) + 91)] * kernel_shared[40]));
+          conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((int)threadIdx.x) + 98)] * kernel_shared[40]));
+          conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((int)threadIdx.x) + 105)] * kernel_shared[40]));
+          conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((int)threadIdx.x) + 112)] * kernel_shared[40]));
+          conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((int)threadIdx.x) + 77)] * kernel_shared[5]));
+          conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((int)threadIdx.x) + 84)] * kernel_shared[5]));
+          conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((int)threadIdx.x) + 91)] * kernel_shared[5]));
+          conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((int)threadIdx.x) + 98)] * kernel_shared[5]));
+          conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((int)threadIdx.x) + 105)] * kernel_shared[5]));
+          conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((int)threadIdx.x) + 112)] * kernel_shared[5]));
+          conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((int)threadIdx.x) + 119)] * kernel_shared[5]));
+          conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((int)threadIdx.x) + 77)] * kernel_shared[29]));
+          conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((int)threadIdx.x) + 84)] * kernel_shared[29]));
+          conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((int)threadIdx.x) + 91)] * kernel_shared[29]));
+          conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((int)threadIdx.x) + 98)] * kernel_shared[29]));
+          conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((int)threadIdx.x) + 105)] * kernel_shared[29]));
+          conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((int)threadIdx.x) + 112)] * kernel_shared[29]));
+          conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((int)threadIdx.x) + 119)] * kernel_shared[29]));
+          conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((int)threadIdx.x) + 77)] * kernel_shared[17]));
+          conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((int)threadIdx.x) + 84)] * kernel_shared[17]));
+          conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((int)threadIdx.x) + 91)] * kernel_shared[17]));
+          conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((int)threadIdx.x) + 98)] * kernel_shared[17]));
+          conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((int)threadIdx.x) + 105)] * kernel_shared[17]));
+          conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((int)threadIdx.x) + 112)] * kernel_shared[17]));
+          conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((int)threadIdx.x) + 119)] * kernel_shared[17]));
+          conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((int)threadIdx.x) + 77)] * kernel_shared[41]));
+          conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((int)threadIdx.x) + 84)] * kernel_shared[41]));
+          conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((int)threadIdx.x) + 91)] * kernel_shared[41]));
+          conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((int)threadIdx.x) + 98)] * kernel_shared[41]));
+          conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((int)threadIdx.x) + 105)] * kernel_shared[41]));
+          conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((int)threadIdx.x) + 112)] * kernel_shared[41]));
+          conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((int)threadIdx.x) + 119)] * kernel_shared[41]));
+          conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((int)threadIdx.x) + 126)] * kernel_shared[6]));
+          conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((int)threadIdx.x) + 133)] * kernel_shared[6]));
+          conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((int)threadIdx.x) + 140)] * kernel_shared[6]));
+          conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((int)threadIdx.x) + 147)] * kernel_shared[6]));
+          conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((int)threadIdx.x) + 154)] * kernel_shared[6]));
+          conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((int)threadIdx.x) + 161)] * kernel_shared[6]));
+          conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((int)threadIdx.x) + 168)] * kernel_shared[6]));
+          conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((int)threadIdx.x) + 126)] * kernel_shared[30]));
+          conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((int)threadIdx.x) + 133)] * kernel_shared[30]));
+          conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((int)threadIdx.x) + 140)] * kernel_shared[30]));
+          conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((int)threadIdx.x) + 147)] * kernel_shared[30]));
+          conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((int)threadIdx.x) + 154)] * kernel_shared[30]));
+          conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((int)threadIdx.x) + 161)] * kernel_shared[30]));
+          conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((int)threadIdx.x) + 168)] * kernel_shared[30]));
+          conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((int)threadIdx.x) + 126)] * kernel_shared[18]));
+          conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((int)threadIdx.x) + 133)] * kernel_shared[18]));
+          conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((int)threadIdx.x) + 140)] * kernel_shared[18]));
+          conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((int)threadIdx.x) + 147)] * kernel_shared[18]));
+          conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((int)threadIdx.x) + 154)] * kernel_shared[18]));
+          conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((int)threadIdx.x) + 161)] * kernel_shared[18]));
+          conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((int)threadIdx.x) + 168)] * kernel_shared[18]));
+          conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((int)threadIdx.x) + 126)] * kernel_shared[42]));
+          conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((int)threadIdx.x) + 133)] * kernel_shared[42]));
+          conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((int)threadIdx.x) + 140)] * kernel_shared[42]));
+          conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((int)threadIdx.x) + 147)] * kernel_shared[42]));
+          conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((int)threadIdx.x) + 154)] * kernel_shared[42]));
+          conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((int)threadIdx.x) + 161)] * kernel_shared[42]));
+          conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((int)threadIdx.x) + 168)] * kernel_shared[42]));
+          conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((int)threadIdx.x) + 133)] * kernel_shared[7]));
+          conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((int)threadIdx.x) + 140)] * kernel_shared[7]));
+          conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((int)threadIdx.x) + 147)] * kernel_shared[7]));
+          conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((int)threadIdx.x) + 154)] * kernel_shared[7]));
+          conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((int)threadIdx.x) + 161)] * kernel_shared[7]));
+          conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((int)threadIdx.x) + 168)] * kernel_shared[7]));
+          conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((int)threadIdx.x) + 175)] * kernel_shared[7]));
+          conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((int)threadIdx.x) + 133)] * kernel_shared[31]));
+          conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((int)threadIdx.x) + 140)] * kernel_shared[31]));
+          conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((int)threadIdx.x) + 147)] * kernel_shared[31]));
+          conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((int)threadIdx.x) + 154)] * kernel_shared[31]));
+          conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((int)threadIdx.x) + 161)] * kernel_shared[31]));
+          conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((int)threadIdx.x) + 168)] * kernel_shared[31]));
+          conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((int)threadIdx.x) + 175)] * kernel_shared[31]));
+          conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((int)threadIdx.x) + 133)] * kernel_shared[19]));
+          conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((int)threadIdx.x) + 140)] * kernel_shared[19]));
+          conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((int)threadIdx.x) + 147)] * kernel_shared[19]));
+          conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((int)threadIdx.x) + 154)] * kernel_shared[19]));
+          conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((int)threadIdx.x) + 161)] * kernel_shared[19]));
+          conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((int)threadIdx.x) + 168)] * kernel_shared[19]));
+          conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((int)threadIdx.x) + 175)] * kernel_shared[19]));
+          conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((int)threadIdx.x) + 133)] * kernel_shared[43]));
+          conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((int)threadIdx.x) + 140)] * kernel_shared[43]));
+          conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((int)threadIdx.x) + 147)] * kernel_shared[43]));
+          conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((int)threadIdx.x) + 154)] * kernel_shared[43]));
+          conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((int)threadIdx.x) + 161)] * kernel_shared[43]));
+          conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((int)threadIdx.x) + 168)] * kernel_shared[43]));
+          conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((int)threadIdx.x) + 175)] * kernel_shared[43]));
+          conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((int)threadIdx.x) + 140)] * kernel_shared[8]));
+          conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((int)threadIdx.x) + 147)] * kernel_shared[8]));
+          conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((int)threadIdx.x) + 154)] * kernel_shared[8]));
+          conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((int)threadIdx.x) + 161)] * kernel_shared[8]));
+          conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((int)threadIdx.x) + 168)] * kernel_shared[8]));
+          conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((int)threadIdx.x) + 175)] * kernel_shared[8]));
+          conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((int)threadIdx.x) + 182)] * kernel_shared[8]));
+          conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((int)threadIdx.x) + 140)] * kernel_shared[32]));
+          conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((int)threadIdx.x) + 147)] * kernel_shared[32]));
+          conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((int)threadIdx.x) + 154)] * kernel_shared[32]));
+          conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((int)threadIdx.x) + 161)] * kernel_shared[32]));
+          conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((int)threadIdx.x) + 168)] * kernel_shared[32]));
+          conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((int)threadIdx.x) + 175)] * kernel_shared[32]));
+          conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((int)threadIdx.x) + 182)] * kernel_shared[32]));
+          conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((int)threadIdx.x) + 140)] * kernel_shared[20]));
+          conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((int)threadIdx.x) + 147)] * kernel_shared[20]));
+          conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((int)threadIdx.x) + 154)] * kernel_shared[20]));
+          conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((int)threadIdx.x) + 161)] * kernel_shared[20]));
+          conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((int)threadIdx.x) + 168)] * kernel_shared[20]));
+          conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((int)threadIdx.x) + 175)] * kernel_shared[20]));
+          conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((int)threadIdx.x) + 182)] * kernel_shared[20]));
+          conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((int)threadIdx.x) + 140)] * kernel_shared[44]));
+          conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((int)threadIdx.x) + 147)] * kernel_shared[44]));
+          conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((int)threadIdx.x) + 154)] * kernel_shared[44]));
+          conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((int)threadIdx.x) + 161)] * kernel_shared[44]));
+          conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((int)threadIdx.x) + 168)] * kernel_shared[44]));
+          conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((int)threadIdx.x) + 175)] * kernel_shared[44]));
+          conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((int)threadIdx.x) + 182)] * kernel_shared[44]));
+          conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((int)threadIdx.x) + 189)] * kernel_shared[9]));
+          conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((int)threadIdx.x) + 196)] * kernel_shared[9]));
+          conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((int)threadIdx.x) + 203)] * kernel_shared[9]));
+          conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((int)threadIdx.x) + 210)] * kernel_shared[9]));
+          conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((int)threadIdx.x) + 217)] * kernel_shared[9]));
+          conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((int)threadIdx.x) + 224)] * kernel_shared[9]));
+          conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((int)threadIdx.x) + 231)] * kernel_shared[9]));
+          conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((int)threadIdx.x) + 189)] * kernel_shared[33]));
+          conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((int)threadIdx.x) + 196)] * kernel_shared[33]));
+          conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((int)threadIdx.x) + 203)] * kernel_shared[33]));
+          conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((int)threadIdx.x) + 210)] * kernel_shared[33]));
+          conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((int)threadIdx.x) + 217)] * kernel_shared[33]));
+          conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((int)threadIdx.x) + 224)] * kernel_shared[33]));
+          conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((int)threadIdx.x) + 231)] * kernel_shared[33]));
+          conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((int)threadIdx.x) + 189)] * kernel_shared[21]));
+          conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((int)threadIdx.x) + 196)] * kernel_shared[21]));
+          conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((int)threadIdx.x) + 203)] * kernel_shared[21]));
+          conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((int)threadIdx.x) + 210)] * kernel_shared[21]));
+          conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((int)threadIdx.x) + 217)] * kernel_shared[21]));
+          conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((int)threadIdx.x) + 224)] * kernel_shared[21]));
+          conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((int)threadIdx.x) + 231)] * kernel_shared[21]));
+          conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((int)threadIdx.x) + 189)] * kernel_shared[45]));
+          conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((int)threadIdx.x) + 196)] * kernel_shared[45]));
+          conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((int)threadIdx.x) + 203)] * kernel_shared[45]));
+          conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((int)threadIdx.x) + 210)] * kernel_shared[45]));
+          conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((int)threadIdx.x) + 217)] * kernel_shared[45]));
+          conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((int)threadIdx.x) + 224)] * kernel_shared[45]));
+          conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((int)threadIdx.x) + 231)] * kernel_shared[45]));
+          conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((int)threadIdx.x) + 196)] * kernel_shared[10]));
+          conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((int)threadIdx.x) + 203)] * kernel_shared[10]));
+          conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((int)threadIdx.x) + 210)] * kernel_shared[10]));
+          conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((int)threadIdx.x) + 217)] * kernel_shared[10]));
+          conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((int)threadIdx.x) + 224)] * kernel_shared[10]));
+          conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((int)threadIdx.x) + 231)] * kernel_shared[10]));
+          conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((int)threadIdx.x) + 238)] * kernel_shared[10]));
+          conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((int)threadIdx.x) + 196)] * kernel_shared[34]));
+          conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((int)threadIdx.x) + 203)] * kernel_shared[34]));
+          conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((int)threadIdx.x) + 210)] * kernel_shared[34]));
+          conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((int)threadIdx.x) + 217)] * kernel_shared[34]));
+          conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((int)threadIdx.x) + 224)] * kernel_shared[34]));
+          conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((int)threadIdx.x) + 231)] * kernel_shared[34]));
+          conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((int)threadIdx.x) + 238)] * kernel_shared[34]));
+          conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((int)threadIdx.x) + 196)] * kernel_shared[22]));
+          conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((int)threadIdx.x) + 203)] * kernel_shared[22]));
+          conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((int)threadIdx.x) + 210)] * kernel_shared[22]));
+          conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((int)threadIdx.x) + 217)] * kernel_shared[22]));
+          conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((int)threadIdx.x) + 224)] * kernel_shared[22]));
+          conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((int)threadIdx.x) + 231)] * kernel_shared[22]));
+          conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((int)threadIdx.x) + 238)] * kernel_shared[22]));
+          conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((int)threadIdx.x) + 196)] * kernel_shared[46]));
+          conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((int)threadIdx.x) + 203)] * kernel_shared[46]));
+          conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((int)threadIdx.x) + 210)] * kernel_shared[46]));
+          conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((int)threadIdx.x) + 217)] * kernel_shared[46]));
+          conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((int)threadIdx.x) + 224)] * kernel_shared[46]));
+          conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((int)threadIdx.x) + 231)] * kernel_shared[46]));
+          conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((int)threadIdx.x) + 238)] * kernel_shared[46]));
+          conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((int)threadIdx.x) + 203)] * kernel_shared[11]));
+          conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((int)threadIdx.x) + 210)] * kernel_shared[11]));
+          conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((int)threadIdx.x) + 217)] * kernel_shared[11]));
+          conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((int)threadIdx.x) + 224)] * kernel_shared[11]));
+          conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((int)threadIdx.x) + 231)] * kernel_shared[11]));
+          conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((int)threadIdx.x) + 238)] * kernel_shared[11]));
+          conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((int)threadIdx.x) + 245)] * kernel_shared[11]));
+          conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((int)threadIdx.x) + 203)] * kernel_shared[35]));
+          conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((int)threadIdx.x) + 210)] * kernel_shared[35]));
+          conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((int)threadIdx.x) + 217)] * kernel_shared[35]));
+          conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((int)threadIdx.x) + 224)] * kernel_shared[35]));
+          conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((int)threadIdx.x) + 231)] * kernel_shared[35]));
+          conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((int)threadIdx.x) + 238)] * kernel_shared[35]));
+          conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((int)threadIdx.x) + 245)] * kernel_shared[35]));
+          conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((int)threadIdx.x) + 203)] * kernel_shared[23]));
+          conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((int)threadIdx.x) + 210)] * kernel_shared[23]));
+          conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((int)threadIdx.x) + 217)] * kernel_shared[23]));
+          conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((int)threadIdx.x) + 224)] * kernel_shared[23]));
+          conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((int)threadIdx.x) + 231)] * kernel_shared[23]));
+          conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((int)threadIdx.x) + 238)] * kernel_shared[23]));
+          conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((int)threadIdx.x) + 245)] * kernel_shared[23]));
+          conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((int)threadIdx.x) + 203)] * kernel_shared[47]));
+          conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((int)threadIdx.x) + 210)] * kernel_shared[47]));
+          conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((int)threadIdx.x) + 217)] * kernel_shared[47]));
+          conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((int)threadIdx.x) + 224)] * kernel_shared[47]));
+          conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((int)threadIdx.x) + 231)] * kernel_shared[47]));
+          conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((int)threadIdx.x) + 238)] * kernel_shared[47]));
+          conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((int)threadIdx.x) + 245)] * kernel_shared[47]));
         }
-        kernel_shared[((int)threadIdx.x)] = kernel[((((((int)blockIdx.x) / 7) * 147456) + (rc_outer_outer * 36)) + ((int)threadIdx.x))];
-        kernel_shared[(((int)threadIdx.x) + 16)] = kernel[(((((((int)blockIdx.x) / 7) * 147456) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 16) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 32)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 32) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 32) % 36) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 48)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 48) / 36) * 4608)) + (rc_outer_outer * 36)) + ((int)threadIdx.x)) + 12)];
-        kernel_shared[(((int)threadIdx.x) + 64)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 64) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 28) % 36) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 80)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 80) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 8) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 96)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 96) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) / 3) + 8) % 12) * 3)) + (((int)threadIdx.x) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 112)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 112) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 4) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 128)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 128) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 20) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 144)] = kernel[(((((((int)blockIdx.x) / 7) * 147456) + (rc_outer_outer * 36)) + ((int)threadIdx.x)) + 18432)];
-        kernel_shared[(((int)threadIdx.x) + 160)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 160) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 16) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 176)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 176) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 32) % 36) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 192)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 192) / 36) * 4608)) + (rc_outer_outer * 36)) + ((int)threadIdx.x)) + 12)];
-        kernel_shared[(((int)threadIdx.x) + 208)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 208) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 28) % 36) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 224)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 224) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 8) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 240)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 240) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) / 3) + 8) % 12) * 3)) + (((int)threadIdx.x) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 256)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 256) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 4) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 272)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 272) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 20) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 288)] = kernel[(((((((int)blockIdx.x) / 7) * 147456) + (rc_outer_outer * 36)) + ((int)threadIdx.x)) + 36864)];
-        kernel_shared[(((int)threadIdx.x) + 304)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 304) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 16) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 320)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 320) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 32) % 36) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 336)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 336) / 36) * 4608)) + (rc_outer_outer * 36)) + ((int)threadIdx.x)) + 12)];
-        kernel_shared[(((int)threadIdx.x) + 352)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 352) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 28) % 36) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 368)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 368) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 8) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 384)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 384) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) / 3) + 8) % 12) * 3)) + (((int)threadIdx.x) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 400)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 400) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 4) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 416)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 416) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 20) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 432)] = kernel[(((((((int)blockIdx.x) / 7) * 147456) + (rc_outer_outer * 36)) + ((int)threadIdx.x)) + 55296)];
-        kernel_shared[(((int)threadIdx.x) + 448)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 448) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 16) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 464)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 464) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 32) % 36) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 480)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 480) / 36) * 4608)) + (rc_outer_outer * 36)) + ((int)threadIdx.x)) + 12)];
-        kernel_shared[(((int)threadIdx.x) + 496)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 496) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 28) % 36) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 512)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 512) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 8) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 528)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 528) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) / 3) + 8) % 12) * 3)) + (((int)threadIdx.x) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 544)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 544) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 4) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 560)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 560) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 20) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 576)] = kernel[(((((((int)blockIdx.x) / 7) * 147456) + (rc_outer_outer * 36)) + ((int)threadIdx.x)) + 73728)];
-        kernel_shared[(((int)threadIdx.x) + 592)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 592) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 16) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 608)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 608) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 32) % 36) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 624)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 624) / 36) * 4608)) + (rc_outer_outer * 36)) + ((int)threadIdx.x)) + 12)];
-        kernel_shared[(((int)threadIdx.x) + 640)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 640) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 28) % 36) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 656)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 656) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 8) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 672)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 672) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) / 3) + 8) % 12) * 3)) + (((int)threadIdx.x) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 688)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 688) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 4) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 704)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 704) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 20) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 720)] = kernel[(((((((int)blockIdx.x) / 7) * 147456) + (rc_outer_outer * 36)) + ((int)threadIdx.x)) + 92160)];
-        kernel_shared[(((int)threadIdx.x) + 736)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 736) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 16) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 752)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 752) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 32) % 36) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 768)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 768) / 36) * 4608)) + (rc_outer_outer * 36)) + ((int)threadIdx.x)) + 12)];
-        kernel_shared[(((int)threadIdx.x) + 784)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 784) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 28) % 36) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 800)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 800) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 8) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 816)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 816) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) / 3) + 8) % 12) * 3)) + (((int)threadIdx.x) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 832)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 832) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 4) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 848)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 848) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 20) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 864)] = kernel[(((((((int)blockIdx.x) / 7) * 147456) + (rc_outer_outer * 36)) + ((int)threadIdx.x)) + 110592)];
-        kernel_shared[(((int)threadIdx.x) + 880)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 880) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 16) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 896)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 896) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 32) % 36) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 912)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 912) / 36) * 4608)) + (rc_outer_outer * 36)) + ((int)threadIdx.x)) + 12)];
-        kernel_shared[(((int)threadIdx.x) + 928)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 928) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 28) % 36) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 944)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 944) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 8) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 960)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 960) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) / 3) + 8) % 12) * 3)) + (((int)threadIdx.x) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 976)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 976) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 4) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 992)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 992) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 20) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 1008)] = kernel[(((((((int)blockIdx.x) / 7) * 147456) + (rc_outer_outer * 36)) + ((int)threadIdx.x)) + 129024)];
-        kernel_shared[(((int)threadIdx.x) + 1024)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 1024) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 16) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 1040)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 1040) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 32) % 36) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 1056)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 1056) / 36) * 4608)) + (rc_outer_outer * 36)) + ((int)threadIdx.x)) + 12)];
-        kernel_shared[(((int)threadIdx.x) + 1072)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 1072) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 28) % 36) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 1088)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 1088) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 8) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 1104)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 1104) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) / 3) + 8) % 12) * 3)) + (((int)threadIdx.x) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 1120)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 1120) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 4) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-        kernel_shared[(((int)threadIdx.x) + 1136)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 1136) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 20) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-        __syncthreads();
-        conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[0] * kernel_shared[(((int)threadIdx.x) * 72)]));
-        conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[1] * kernel_shared[((((int)threadIdx.x) * 72) + 1)]));
-        conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[2] * kernel_shared[((((int)threadIdx.x) * 72) + 2)]));
-        conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[3] * kernel_shared[((((int)threadIdx.x) * 72) + 3)]));
-        conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[4] * kernel_shared[((((int)threadIdx.x) * 72) + 4)]));
-        conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[5] * kernel_shared[((((int)threadIdx.x) * 72) + 5)]));
-        conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[6] * kernel_shared[((((int)threadIdx.x) * 72) + 6)]));
-        conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[7] * kernel_shared[((((int)threadIdx.x) * 72) + 7)]));
-        conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[8] * kernel_shared[((((int)threadIdx.x) * 72) + 8)]));
-        conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[3] * kernel_shared[(((int)threadIdx.x) * 72)]));
-        conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[4] * kernel_shared[((((int)threadIdx.x) * 72) + 1)]));
-        conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[5] * kernel_shared[((((int)threadIdx.x) * 72) + 2)]));
-        conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[6] * kernel_shared[((((int)threadIdx.x) * 72) + 3)]));
-        conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[7] * kernel_shared[((((int)threadIdx.x) * 72) + 4)]));
-        conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[8] * kernel_shared[((((int)threadIdx.x) * 72) + 5)]));
-        conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[9] * kernel_shared[((((int)threadIdx.x) * 72) + 6)]));
-        conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[10] * kernel_shared[((((int)threadIdx.x) * 72) + 7)]));
-        conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[11] * kernel_shared[((((int)threadIdx.x) * 72) + 8)]));
-        conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[6] * kernel_shared[(((int)threadIdx.x) * 72)]));
-        conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[7] * kernel_shared[((((int)threadIdx.x) * 72) + 1)]));
-        conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[8] * kernel_shared[((((int)threadIdx.x) * 72) + 2)]));
-        conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[9] * kernel_shared[((((int)threadIdx.x) * 72) + 3)]));
-        conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[10] * kernel_shared[((((int)threadIdx.x) * 72) + 4)]));
-        conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[11] * kernel_shared[((((int)threadIdx.x) * 72) + 5)]));
-        conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[12] * kernel_shared[((((int)threadIdx.x) * 72) + 6)]));
-        conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[13] * kernel_shared[((((int)threadIdx.x) * 72) + 7)]));
-        conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[14] * kernel_shared[((((int)threadIdx.x) * 72) + 8)]));
-        conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[9] * kernel_shared[(((int)threadIdx.x) * 72)]));
-        conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[10] * kernel_shared[((((int)threadIdx.x) * 72) + 1)]));
-        conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[11] * kernel_shared[((((int)threadIdx.x) * 72) + 2)]));
-        conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[12] * kernel_shared[((((int)threadIdx.x) * 72) + 3)]));
-        conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[13] * kernel_shared[((((int)threadIdx.x) * 72) + 4)]));
-        conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[14] * kernel_shared[((((int)threadIdx.x) * 72) + 5)]));
-        conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[15] * kernel_shared[((((int)threadIdx.x) * 72) + 6)]));
-        conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[16] * kernel_shared[((((int)threadIdx.x) * 72) + 7)]));
-        conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[17] * kernel_shared[((((int)threadIdx.x) * 72) + 8)]));
-        conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[12] * kernel_shared[(((int)threadIdx.x) * 72)]));
-        conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[13] * kernel_shared[((((int)threadIdx.x) * 72) + 1)]));
-        conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[14] * kernel_shared[((((int)threadIdx.x) * 72) + 2)]));
-        conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[15] * kernel_shared[((((int)threadIdx.x) * 72) + 3)]));
-        conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[16] * kernel_shared[((((int)threadIdx.x) * 72) + 4)]));
-        conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[17] * kernel_shared[((((int)threadIdx.x) * 72) + 5)]));
-        conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[18] * kernel_shared[((((int)threadIdx.x) * 72) + 6)]));
-        conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[19] * kernel_shared[((((int)threadIdx.x) * 72) + 7)]));
-        conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[20] * kernel_shared[((((int)threadIdx.x) * 72) + 8)]));
-        conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[15] * kernel_shared[(((int)threadIdx.x) * 72)]));
-        conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[16] * kernel_shared[((((int)threadIdx.x) * 72) + 1)]));
-        conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[17] * kernel_shared[((((int)threadIdx.x) * 72) + 2)]));
-        conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[18] * kernel_shared[((((int)threadIdx.x) * 72) + 3)]));
-        conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[19] * kernel_shared[((((int)threadIdx.x) * 72) + 4)]));
-        conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[20] * kernel_shared[((((int)threadIdx.x) * 72) + 5)]));
-        conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[21] * kernel_shared[((((int)threadIdx.x) * 72) + 6)]));
-        conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[22] * kernel_shared[((((int)threadIdx.x) * 72) + 7)]));
-        conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[23] * kernel_shared[((((int)threadIdx.x) * 72) + 8)]));
-        conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[18] * kernel_shared[(((int)threadIdx.x) * 72)]));
-        conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[19] * kernel_shared[((((int)threadIdx.x) * 72) + 1)]));
-        conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[20] * kernel_shared[((((int)threadIdx.x) * 72) + 2)]));
-        conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[21] * kernel_shared[((((int)threadIdx.x) * 72) + 3)]));
-        conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[22] * kernel_shared[((((int)threadIdx.x) * 72) + 4)]));
-        conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[23] * kernel_shared[((((int)threadIdx.x) * 72) + 5)]));
-        conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[24] * kernel_shared[((((int)threadIdx.x) * 72) + 6)]));
-        conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[25] * kernel_shared[((((int)threadIdx.x) * 72) + 7)]));
-        conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[26] * kernel_shared[((((int)threadIdx.x) * 72) + 8)]));
-        conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[0] * kernel_shared[((((int)threadIdx.x) * 72) + 36)]));
-        conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[1] * kernel_shared[((((int)threadIdx.x) * 72) + 37)]));
-        conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[2] * kernel_shared[((((int)threadIdx.x) * 72) + 38)]));
-        conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[3] * kernel_shared[((((int)threadIdx.x) * 72) + 39)]));
-        conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[4] * kernel_shared[((((int)threadIdx.x) * 72) + 40)]));
-        conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[5] * kernel_shared[((((int)threadIdx.x) * 72) + 41)]));
-        conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[6] * kernel_shared[((((int)threadIdx.x) * 72) + 42)]));
-        conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[7] * kernel_shared[((((int)threadIdx.x) * 72) + 43)]));
-        conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[8] * kernel_shared[((((int)threadIdx.x) * 72) + 44)]));
-        conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[3] * kernel_shared[((((int)threadIdx.x) * 72) + 36)]));
-        conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[4] * kernel_shared[((((int)threadIdx.x) * 72) + 37)]));
-        conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[5] * kernel_shared[((((int)threadIdx.x) * 72) + 38)]));
-        conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[6] * kernel_shared[((((int)threadIdx.x) * 72) + 39)]));
-        conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[7] * kernel_shared[((((int)threadIdx.x) * 72) + 40)]));
-        conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[8] * kernel_shared[((((int)threadIdx.x) * 72) + 41)]));
-        conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[9] * kernel_shared[((((int)threadIdx.x) * 72) + 42)]));
-        conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[10] * kernel_shared[((((int)threadIdx.x) * 72) + 43)]));
-        conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[11] * kernel_shared[((((int)threadIdx.x) * 72) + 44)]));
-        conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[6] * kernel_shared[((((int)threadIdx.x) * 72) + 36)]));
-        conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[7] * kernel_shared[((((int)threadIdx.x) * 72) + 37)]));
-        conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[8] * kernel_shared[((((int)threadIdx.x) * 72) + 38)]));
-        conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[9] * kernel_shared[((((int)threadIdx.x) * 72) + 39)]));
-        conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[10] * kernel_shared[((((int)threadIdx.x) * 72) + 40)]));
-        conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[11] * kernel_shared[((((int)threadIdx.x) * 72) + 41)]));
-        conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[12] * kernel_shared[((((int)threadIdx.x) * 72) + 42)]));
-        conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[13] * kernel_shared[((((int)threadIdx.x) * 72) + 43)]));
-        conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[14] * kernel_shared[((((int)threadIdx.x) * 72) + 44)]));
-        conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[9] * kernel_shared[((((int)threadIdx.x) * 72) + 36)]));
-        conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[10] * kernel_shared[((((int)threadIdx.x) * 72) + 37)]));
-        conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[11] * kernel_shared[((((int)threadIdx.x) * 72) + 38)]));
-        conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[12] * kernel_shared[((((int)threadIdx.x) * 72) + 39)]));
-        conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[13] * kernel_shared[((((int)threadIdx.x) * 72) + 40)]));
-        conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[14] * kernel_shared[((((int)threadIdx.x) * 72) + 41)]));
-        conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[15] * kernel_shared[((((int)threadIdx.x) * 72) + 42)]));
-        conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[16] * kernel_shared[((((int)threadIdx.x) * 72) + 43)]));
-        conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[17] * kernel_shared[((((int)threadIdx.x) * 72) + 44)]));
-        conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[12] * kernel_shared[((((int)threadIdx.x) * 72) + 36)]));
-        conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[13] * kernel_shared[((((int)threadIdx.x) * 72) + 37)]));
-        conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[14] * kernel_shared[((((int)threadIdx.x) * 72) + 38)]));
-        conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[15] * kernel_shared[((((int)threadIdx.x) * 72) + 39)]));
-        conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[16] * kernel_shared[((((int)threadIdx.x) * 72) + 40)]));
-        conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[17] * kernel_shared[((((int)threadIdx.x) * 72) + 41)]));
-        conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[18] * kernel_shared[((((int)threadIdx.x) * 72) + 42)]));
-        conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[19] * kernel_shared[((((int)threadIdx.x) * 72) + 43)]));
-        conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[20] * kernel_shared[((((int)threadIdx.x) * 72) + 44)]));
-        conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[15] * kernel_shared[((((int)threadIdx.x) * 72) + 36)]));
-        conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[16] * kernel_shared[((((int)threadIdx.x) * 72) + 37)]));
-        conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[17] * kernel_shared[((((int)threadIdx.x) * 72) + 38)]));
-        conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[18] * kernel_shared[((((int)threadIdx.x) * 72) + 39)]));
-        conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[19] * kernel_shared[((((int)threadIdx.x) * 72) + 40)]));
-        conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[20] * kernel_shared[((((int)threadIdx.x) * 72) + 41)]));
-        conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[21] * kernel_shared[((((int)threadIdx.x) * 72) + 42)]));
-        conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[22] * kernel_shared[((((int)threadIdx.x) * 72) + 43)]));
-        conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[23] * kernel_shared[((((int)threadIdx.x) * 72) + 44)]));
-        conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[18] * kernel_shared[((((int)threadIdx.x) * 72) + 36)]));
-        conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[19] * kernel_shared[((((int)threadIdx.x) * 72) + 37)]));
-        conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[20] * kernel_shared[((((int)threadIdx.x) * 72) + 38)]));
-        conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[21] * kernel_shared[((((int)threadIdx.x) * 72) + 39)]));
-        conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[22] * kernel_shared[((((int)threadIdx.x) * 72) + 40)]));
-        conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[23] * kernel_shared[((((int)threadIdx.x) * 72) + 41)]));
-        conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[24] * kernel_shared[((((int)threadIdx.x) * 72) + 42)]));
-        conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[25] * kernel_shared[((((int)threadIdx.x) * 72) + 43)]));
-        conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[26] * kernel_shared[((((int)threadIdx.x) * 72) + 44)]));
-        conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[27] * kernel_shared[((((int)threadIdx.x) * 72) + 9)]));
-        conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[28] * kernel_shared[((((int)threadIdx.x) * 72) + 10)]));
-        conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[29] * kernel_shared[((((int)threadIdx.x) * 72) + 11)]));
-        conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[30] * kernel_shared[((((int)threadIdx.x) * 72) + 12)]));
-        conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[31] * kernel_shared[((((int)threadIdx.x) * 72) + 13)]));
-        conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[32] * kernel_shared[((((int)threadIdx.x) * 72) + 14)]));
-        conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[33] * kernel_shared[((((int)threadIdx.x) * 72) + 15)]));
-        conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[34] * kernel_shared[((((int)threadIdx.x) * 72) + 16)]));
-        conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[35] * kernel_shared[((((int)threadIdx.x) * 72) + 17)]));
-        conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[30] * kernel_shared[((((int)threadIdx.x) * 72) + 9)]));
-        conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[31] * kernel_shared[((((int)threadIdx.x) * 72) + 10)]));
-        conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[32] * kernel_shared[((((int)threadIdx.x) * 72) + 11)]));
-        conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[33] * kernel_shared[((((int)threadIdx.x) * 72) + 12)]));
-        conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[34] * kernel_shared[((((int)threadIdx.x) * 72) + 13)]));
-        conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[35] * kernel_shared[((((int)threadIdx.x) * 72) + 14)]));
-        conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[36] * kernel_shared[((((int)threadIdx.x) * 72) + 15)]));
-        conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[37] * kernel_shared[((((int)threadIdx.x) * 72) + 16)]));
-        conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[38] * kernel_shared[((((int)threadIdx.x) * 72) + 17)]));
-        conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[33] * kernel_shared[((((int)threadIdx.x) * 72) + 9)]));
-        conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[34] * kernel_shared[((((int)threadIdx.x) * 72) + 10)]));
-        conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[35] * kernel_shared[((((int)threadIdx.x) * 72) + 11)]));
-        conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[36] * kernel_shared[((((int)threadIdx.x) * 72) + 12)]));
-        conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[37] * kernel_shared[((((int)threadIdx.x) * 72) + 13)]));
-        conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[38] * kernel_shared[((((int)threadIdx.x) * 72) + 14)]));
-        conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[39] * kernel_shared[((((int)threadIdx.x) * 72) + 15)]));
-        conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[40] * kernel_shared[((((int)threadIdx.x) * 72) + 16)]));
-        conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[41] * kernel_shared[((((int)threadIdx.x) * 72) + 17)]));
-        conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[36] * kernel_shared[((((int)threadIdx.x) * 72) + 9)]));
-        conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[37] * kernel_shared[((((int)threadIdx.x) * 72) + 10)]));
-        conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[38] * kernel_shared[((((int)threadIdx.x) * 72) + 11)]));
-        conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[39] * kernel_shared[((((int)threadIdx.x) * 72) + 12)]));
-        conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[40] * kernel_shared[((((int)threadIdx.x) * 72) + 13)]));
-        conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[41] * kernel_shared[((((int)threadIdx.x) * 72) + 14)]));
-        conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[42] * kernel_shared[((((int)threadIdx.x) * 72) + 15)]));
-        conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[43] * kernel_shared[((((int)threadIdx.x) * 72) + 16)]));
-        conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[44] * kernel_shared[((((int)threadIdx.x) * 72) + 17)]));
-        conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[39] * kernel_shared[((((int)threadIdx.x) * 72) + 9)]));
-        conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[40] * kernel_shared[((((int)threadIdx.x) * 72) + 10)]));
-        conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[41] * kernel_shared[((((int)threadIdx.x) * 72) + 11)]));
-        conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[42] * kernel_shared[((((int)threadIdx.x) * 72) + 12)]));
-        conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[43] * kernel_shared[((((int)threadIdx.x) * 72) + 13)]));
-        conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[44] * kernel_shared[((((int)threadIdx.x) * 72) + 14)]));
-        conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[45] * kernel_shared[((((int)threadIdx.x) * 72) + 15)]));
-        conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[46] * kernel_shared[((((int)threadIdx.x) * 72) + 16)]));
-        conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[47] * kernel_shared[((((int)threadIdx.x) * 72) + 17)]));
-        conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[42] * kernel_shared[((((int)threadIdx.x) * 72) + 9)]));
-        conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[43] * kernel_shared[((((int)threadIdx.x) * 72) + 10)]));
-        conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[44] * kernel_shared[((((int)threadIdx.x) * 72) + 11)]));
-        conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[45] * kernel_shared[((((int)threadIdx.x) * 72) + 12)]));
-        conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[46] * kernel_shared[((((int)threadIdx.x) * 72) + 13)]));
-        conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[47] * kernel_shared[((((int)threadIdx.x) * 72) + 14)]));
-        conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[48] * kernel_shared[((((int)threadIdx.x) * 72) + 15)]));
-        conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[49] * kernel_shared[((((int)threadIdx.x) * 72) + 16)]));
-        conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[50] * kernel_shared[((((int)threadIdx.x) * 72) + 17)]));
-        conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[45] * kernel_shared[((((int)threadIdx.x) * 72) + 9)]));
-        conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[46] * kernel_shared[((((int)threadIdx.x) * 72) + 10)]));
-        conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[47] * kernel_shared[((((int)threadIdx.x) * 72) + 11)]));
-        conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[48] * kernel_shared[((((int)threadIdx.x) * 72) + 12)]));
-        conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[49] * kernel_shared[((((int)threadIdx.x) * 72) + 13)]));
-        conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[50] * kernel_shared[((((int)threadIdx.x) * 72) + 14)]));
-        conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[51] * kernel_shared[((((int)threadIdx.x) * 72) + 15)]));
-        conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[52] * kernel_shared[((((int)threadIdx.x) * 72) + 16)]));
-        conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[53] * kernel_shared[((((int)threadIdx.x) * 72) + 17)]));
-        conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[27] * kernel_shared[((((int)threadIdx.x) * 72) + 45)]));
-        conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[28] * kernel_shared[((((int)threadIdx.x) * 72) + 46)]));
-        conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[29] * kernel_shared[((((int)threadIdx.x) * 72) + 47)]));
-        conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[30] * kernel_shared[((((int)threadIdx.x) * 72) + 48)]));
-        conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[31] * kernel_shared[((((int)threadIdx.x) * 72) + 49)]));
-        conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[32] * kernel_shared[((((int)threadIdx.x) * 72) + 50)]));
-        conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[33] * kernel_shared[((((int)threadIdx.x) * 72) + 51)]));
-        conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[34] * kernel_shared[((((int)threadIdx.x) * 72) + 52)]));
-        conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[35] * kernel_shared[((((int)threadIdx.x) * 72) + 53)]));
-        conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[30] * kernel_shared[((((int)threadIdx.x) * 72) + 45)]));
-        conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[31] * kernel_shared[((((int)threadIdx.x) * 72) + 46)]));
-        conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[32] * kernel_shared[((((int)threadIdx.x) * 72) + 47)]));
-        conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[33] * kernel_shared[((((int)threadIdx.x) * 72) + 48)]));
-        conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[34] * kernel_shared[((((int)threadIdx.x) * 72) + 49)]));
-        conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[35] * kernel_shared[((((int)threadIdx.x) * 72) + 50)]));
-        conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[36] * kernel_shared[((((int)threadIdx.x) * 72) + 51)]));
-        conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[37] * kernel_shared[((((int)threadIdx.x) * 72) + 52)]));
-        conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[38] * kernel_shared[((((int)threadIdx.x) * 72) + 53)]));
-        conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[33] * kernel_shared[((((int)threadIdx.x) * 72) + 45)]));
-        conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[34] * kernel_shared[((((int)threadIdx.x) * 72) + 46)]));
-        conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[35] * kernel_shared[((((int)threadIdx.x) * 72) + 47)]));
-        conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[36] * kernel_shared[((((int)threadIdx.x) * 72) + 48)]));
-        conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[37] * kernel_shared[((((int)threadIdx.x) * 72) + 49)]));
-        conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[38] * kernel_shared[((((int)threadIdx.x) * 72) + 50)]));
-        conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[39] * kernel_shared[((((int)threadIdx.x) * 72) + 51)]));
-        conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[40] * kernel_shared[((((int)threadIdx.x) * 72) + 52)]));
-        conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[41] * kernel_shared[((((int)threadIdx.x) * 72) + 53)]));
-        conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[36] * kernel_shared[((((int)threadIdx.x) * 72) + 45)]));
-        conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[37] * kernel_shared[((((int)threadIdx.x) * 72) + 46)]));
-        conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[38] * kernel_shared[((((int)threadIdx.x) * 72) + 47)]));
-        conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[39] * kernel_shared[((((int)threadIdx.x) * 72) + 48)]));
-        conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[40] * kernel_shared[((((int)threadIdx.x) * 72) + 49)]));
-        conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[41] * kernel_shared[((((int)threadIdx.x) * 72) + 50)]));
-        conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[42] * kernel_shared[((((int)threadIdx.x) * 72) + 51)]));
-        conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[43] * kernel_shared[((((int)threadIdx.x) * 72) + 52)]));
-        conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[44] * kernel_shared[((((int)threadIdx.x) * 72) + 53)]));
-        conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[39] * kernel_shared[((((int)threadIdx.x) * 72) + 45)]));
-        conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[40] * kernel_shared[((((int)threadIdx.x) * 72) + 46)]));
-        conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[41] * kernel_shared[((((int)threadIdx.x) * 72) + 47)]));
-        conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[42] * kernel_shared[((((int)threadIdx.x) * 72) + 48)]));
-        conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[43] * kernel_shared[((((int)threadIdx.x) * 72) + 49)]));
-        conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[44] * kernel_shared[((((int)threadIdx.x) * 72) + 50)]));
-        conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[45] * kernel_shared[((((int)threadIdx.x) * 72) + 51)]));
-        conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[46] * kernel_shared[((((int)threadIdx.x) * 72) + 52)]));
-        conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[47] * kernel_shared[((((int)threadIdx.x) * 72) + 53)]));
-        conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[42] * kernel_shared[((((int)threadIdx.x) * 72) + 45)]));
-        conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[43] * kernel_shared[((((int)threadIdx.x) * 72) + 46)]));
-        conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[44] * kernel_shared[((((int)threadIdx.x) * 72) + 47)]));
-        conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[45] * kernel_shared[((((int)threadIdx.x) * 72) + 48)]));
-        conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[46] * kernel_shared[((((int)threadIdx.x) * 72) + 49)]));
-        conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[47] * kernel_shared[((((int)threadIdx.x) * 72) + 50)]));
-        conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[48] * kernel_shared[((((int)threadIdx.x) * 72) + 51)]));
-        conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[49] * kernel_shared[((((int)threadIdx.x) * 72) + 52)]));
-        conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[50] * kernel_shared[((((int)threadIdx.x) * 72) + 53)]));
-        conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[45] * kernel_shared[((((int)threadIdx.x) * 72) + 45)]));
-        conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[46] * kernel_shared[((((int)threadIdx.x) * 72) + 46)]));
-        conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[47] * kernel_shared[((((int)threadIdx.x) * 72) + 47)]));
-        conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[48] * kernel_shared[((((int)threadIdx.x) * 72) + 48)]));
-        conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[49] * kernel_shared[((((int)threadIdx.x) * 72) + 49)]));
-        conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[50] * kernel_shared[((((int)threadIdx.x) * 72) + 50)]));
-        conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[51] * kernel_shared[((((int)threadIdx.x) * 72) + 51)]));
-        conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[52] * kernel_shared[((((int)threadIdx.x) * 72) + 52)]));
-        conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[53] * kernel_shared[((((int)threadIdx.x) * 72) + 53)]));
-        conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[54] * kernel_shared[((((int)threadIdx.x) * 72) + 18)]));
-        conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[55] * kernel_shared[((((int)threadIdx.x) * 72) + 19)]));
-        conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[56] * kernel_shared[((((int)threadIdx.x) * 72) + 20)]));
-        conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[57] * kernel_shared[((((int)threadIdx.x) * 72) + 21)]));
-        conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[58] * kernel_shared[((((int)threadIdx.x) * 72) + 22)]));
-        conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[59] * kernel_shared[((((int)threadIdx.x) * 72) + 23)]));
-        conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[60] * kernel_shared[((((int)threadIdx.x) * 72) + 24)]));
-        conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[61] * kernel_shared[((((int)threadIdx.x) * 72) + 25)]));
-        conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[62] * kernel_shared[((((int)threadIdx.x) * 72) + 26)]));
-        conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[57] * kernel_shared[((((int)threadIdx.x) * 72) + 18)]));
-        conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[58] * kernel_shared[((((int)threadIdx.x) * 72) + 19)]));
-        conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[59] * kernel_shared[((((int)threadIdx.x) * 72) + 20)]));
-        conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[60] * kernel_shared[((((int)threadIdx.x) * 72) + 21)]));
-        conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[61] * kernel_shared[((((int)threadIdx.x) * 72) + 22)]));
-        conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[62] * kernel_shared[((((int)threadIdx.x) * 72) + 23)]));
-        conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[63] * kernel_shared[((((int)threadIdx.x) * 72) + 24)]));
-        conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[64] * kernel_shared[((((int)threadIdx.x) * 72) + 25)]));
-        conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[65] * kernel_shared[((((int)threadIdx.x) * 72) + 26)]));
-        conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[60] * kernel_shared[((((int)threadIdx.x) * 72) + 18)]));
-        conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[61] * kernel_shared[((((int)threadIdx.x) * 72) + 19)]));
-        conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[62] * kernel_shared[((((int)threadIdx.x) * 72) + 20)]));
-        conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[63] * kernel_shared[((((int)threadIdx.x) * 72) + 21)]));
-        conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[64] * kernel_shared[((((int)threadIdx.x) * 72) + 22)]));
-        conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[65] * kernel_shared[((((int)threadIdx.x) * 72) + 23)]));
-        conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[66] * kernel_shared[((((int)threadIdx.x) * 72) + 24)]));
-        conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[67] * kernel_shared[((((int)threadIdx.x) * 72) + 25)]));
-        conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[68] * kernel_shared[((((int)threadIdx.x) * 72) + 26)]));
-        conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[63] * kernel_shared[((((int)threadIdx.x) * 72) + 18)]));
-        conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[64] * kernel_shared[((((int)threadIdx.x) * 72) + 19)]));
-        conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[65] * kernel_shared[((((int)threadIdx.x) * 72) + 20)]));
-        conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[66] * kernel_shared[((((int)threadIdx.x) * 72) + 21)]));
-        conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[67] * kernel_shared[((((int)threadIdx.x) * 72) + 22)]));
-        conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[68] * kernel_shared[((((int)threadIdx.x) * 72) + 23)]));
-        conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[69] * kernel_shared[((((int)threadIdx.x) * 72) + 24)]));
-        conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[70] * kernel_shared[((((int)threadIdx.x) * 72) + 25)]));
-        conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[71] * kernel_shared[((((int)threadIdx.x) * 72) + 26)]));
-        conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[66] * kernel_shared[((((int)threadIdx.x) * 72) + 18)]));
-        conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[67] * kernel_shared[((((int)threadIdx.x) * 72) + 19)]));
-        conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[68] * kernel_shared[((((int)threadIdx.x) * 72) + 20)]));
-        conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[69] * kernel_shared[((((int)threadIdx.x) * 72) + 21)]));
-        conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[70] * kernel_shared[((((int)threadIdx.x) * 72) + 22)]));
-        conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[71] * kernel_shared[((((int)threadIdx.x) * 72) + 23)]));
-        conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[72] * kernel_shared[((((int)threadIdx.x) * 72) + 24)]));
-        conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[73] * kernel_shared[((((int)threadIdx.x) * 72) + 25)]));
-        conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[74] * kernel_shared[((((int)threadIdx.x) * 72) + 26)]));
-        conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[69] * kernel_shared[((((int)threadIdx.x) * 72) + 18)]));
-        conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[70] * kernel_shared[((((int)threadIdx.x) * 72) + 19)]));
-        conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[71] * kernel_shared[((((int)threadIdx.x) * 72) + 20)]));
-        conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[72] * kernel_shared[((((int)threadIdx.x) * 72) + 21)]));
-        conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[73] * kernel_shared[((((int)threadIdx.x) * 72) + 22)]));
-        conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[74] * kernel_shared[((((int)threadIdx.x) * 72) + 23)]));
-        conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[75] * kernel_shared[((((int)threadIdx.x) * 72) + 24)]));
-        conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[76] * kernel_shared[((((int)threadIdx.x) * 72) + 25)]));
-        conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[77] * kernel_shared[((((int)threadIdx.x) * 72) + 26)]));
-        conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[72] * kernel_shared[((((int)threadIdx.x) * 72) + 18)]));
-        conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[73] * kernel_shared[((((int)threadIdx.x) * 72) + 19)]));
-        conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[74] * kernel_shared[((((int)threadIdx.x) * 72) + 20)]));
-        conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[75] * kernel_shared[((((int)threadIdx.x) * 72) + 21)]));
-        conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[76] * kernel_shared[((((int)threadIdx.x) * 72) + 22)]));
-        conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[77] * kernel_shared[((((int)threadIdx.x) * 72) + 23)]));
-        conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[78] * kernel_shared[((((int)threadIdx.x) * 72) + 24)]));
-        conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[79] * kernel_shared[((((int)threadIdx.x) * 72) + 25)]));
-        conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[80] * kernel_shared[((((int)threadIdx.x) * 72) + 26)]));
-        conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[54] * kernel_shared[((((int)threadIdx.x) * 72) + 54)]));
-        conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[55] * kernel_shared[((((int)threadIdx.x) * 72) + 55)]));
-        conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[56] * kernel_shared[((((int)threadIdx.x) * 72) + 56)]));
-        conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[57] * kernel_shared[((((int)threadIdx.x) * 72) + 57)]));
-        conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[58] * kernel_shared[((((int)threadIdx.x) * 72) + 58)]));
-        conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[59] * kernel_shared[((((int)threadIdx.x) * 72) + 59)]));
-        conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[60] * kernel_shared[((((int)threadIdx.x) * 72) + 60)]));
-        conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[61] * kernel_shared[((((int)threadIdx.x) * 72) + 61)]));
-        conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[62] * kernel_shared[((((int)threadIdx.x) * 72) + 62)]));
-        conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[57] * kernel_shared[((((int)threadIdx.x) * 72) + 54)]));
-        conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[58] * kernel_shared[((((int)threadIdx.x) * 72) + 55)]));
-        conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[59] * kernel_shared[((((int)threadIdx.x) * 72) + 56)]));
-        conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[60] * kernel_shared[((((int)threadIdx.x) * 72) + 57)]));
-        conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[61] * kernel_shared[((((int)threadIdx.x) * 72) + 58)]));
-        conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[62] * kernel_shared[((((int)threadIdx.x) * 72) + 59)]));
-        conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[63] * kernel_shared[((((int)threadIdx.x) * 72) + 60)]));
-        conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[64] * kernel_shared[((((int)threadIdx.x) * 72) + 61)]));
-        conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[65] * kernel_shared[((((int)threadIdx.x) * 72) + 62)]));
-        conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[60] * kernel_shared[((((int)threadIdx.x) * 72) + 54)]));
-        conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[61] * kernel_shared[((((int)threadIdx.x) * 72) + 55)]));
-        conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[62] * kernel_shared[((((int)threadIdx.x) * 72) + 56)]));
-        conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[63] * kernel_shared[((((int)threadIdx.x) * 72) + 57)]));
-        conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[64] * kernel_shared[((((int)threadIdx.x) * 72) + 58)]));
-        conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[65] * kernel_shared[((((int)threadIdx.x) * 72) + 59)]));
-        conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[66] * kernel_shared[((((int)threadIdx.x) * 72) + 60)]));
-        conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[67] * kernel_shared[((((int)threadIdx.x) * 72) + 61)]));
-        conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[68] * kernel_shared[((((int)threadIdx.x) * 72) + 62)]));
-        conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[63] * kernel_shared[((((int)threadIdx.x) * 72) + 54)]));
-        conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[64] * kernel_shared[((((int)threadIdx.x) * 72) + 55)]));
-        conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[65] * kernel_shared[((((int)threadIdx.x) * 72) + 56)]));
-        conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[66] * kernel_shared[((((int)threadIdx.x) * 72) + 57)]));
-        conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[67] * kernel_shared[((((int)threadIdx.x) * 72) + 58)]));
-        conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[68] * kernel_shared[((((int)threadIdx.x) * 72) + 59)]));
-        conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[69] * kernel_shared[((((int)threadIdx.x) * 72) + 60)]));
-        conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[70] * kernel_shared[((((int)threadIdx.x) * 72) + 61)]));
-        conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[71] * kernel_shared[((((int)threadIdx.x) * 72) + 62)]));
-        conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[66] * kernel_shared[((((int)threadIdx.x) * 72) + 54)]));
-        conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[67] * kernel_shared[((((int)threadIdx.x) * 72) + 55)]));
-        conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[68] * kernel_shared[((((int)threadIdx.x) * 72) + 56)]));
-        conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[69] * kernel_shared[((((int)threadIdx.x) * 72) + 57)]));
-        conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[70] * kernel_shared[((((int)threadIdx.x) * 72) + 58)]));
-        conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[71] * kernel_shared[((((int)threadIdx.x) * 72) + 59)]));
-        conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[72] * kernel_shared[((((int)threadIdx.x) * 72) + 60)]));
-        conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[73] * kernel_shared[((((int)threadIdx.x) * 72) + 61)]));
-        conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[74] * kernel_shared[((((int)threadIdx.x) * 72) + 62)]));
-        conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[69] * kernel_shared[((((int)threadIdx.x) * 72) + 54)]));
-        conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[70] * kernel_shared[((((int)threadIdx.x) * 72) + 55)]));
-        conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[71] * kernel_shared[((((int)threadIdx.x) * 72) + 56)]));
-        conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[72] * kernel_shared[((((int)threadIdx.x) * 72) + 57)]));
-        conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[73] * kernel_shared[((((int)threadIdx.x) * 72) + 58)]));
-        conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[74] * kernel_shared[((((int)threadIdx.x) * 72) + 59)]));
-        conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[75] * kernel_shared[((((int)threadIdx.x) * 72) + 60)]));
-        conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[76] * kernel_shared[((((int)threadIdx.x) * 72) + 61)]));
-        conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[77] * kernel_shared[((((int)threadIdx.x) * 72) + 62)]));
-        conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[72] * kernel_shared[((((int)threadIdx.x) * 72) + 54)]));
-        conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[73] * kernel_shared[((((int)threadIdx.x) * 72) + 55)]));
-        conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[74] * kernel_shared[((((int)threadIdx.x) * 72) + 56)]));
-        conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[75] * kernel_shared[((((int)threadIdx.x) * 72) + 57)]));
-        conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[76] * kernel_shared[((((int)threadIdx.x) * 72) + 58)]));
-        conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[77] * kernel_shared[((((int)threadIdx.x) * 72) + 59)]));
-        conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[78] * kernel_shared[((((int)threadIdx.x) * 72) + 60)]));
-        conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[79] * kernel_shared[((((int)threadIdx.x) * 72) + 61)]));
-        conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[80] * kernel_shared[((((int)threadIdx.x) * 72) + 62)]));
-        conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[81] * kernel_shared[((((int)threadIdx.x) * 72) + 27)]));
-        conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[82] * kernel_shared[((((int)threadIdx.x) * 72) + 28)]));
-        conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[83] * kernel_shared[((((int)threadIdx.x) * 72) + 29)]));
-        conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[84] * kernel_shared[((((int)threadIdx.x) * 72) + 30)]));
-        conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[85] * kernel_shared[((((int)threadIdx.x) * 72) + 31)]));
-        conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[86] * kernel_shared[((((int)threadIdx.x) * 72) + 32)]));
-        conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[87] * kernel_shared[((((int)threadIdx.x) * 72) + 33)]));
-        conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[88] * kernel_shared[((((int)threadIdx.x) * 72) + 34)]));
-        conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[89] * kernel_shared[((((int)threadIdx.x) * 72) + 35)]));
-        conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[84] * kernel_shared[((((int)threadIdx.x) * 72) + 27)]));
-        conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[85] * kernel_shared[((((int)threadIdx.x) * 72) + 28)]));
-        conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[86] * kernel_shared[((((int)threadIdx.x) * 72) + 29)]));
-        conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[87] * kernel_shared[((((int)threadIdx.x) * 72) + 30)]));
-        conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[88] * kernel_shared[((((int)threadIdx.x) * 72) + 31)]));
-        conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[89] * kernel_shared[((((int)threadIdx.x) * 72) + 32)]));
-        conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[90] * kernel_shared[((((int)threadIdx.x) * 72) + 33)]));
-        conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[91] * kernel_shared[((((int)threadIdx.x) * 72) + 34)]));
-        conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[92] * kernel_shared[((((int)threadIdx.x) * 72) + 35)]));
-        conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[87] * kernel_shared[((((int)threadIdx.x) * 72) + 27)]));
-        conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[88] * kernel_shared[((((int)threadIdx.x) * 72) + 28)]));
-        conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[89] * kernel_shared[((((int)threadIdx.x) * 72) + 29)]));
-        conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[90] * kernel_shared[((((int)threadIdx.x) * 72) + 30)]));
-        conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[91] * kernel_shared[((((int)threadIdx.x) * 72) + 31)]));
-        conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[92] * kernel_shared[((((int)threadIdx.x) * 72) + 32)]));
-        conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[93] * kernel_shared[((((int)threadIdx.x) * 72) + 33)]));
-        conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[94] * kernel_shared[((((int)threadIdx.x) * 72) + 34)]));
-        conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[95] * kernel_shared[((((int)threadIdx.x) * 72) + 35)]));
-        conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[90] * kernel_shared[((((int)threadIdx.x) * 72) + 27)]));
-        conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[91] * kernel_shared[((((int)threadIdx.x) * 72) + 28)]));
-        conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[92] * kernel_shared[((((int)threadIdx.x) * 72) + 29)]));
-        conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[93] * kernel_shared[((((int)threadIdx.x) * 72) + 30)]));
-        conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[94] * kernel_shared[((((int)threadIdx.x) * 72) + 31)]));
-        conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[95] * kernel_shared[((((int)threadIdx.x) * 72) + 32)]));
-        conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[96] * kernel_shared[((((int)threadIdx.x) * 72) + 33)]));
-        conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[97] * kernel_shared[((((int)threadIdx.x) * 72) + 34)]));
-        conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[98] * kernel_shared[((((int)threadIdx.x) * 72) + 35)]));
-        conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[93] * kernel_shared[((((int)threadIdx.x) * 72) + 27)]));
-        conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[94] * kernel_shared[((((int)threadIdx.x) * 72) + 28)]));
-        conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[95] * kernel_shared[((((int)threadIdx.x) * 72) + 29)]));
-        conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[96] * kernel_shared[((((int)threadIdx.x) * 72) + 30)]));
-        conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[97] * kernel_shared[((((int)threadIdx.x) * 72) + 31)]));
-        conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[98] * kernel_shared[((((int)threadIdx.x) * 72) + 32)]));
-        conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[99] * kernel_shared[((((int)threadIdx.x) * 72) + 33)]));
-        conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[100] * kernel_shared[((((int)threadIdx.x) * 72) + 34)]));
-        conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[101] * kernel_shared[((((int)threadIdx.x) * 72) + 35)]));
-        conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[96] * kernel_shared[((((int)threadIdx.x) * 72) + 27)]));
-        conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[97] * kernel_shared[((((int)threadIdx.x) * 72) + 28)]));
-        conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[98] * kernel_shared[((((int)threadIdx.x) * 72) + 29)]));
-        conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[99] * kernel_shared[((((int)threadIdx.x) * 72) + 30)]));
-        conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[100] * kernel_shared[((((int)threadIdx.x) * 72) + 31)]));
-        conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[101] * kernel_shared[((((int)threadIdx.x) * 72) + 32)]));
-        conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[102] * kernel_shared[((((int)threadIdx.x) * 72) + 33)]));
-        conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[103] * kernel_shared[((((int)threadIdx.x) * 72) + 34)]));
-        conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[104] * kernel_shared[((((int)threadIdx.x) * 72) + 35)]));
-        conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[99] * kernel_shared[((((int)threadIdx.x) * 72) + 27)]));
-        conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[100] * kernel_shared[((((int)threadIdx.x) * 72) + 28)]));
-        conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[101] * kernel_shared[((((int)threadIdx.x) * 72) + 29)]));
-        conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[102] * kernel_shared[((((int)threadIdx.x) * 72) + 30)]));
-        conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[103] * kernel_shared[((((int)threadIdx.x) * 72) + 31)]));
-        conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[104] * kernel_shared[((((int)threadIdx.x) * 72) + 32)]));
-        conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[105] * kernel_shared[((((int)threadIdx.x) * 72) + 33)]));
-        conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[106] * kernel_shared[((((int)threadIdx.x) * 72) + 34)]));
-        conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[107] * kernel_shared[((((int)threadIdx.x) * 72) + 35)]));
-        conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[81] * kernel_shared[((((int)threadIdx.x) * 72) + 63)]));
-        conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[82] * kernel_shared[((((int)threadIdx.x) * 72) + 64)]));
-        conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[83] * kernel_shared[((((int)threadIdx.x) * 72) + 65)]));
-        conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[84] * kernel_shared[((((int)threadIdx.x) * 72) + 66)]));
-        conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[85] * kernel_shared[((((int)threadIdx.x) * 72) + 67)]));
-        conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[86] * kernel_shared[((((int)threadIdx.x) * 72) + 68)]));
-        conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[87] * kernel_shared[((((int)threadIdx.x) * 72) + 69)]));
-        conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[88] * kernel_shared[((((int)threadIdx.x) * 72) + 70)]));
-        conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[89] * kernel_shared[((((int)threadIdx.x) * 72) + 71)]));
-        conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[84] * kernel_shared[((((int)threadIdx.x) * 72) + 63)]));
-        conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[85] * kernel_shared[((((int)threadIdx.x) * 72) + 64)]));
-        conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[86] * kernel_shared[((((int)threadIdx.x) * 72) + 65)]));
-        conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[87] * kernel_shared[((((int)threadIdx.x) * 72) + 66)]));
-        conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[88] * kernel_shared[((((int)threadIdx.x) * 72) + 67)]));
-        conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[89] * kernel_shared[((((int)threadIdx.x) * 72) + 68)]));
-        conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[90] * kernel_shared[((((int)threadIdx.x) * 72) + 69)]));
-        conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[91] * kernel_shared[((((int)threadIdx.x) * 72) + 70)]));
-        conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[92] * kernel_shared[((((int)threadIdx.x) * 72) + 71)]));
-        conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[87] * kernel_shared[((((int)threadIdx.x) * 72) + 63)]));
-        conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[88] * kernel_shared[((((int)threadIdx.x) * 72) + 64)]));
-        conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[89] * kernel_shared[((((int)threadIdx.x) * 72) + 65)]));
-        conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[90] * kernel_shared[((((int)threadIdx.x) * 72) + 66)]));
-        conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[91] * kernel_shared[((((int)threadIdx.x) * 72) + 67)]));
-        conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[92] * kernel_shared[((((int)threadIdx.x) * 72) + 68)]));
-        conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[93] * kernel_shared[((((int)threadIdx.x) * 72) + 69)]));
-        conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[94] * kernel_shared[((((int)threadIdx.x) * 72) + 70)]));
-        conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[95] * kernel_shared[((((int)threadIdx.x) * 72) + 71)]));
-        conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[90] * kernel_shared[((((int)threadIdx.x) * 72) + 63)]));
-        conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[91] * kernel_shared[((((int)threadIdx.x) * 72) + 64)]));
-        conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[92] * kernel_shared[((((int)threadIdx.x) * 72) + 65)]));
-        conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[93] * kernel_shared[((((int)threadIdx.x) * 72) + 66)]));
-        conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[94] * kernel_shared[((((int)threadIdx.x) * 72) + 67)]));
-        conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[95] * kernel_shared[((((int)threadIdx.x) * 72) + 68)]));
-        conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[96] * kernel_shared[((((int)threadIdx.x) * 72) + 69)]));
-        conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[97] * kernel_shared[((((int)threadIdx.x) * 72) + 70)]));
-        conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[98] * kernel_shared[((((int)threadIdx.x) * 72) + 71)]));
-        conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[93] * kernel_shared[((((int)threadIdx.x) * 72) + 63)]));
-        conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[94] * kernel_shared[((((int)threadIdx.x) * 72) + 64)]));
-        conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[95] * kernel_shared[((((int)threadIdx.x) * 72) + 65)]));
-        conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[96] * kernel_shared[((((int)threadIdx.x) * 72) + 66)]));
-        conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[97] * kernel_shared[((((int)threadIdx.x) * 72) + 67)]));
-        conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[98] * kernel_shared[((((int)threadIdx.x) * 72) + 68)]));
-        conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[99] * kernel_shared[((((int)threadIdx.x) * 72) + 69)]));
-        conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[100] * kernel_shared[((((int)threadIdx.x) * 72) + 70)]));
-        conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[101] * kernel_shared[((((int)threadIdx.x) * 72) + 71)]));
-        conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[96] * kernel_shared[((((int)threadIdx.x) * 72) + 63)]));
-        conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[97] * kernel_shared[((((int)threadIdx.x) * 72) + 64)]));
-        conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[98] * kernel_shared[((((int)threadIdx.x) * 72) + 65)]));
-        conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[99] * kernel_shared[((((int)threadIdx.x) * 72) + 66)]));
-        conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[100] * kernel_shared[((((int)threadIdx.x) * 72) + 67)]));
-        conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[101] * kernel_shared[((((int)threadIdx.x) * 72) + 68)]));
-        conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[102] * kernel_shared[((((int)threadIdx.x) * 72) + 69)]));
-        conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[103] * kernel_shared[((((int)threadIdx.x) * 72) + 70)]));
-        conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[104] * kernel_shared[((((int)threadIdx.x) * 72) + 71)]));
-        conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[99] * kernel_shared[((((int)threadIdx.x) * 72) + 63)]));
-        conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[100] * kernel_shared[((((int)threadIdx.x) * 72) + 64)]));
-        conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[101] * kernel_shared[((((int)threadIdx.x) * 72) + 65)]));
-        conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[102] * kernel_shared[((((int)threadIdx.x) * 72) + 66)]));
-        conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[103] * kernel_shared[((((int)threadIdx.x) * 72) + 67)]));
-        conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[104] * kernel_shared[((((int)threadIdx.x) * 72) + 68)]));
-        conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[105] * kernel_shared[((((int)threadIdx.x) * 72) + 69)]));
-        conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[106] * kernel_shared[((((int)threadIdx.x) * 72) + 70)]));
-        conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[107] * kernel_shared[((((int)threadIdx.x) * 72) + 71)]));
       }
       for (int i1_inner = 0; i1_inner < 2; ++i1_inner) {
-        for (int i2_inner = 0; i2_inner < 7; ++i2_inner) {
-          compute[((((((((int)blockIdx.x) / 7) * 1568) + (((int)threadIdx.x) * 98)) + (i1_inner * 49)) + (i2_inner * 7)) + (((int)blockIdx.x) % 7))] = max((conv2d_nchw[((i1_inner * 7) + i2_inner)] + bias[((((((int)blockIdx.x) / 7) * 32) + (((int)threadIdx.x) * 2)) + i1_inner)]), 0.000000e+00f);
-        }
+        compute[(((((int)blockIdx.x) * 196) + (i1_inner * 49)) + ((int)threadIdx.x))] = max((conv2d_nchw[i1_inner] + bias[((((int)blockIdx.x) * 4) + i1_inner)]), 0.000000e+00f);
+        compute[((((((int)blockIdx.x) * 196) + (i1_inner * 49)) + ((int)threadIdx.x)) + 7)] = max((conv2d_nchw[(i1_inner + 2)] + bias[((((int)blockIdx.x) * 4) + i1_inner)]), 0.000000e+00f);
+        compute[((((((int)blockIdx.x) * 196) + (i1_inner * 49)) + ((int)threadIdx.x)) + 14)] = max((conv2d_nchw[(i1_inner + 4)] + bias[((((int)blockIdx.x) * 4) + i1_inner)]), 0.000000e+00f);
+        compute[((((((int)blockIdx.x) * 196) + (i1_inner * 49)) + ((int)threadIdx.x)) + 21)] = max((conv2d_nchw[(i1_inner + 6)] + bias[((((int)blockIdx.x) * 4) + i1_inner)]), 0.000000e+00f);
+        compute[((((((int)blockIdx.x) * 196) + (i1_inner * 49)) + ((int)threadIdx.x)) + 28)] = max((conv2d_nchw[(i1_inner + 8)] + bias[((((int)blockIdx.x) * 4) + i1_inner)]), 0.000000e+00f);
+        compute[((((((int)blockIdx.x) * 196) + (i1_inner * 49)) + ((int)threadIdx.x)) + 35)] = max((conv2d_nchw[(i1_inner + 10)] + bias[((((int)blockIdx.x) * 4) + i1_inner)]), 0.000000e+00f);
+        compute[((((((int)blockIdx.x) * 196) + (i1_inner * 49)) + ((int)threadIdx.x)) + 42)] = max((conv2d_nchw[(i1_inner + 12)] + bias[((((int)blockIdx.x) * 4) + i1_inner)]), 0.000000e+00f);
+        compute[((((((int)blockIdx.x) * 196) + (i1_inner * 49)) + ((int)threadIdx.x)) + 98)] = max((conv2d_nchw[(i1_inner + 14)] + bias[(((((int)blockIdx.x) * 4) + i1_inner) + 2)]), 0.000000e+00f);
+        compute[((((((int)blockIdx.x) * 196) + (i1_inner * 49)) + ((int)threadIdx.x)) + 105)] = max((conv2d_nchw[(i1_inner + 16)] + bias[(((((int)blockIdx.x) * 4) + i1_inner) + 2)]), 0.000000e+00f);
+        compute[((((((int)blockIdx.x) * 196) + (i1_inner * 49)) + ((int)threadIdx.x)) + 112)] = max((conv2d_nchw[(i1_inner + 18)] + bias[(((((int)blockIdx.x) * 4) + i1_inner) + 2)]), 0.000000e+00f);
+        compute[((((((int)blockIdx.x) * 196) + (i1_inner * 49)) + ((int)threadIdx.x)) + 119)] = max((conv2d_nchw[(i1_inner + 20)] + bias[(((((int)blockIdx.x) * 4) + i1_inner) + 2)]), 0.000000e+00f);
+        compute[((((((int)blockIdx.x) * 196) + (i1_inner * 49)) + ((int)threadIdx.x)) + 126)] = max((conv2d_nchw[(i1_inner + 22)] + bias[(((((int)blockIdx.x) * 4) + i1_inner) + 2)]), 0.000000e+00f);
+        compute[((((((int)blockIdx.x) * 196) + (i1_inner * 49)) + ((int)threadIdx.x)) + 133)] = max((conv2d_nchw[(i1_inner + 24)] + bias[(((((int)blockIdx.x) * 4) + i1_inner) + 2)]), 0.000000e+00f);
+        compute[((((((int)blockIdx.x) * 196) + (i1_inner * 49)) + ((int)threadIdx.x)) + 140)] = max((conv2d_nchw[(i1_inner + 26)] + bias[(((((int)blockIdx.x) * 4) + i1_inner) + 2)]), 0.000000e+00f);
       }
     }
 
@@ -1781,7 +1390,7 @@ In the example below we resume the status and do more 5 trials.
 
 .. rst-class:: sphx-glr-timing
 
-   **Total running time of the script:** ( 3 minutes  11.157 seconds)
+   **Total running time of the script:** ( 3 minutes  10.386 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 699fb6b85..e812c71cf 100644
--- a/docs/_sources/how_to/tune_with_autoscheduler/tune_network_cuda.rst.txt
+++ b/docs/_sources/how_to/tune_with_autoscheduler/tune_network_cuda.rst.txt
@@ -647,7 +647,7 @@ so we can read the log file and load the best schedules.
     Evaluate inference time cost...
     Execution time summary:
      mean (ms)   median (ms)    max (ms)     min (ms)     std (ms)  
-       9.7592       9.7663       9.7725       9.7386       0.0147   
+       9.6540       9.6583       9.6798       9.6239       0.0230   
                
 
 
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 5d488e829..718158147 100644
--- a/docs/_sources/how_to/tune_with_autoscheduler/tune_network_x86.rst.txt
+++ b/docs/_sources/how_to/tune_with_autoscheduler/tune_network_x86.rst.txt
@@ -666,7 +666,7 @@ so we can read the log file and load the best schedules.
     Evaluate inference time cost...
     Execution time summary:
      mean (ms)   median (ms)    max (ms)     min (ms)     std (ms)  
-      762.3544     762.0270     763.7512     761.2850      1.0331   
+      762.9937     760.9339     767.2873     760.7600      3.0368   
                
 
 
@@ -694,7 +694,7 @@ Other Tips
 
 .. rst-class:: sphx-glr-timing
 
-   **Total running time of the script:** ( 1 minutes  21.967 seconds)
+   **Total running time of the script:** ( 1 minutes  22.515 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 2a130e915..f15d7aeef 100644
--- a/docs/_sources/how_to/tune_with_autoscheduler/tune_sparse_x86.rst.txt
+++ b/docs/_sources/how_to/tune_with_autoscheduler/tune_sparse_x86.rst.txt
@@ -397,28 +397,659 @@ layout transformation, parallelization, vectorization, unrolling, and operator f
                  placeholder_4: Buffer(placeholder_14: Pointer(float32), float32, [65536], []),
                  compute: Buffer(compute_2: Pointer(float32), float32, [65536], [])}
       buffer_map = {placeholder_5: placeholder, placeholder_6: placeholder_1, placeholder_7: placeholder_2, placeholder_8: placeholder_3, placeholder_9: placeholder_4, compute_1: compute}
-      preflattened_buffer_map = {compute_1: compute_3: Buffer(compute_2, float32, [128, 512], []), placeholder_8: placeholder_15: Buffer(placeholder_13, int32, [33], []), placeholder_5: placeholder_16: Buffer(placeholder_10, float32, [128, 256], []), placeholder_6: placeholder_17: Buffer(placeholder_11, float32, [4916, 16, 1], []), placeholder_7: placeholder_18: Buffer(placeholder_12, int32, [4916], []), placeholder_9: placeholder_19: Buffer(placeholder_14, float32, [128, 512], [])} {
-      for (i0.outer.i1.outer.fused: int32, 0, 16) "parallel" {
-        allocate(compute_4: Pointer(global float32), float32, [4096]), storage_scope = global {
-          for (i.outer.inner: int32, 0, 128) {
-            for (nb_j.inner: int32, 0, 2) {
-              for (j.init: int32, 0, 16) {
-                compute_5: Buffer(compute_4, float32, [4096], [])[(((i.outer.inner*32) + (nb_j.inner*16)) + j.init)] = 0f32
-              }
-              for (elem_idx: int32, 0, let cse_var_1: int32 = ((i0.outer.i1.outer.fused*2) + nb_j.inner) in (placeholder_3[(cse_var_1 + 1)] - placeholder_3[cse_var_1])) {
-                for (j: int32, 0, 16) {
-                  let cse_var_3: int32 = ((i0.outer.i1.outer.fused*2) + nb_j.inner)
-                  let cse_var_2: int32 = (((i.outer.inner*32) + (nb_j.inner*16)) + j)
-                  compute_5[cse_var_2] = (compute_5[cse_var_2] + (placeholder_1[(((placeholder_3[cse_var_3]*16) + (elem_idx*16)) + j)]*max(placeholder[((i.outer.inner*256) + placeholder_2[(placeholder_3[cse_var_3] + elem_idx)])], 0f32)))
+      preflattened_buffer_map = {placeholder_5: placeholder_15: Buffer(placeholder_10, float32, [128, 256], []), placeholder_6: placeholder_16: Buffer(placeholder_11, float32, [4916, 16, 1], []), placeholder_8: placeholder_17: Buffer(placeholder_13, int32, [33], []), placeholder_9: placeholder_18: Buffer(placeholder_14, float32, [128, 512], []), placeholder_7: placeholder_19: Buffer(placeholder_12, int32, [4916], []), compute_1: compute_3: Buffer(compute_2, float32, [128, 512], [])} {
+      for (i0.outer.i1.outer.fused: int32, 0, 128) "parallel" {
+        allocate(compute_4: Pointer(global float32), float32, [512]), storage_scope = global {
+          for (i.outer.inner: int32, 0, 4) {
+            let cse_var_2: int32 = floormod(i0.outer.i1.outer.fused, 32)
+            let cse_var_1: int32 = (i.outer.inner*128)
+             {
+              compute_5: Buffer(compute_4, float32, [512], [])[cse_var_1] = 0f32
+              compute_5[(cse_var_1 + 1)] = 0f32
+              compute_5[(cse_var_1 + 2)] = 0f32
+              compute_5[(cse_var_1 + 3)] = 0f32
+              compute_5[(cse_var_1 + 4)] = 0f32
+              compute_5[(cse_var_1 + 5)] = 0f32
+              compute_5[(cse_var_1 + 6)] = 0f32
+              compute_5[(cse_var_1 + 7)] = 0f32
+              compute_5[(cse_var_1 + 8)] = 0f32
+              compute_5[(cse_var_1 + 9)] = 0f32
+              compute_5[(cse_var_1 + 10)] = 0f32
+              compute_5[(cse_var_1 + 11)] = 0f32
+              compute_5[(cse_var_1 + 12)] = 0f32
+              compute_5[(cse_var_1 + 13)] = 0f32
+              compute_5[(cse_var_1 + 14)] = 0f32
+              compute_5[(cse_var_1 + 15)] = 0f32
+              compute_5[(cse_var_1 + 16)] = 0f32
+              compute_5[(cse_var_1 + 17)] = 0f32
+              compute_5[(cse_var_1 + 18)] = 0f32
+              compute_5[(cse_var_1 + 19)] = 0f32
+              compute_5[(cse_var_1 + 20)] = 0f32
+              compute_5[(cse_var_1 + 21)] = 0f32
+              compute_5[(cse_var_1 + 22)] = 0f32
+              compute_5[(cse_var_1 + 23)] = 0f32
+              compute_5[(cse_var_1 + 24)] = 0f32
+              compute_5[(cse_var_1 + 25)] = 0f32
+              compute_5[(cse_var_1 + 26)] = 0f32
+              compute_5[(cse_var_1 + 27)] = 0f32
+              compute_5[(cse_var_1 + 28)] = 0f32
+              compute_5[(cse_var_1 + 29)] = 0f32
+              compute_5[(cse_var_1 + 30)] = 0f32
+              compute_5[(cse_var_1 + 31)] = 0f32
+              compute_5[(cse_var_1 + 32)] = 0f32
+              compute_5[(cse_var_1 + 33)] = 0f32
+              compute_5[(cse_var_1 + 34)] = 0f32
+              compute_5[(cse_var_1 + 35)] = 0f32
+              compute_5[(cse_var_1 + 36)] = 0f32
+              compute_5[(cse_var_1 + 37)] = 0f32
+              compute_5[(cse_var_1 + 38)] = 0f32
+              compute_5[(cse_var_1 + 39)] = 0f32
+              compute_5[(cse_var_1 + 40)] = 0f32
+              compute_5[(cse_var_1 + 41)] = 0f32
+              compute_5[(cse_var_1 + 42)] = 0f32
+              compute_5[(cse_var_1 + 43)] = 0f32
+              compute_5[(cse_var_1 + 44)] = 0f32
+              compute_5[(cse_var_1 + 45)] = 0f32
+              compute_5[(cse_var_1 + 46)] = 0f32
+              compute_5[(cse_var_1 + 47)] = 0f32
+              compute_5[(cse_var_1 + 48)] = 0f32
+              compute_5[(cse_var_1 + 49)] = 0f32
+              compute_5[(cse_var_1 + 50)] = 0f32
+              compute_5[(cse_var_1 + 51)] = 0f32
+              compute_5[(cse_var_1 + 52)] = 0f32
+              compute_5[(cse_var_1 + 53)] = 0f32
+              compute_5[(cse_var_1 + 54)] = 0f32
+              compute_5[(cse_var_1 + 55)] = 0f32
+              compute_5[(cse_var_1 + 56)] = 0f32
+              compute_5[(cse_var_1 + 57)] = 0f32
+              compute_5[(cse_var_1 + 58)] = 0f32
+              compute_5[(cse_var_1 + 59)] = 0f32
+              compute_5[(cse_var_1 + 60)] = 0f32
+              compute_5[(cse_var_1 + 61)] = 0f32
+              compute_5[(cse_var_1 + 62)] = 0f32
+              compute_5[(cse_var_1 + 63)] = 0f32
+              compute_5[(cse_var_1 + 64)] = 0f32
+              compute_5[(cse_var_1 + 65)] = 0f32
+              compute_5[(cse_var_1 + 66)] = 0f32
+              compute_5[(cse_var_1 + 67)] = 0f32
+              compute_5[(cse_var_1 + 68)] = 0f32
+              compute_5[(cse_var_1 + 69)] = 0f32
+              compute_5[(cse_var_1 + 70)] = 0f32
+              compute_5[(cse_var_1 + 71)] = 0f32
+              compute_5[(cse_var_1 + 72)] = 0f32
+              compute_5[(cse_var_1 + 73)] = 0f32
+              compute_5[(cse_var_1 + 74)] = 0f32
+              compute_5[(cse_var_1 + 75)] = 0f32
+              compute_5[(cse_var_1 + 76)] = 0f32
+              compute_5[(cse_var_1 + 77)] = 0f32
+              compute_5[(cse_var_1 + 78)] = 0f32
+              compute_5[(cse_var_1 + 79)] = 0f32
+              compute_5[(cse_var_1 + 80)] = 0f32
+              compute_5[(cse_var_1 + 81)] = 0f32
+              compute_5[(cse_var_1 + 82)] = 0f32
+              compute_5[(cse_var_1 + 83)] = 0f32
+              compute_5[(cse_var_1 + 84)] = 0f32
+              compute_5[(cse_var_1 + 85)] = 0f32
+              compute_5[(cse_var_1 + 86)] = 0f32
+              compute_5[(cse_var_1 + 87)] = 0f32
+              compute_5[(cse_var_1 + 88)] = 0f32
+              compute_5[(cse_var_1 + 89)] = 0f32
+              compute_5[(cse_var_1 + 90)] = 0f32
+              compute_5[(cse_var_1 + 91)] = 0f32
+              compute_5[(cse_var_1 + 92)] = 0f32
+              compute_5[(cse_var_1 + 93)] = 0f32
+              compute_5[(cse_var_1 + 94)] = 0f32
+              compute_5[(cse_var_1 + 95)] = 0f32
+              compute_5[(cse_var_1 + 96)] = 0f32
+              compute_5[(cse_var_1 + 97)] = 0f32
+              compute_5[(cse_var_1 + 98)] = 0f32
+              compute_5[(cse_var_1 + 99)] = 0f32
+              compute_5[(cse_var_1 + 100)] = 0f32
+              compute_5[(cse_var_1 + 101)] = 0f32
+              compute_5[(cse_var_1 + 102)] = 0f32
+              compute_5[(cse_var_1 + 103)] = 0f32
+              compute_5[(cse_var_1 + 104)] = 0f32
+              compute_5[(cse_var_1 + 105)] = 0f32
+              compute_5[(cse_var_1 + 106)] = 0f32
+              compute_5[(cse_var_1 + 107)] = 0f32
+              compute_5[(cse_var_1 + 108)] = 0f32
+              compute_5[(cse_var_1 + 109)] = 0f32
+              compute_5[(cse_var_1 + 110)] = 0f32
+              compute_5[(cse_var_1 + 111)] = 0f32
+              compute_5[(cse_var_1 + 112)] = 0f32
+              compute_5[(cse_var_1 + 113)] = 0f32
+              compute_5[(cse_var_1 + 114)] = 0f32
+              compute_5[(cse_var_1 + 115)] = 0f32
+              compute_5[(cse_var_1 + 116)] = 0f32
+              compute_5[(cse_var_1 + 117)] = 0f32
+              compute_5[(cse_var_1 + 118)] = 0f32
+              compute_5[(cse_var_1 + 119)] = 0f32
+              compute_5[(cse_var_1 + 120)] = 0f32
+              compute_5[(cse_var_1 + 121)] = 0f32
+              compute_5[(cse_var_1 + 122)] = 0f32
+              compute_5[(cse_var_1 + 123)] = 0f32
+              compute_5[(cse_var_1 + 124)] = 0f32
+              compute_5[(cse_var_1 + 125)] = 0f32
+              compute_5[(cse_var_1 + 126)] = 0f32
+              compute_5[(cse_var_1 + 127)] = 0f32
+              for (elem_idx: int32, 0, (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])) {
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  compute_5[cse_var_1] = (compute_5[cse_var_1] + (placeholder_1[((placeholder_3[cse_var_2]*16) + (elem_idx*16))]*max(placeholder[(((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)])], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_3: int32 = (cse_var_1 + 1)
+                  compute_5[cse_var_3] = (compute_5[cse_var_3] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 1)]*max(placeholder[(((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)])], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_4: int32 = (cse_var_1 + 2)
+                  compute_5[cse_var_4] = (compute_5[cse_var_4] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 2)]*max(placeholder[(((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)])], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_5: int32 = (cse_var_1 + 3)
+                  compute_5[cse_var_5] = (compute_5[cse_var_5] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 3)]*max(placeholder[(((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)])], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_6: int32 = (cse_var_1 + 4)
+                  compute_5[cse_var_6] = (compute_5[cse_var_6] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 4)]*max(placeholder[(((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)])], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_7: int32 = (cse_var_1 + 5)
+                  compute_5[cse_var_7] = (compute_5[cse_var_7] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 5)]*max(placeholder[(((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)])], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_8: int32 = (cse_var_1 + 6)
+                  compute_5[cse_var_8] = (compute_5[cse_var_8] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 6)]*max(placeholder[(((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)])], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_9: int32 = (cse_var_1 + 7)
+                  compute_5[cse_var_9] = (compute_5[cse_var_9] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 7)]*max(placeholder[(((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)])], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_10: int32 = (cse_var_1 + 8)
+                  compute_5[cse_var_10] = (compute_5[cse_var_10] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 8)]*max(placeholder[(((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)])], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_11: int32 = (cse_var_1 + 9)
+                  compute_5[cse_var_11] = (compute_5[cse_var_11] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 9)]*max(placeholder[(((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)])], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_12: int32 = (cse_var_1 + 10)
+                  compute_5[cse_var_12] = (compute_5[cse_var_12] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 10)]*max(placeholder[(((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)])], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_13: int32 = (cse_var_1 + 11)
+                  compute_5[cse_var_13] = (compute_5[cse_var_13] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 11)]*max(placeholder[(((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)])], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_14: int32 = (cse_var_1 + 12)
+                  compute_5[cse_var_14] = (compute_5[cse_var_14] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 12)]*max(placeholder[(((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)])], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_15: int32 = (cse_var_1 + 13)
+                  compute_5[cse_var_15] = (compute_5[cse_var_15] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 13)]*max(placeholder[(((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)])], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_16: int32 = (cse_var_1 + 14)
+                  compute_5[cse_var_16] = (compute_5[cse_var_16] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 14)]*max(placeholder[(((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)])], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_17: int32 = (cse_var_1 + 15)
+                  compute_5[cse_var_17] = (compute_5[cse_var_17] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 15)]*max(placeholder[(((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)])], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_18: int32 = (cse_var_1 + 16)
+                  compute_5[cse_var_18] = (compute_5[cse_var_18] + (placeholder_1[((placeholder_3[cse_var_2]*16) + (elem_idx*16))]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 256)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_19: int32 = (cse_var_1 + 17)
+                  compute_5[cse_var_19] = (compute_5[cse_var_19] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 1)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 256)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_20: int32 = (cse_var_1 + 18)
+                  compute_5[cse_var_20] = (compute_5[cse_var_20] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 2)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 256)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_21: int32 = (cse_var_1 + 19)
+                  compute_5[cse_var_21] = (compute_5[cse_var_21] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 3)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 256)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_22: int32 = (cse_var_1 + 20)
+                  compute_5[cse_var_22] = (compute_5[cse_var_22] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 4)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 256)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_23: int32 = (cse_var_1 + 21)
+                  compute_5[cse_var_23] = (compute_5[cse_var_23] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 5)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 256)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_24: int32 = (cse_var_1 + 22)
+                  compute_5[cse_var_24] = (compute_5[cse_var_24] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 6)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 256)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_25: int32 = (cse_var_1 + 23)
+                  compute_5[cse_var_25] = (compute_5[cse_var_25] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 7)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 256)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_26: int32 = (cse_var_1 + 24)
+                  compute_5[cse_var_26] = (compute_5[cse_var_26] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 8)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 256)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_27: int32 = (cse_var_1 + 25)
+                  compute_5[cse_var_27] = (compute_5[cse_var_27] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 9)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 256)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_28: int32 = (cse_var_1 + 26)
+                  compute_5[cse_var_28] = (compute_5[cse_var_28] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 10)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 256)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_29: int32 = (cse_var_1 + 27)
+                  compute_5[cse_var_29] = (compute_5[cse_var_29] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 11)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 256)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_30: int32 = (cse_var_1 + 28)
+                  compute_5[cse_var_30] = (compute_5[cse_var_30] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 12)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 256)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_31: int32 = (cse_var_1 + 29)
+                  compute_5[cse_var_31] = (compute_5[cse_var_31] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 13)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 256)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_32: int32 = (cse_var_1 + 30)
+                  compute_5[cse_var_32] = (compute_5[cse_var_32] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 14)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 256)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_33: int32 = (cse_var_1 + 31)
+                  compute_5[cse_var_33] = (compute_5[cse_var_33] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 15)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 256)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_34: int32 = (cse_var_1 + 32)
+                  compute_5[cse_var_34] = (compute_5[cse_var_34] + (placeholder_1[((placeholder_3[cse_var_2]*16) + (elem_idx*16))]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 512)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_35: int32 = (cse_var_1 + 33)
+                  compute_5[cse_var_35] = (compute_5[cse_var_35] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 1)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 512)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_36: int32 = (cse_var_1 + 34)
+                  compute_5[cse_var_36] = (compute_5[cse_var_36] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 2)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 512)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_37: int32 = (cse_var_1 + 35)
+                  compute_5[cse_var_37] = (compute_5[cse_var_37] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 3)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 512)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_38: int32 = (cse_var_1 + 36)
+                  compute_5[cse_var_38] = (compute_5[cse_var_38] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 4)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 512)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_39: int32 = (cse_var_1 + 37)
+                  compute_5[cse_var_39] = (compute_5[cse_var_39] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 5)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 512)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_40: int32 = (cse_var_1 + 38)
+                  compute_5[cse_var_40] = (compute_5[cse_var_40] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 6)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 512)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_41: int32 = (cse_var_1 + 39)
+                  compute_5[cse_var_41] = (compute_5[cse_var_41] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 7)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 512)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_42: int32 = (cse_var_1 + 40)
+                  compute_5[cse_var_42] = (compute_5[cse_var_42] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 8)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 512)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_43: int32 = (cse_var_1 + 41)
+                  compute_5[cse_var_43] = (compute_5[cse_var_43] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 9)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 512)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_44: int32 = (cse_var_1 + 42)
+                  compute_5[cse_var_44] = (compute_5[cse_var_44] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 10)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 512)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_45: int32 = (cse_var_1 + 43)
+                  compute_5[cse_var_45] = (compute_5[cse_var_45] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 11)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 512)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_46: int32 = (cse_var_1 + 44)
+                  compute_5[cse_var_46] = (compute_5[cse_var_46] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 12)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 512)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_47: int32 = (cse_var_1 + 45)
+                  compute_5[cse_var_47] = (compute_5[cse_var_47] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 13)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 512)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_48: int32 = (cse_var_1 + 46)
+                  compute_5[cse_var_48] = (compute_5[cse_var_48] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 14)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 512)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_49: int32 = (cse_var_1 + 47)
+                  compute_5[cse_var_49] = (compute_5[cse_var_49] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 15)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 512)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_50: int32 = (cse_var_1 + 48)
+                  compute_5[cse_var_50] = (compute_5[cse_var_50] + (placeholder_1[((placeholder_3[cse_var_2]*16) + (elem_idx*16))]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 768)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_51: int32 = (cse_var_1 + 49)
+                  compute_5[cse_var_51] = (compute_5[cse_var_51] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 1)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 768)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_52: int32 = (cse_var_1 + 50)
+                  compute_5[cse_var_52] = (compute_5[cse_var_52] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 2)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 768)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_53: int32 = (cse_var_1 + 51)
+                  compute_5[cse_var_53] = (compute_5[cse_var_53] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 3)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 768)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_54: int32 = (cse_var_1 + 52)
+                  compute_5[cse_var_54] = (compute_5[cse_var_54] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 4)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 768)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_55: int32 = (cse_var_1 + 53)
+                  compute_5[cse_var_55] = (compute_5[cse_var_55] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 5)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 768)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_56: int32 = (cse_var_1 + 54)
+                  compute_5[cse_var_56] = (compute_5[cse_var_56] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 6)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 768)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_57: int32 = (cse_var_1 + 55)
+                  compute_5[cse_var_57] = (compute_5[cse_var_57] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 7)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 768)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_58: int32 = (cse_var_1 + 56)
+                  compute_5[cse_var_58] = (compute_5[cse_var_58] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 8)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 768)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_59: int32 = (cse_var_1 + 57)
+                  compute_5[cse_var_59] = (compute_5[cse_var_59] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 9)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 768)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_60: int32 = (cse_var_1 + 58)
+                  compute_5[cse_var_60] = (compute_5[cse_var_60] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 10)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 768)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_61: int32 = (cse_var_1 + 59)
+                  compute_5[cse_var_61] = (compute_5[cse_var_61] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 11)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 768)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_62: int32 = (cse_var_1 + 60)
+                  compute_5[cse_var_62] = (compute_5[cse_var_62] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 12)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 768)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_63: int32 = (cse_var_1 + 61)
+                  compute_5[cse_var_63] = (compute_5[cse_var_63] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 13)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 768)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_64: int32 = (cse_var_1 + 62)
+                  compute_5[cse_var_64] = (compute_5[cse_var_64] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 14)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 768)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_65: int32 = (cse_var_1 + 63)
+                  compute_5[cse_var_65] = (compute_5[cse_var_65] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 15)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 768)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_66: int32 = (cse_var_1 + 64)
+                  compute_5[cse_var_66] = (compute_5[cse_var_66] + (placeholder_1[((placeholder_3[cse_var_2]*16) + (elem_idx*16))]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1024)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_67: int32 = (cse_var_1 + 65)
+                  compute_5[cse_var_67] = (compute_5[cse_var_67] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 1)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1024)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_68: int32 = (cse_var_1 + 66)
+                  compute_5[cse_var_68] = (compute_5[cse_var_68] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 2)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1024)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_69: int32 = (cse_var_1 + 67)
+                  compute_5[cse_var_69] = (compute_5[cse_var_69] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 3)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1024)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_70: int32 = (cse_var_1 + 68)
+                  compute_5[cse_var_70] = (compute_5[cse_var_70] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 4)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1024)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_71: int32 = (cse_var_1 + 69)
+                  compute_5[cse_var_71] = (compute_5[cse_var_71] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 5)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1024)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_72: int32 = (cse_var_1 + 70)
+                  compute_5[cse_var_72] = (compute_5[cse_var_72] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 6)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1024)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_73: int32 = (cse_var_1 + 71)
+                  compute_5[cse_var_73] = (compute_5[cse_var_73] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 7)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1024)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_74: int32 = (cse_var_1 + 72)
+                  compute_5[cse_var_74] = (compute_5[cse_var_74] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 8)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1024)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_75: int32 = (cse_var_1 + 73)
+                  compute_5[cse_var_75] = (compute_5[cse_var_75] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 9)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1024)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_76: int32 = (cse_var_1 + 74)
+                  compute_5[cse_var_76] = (compute_5[cse_var_76] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 10)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1024)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_77: int32 = (cse_var_1 + 75)
+                  compute_5[cse_var_77] = (compute_5[cse_var_77] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 11)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1024)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_78: int32 = (cse_var_1 + 76)
+                  compute_5[cse_var_78] = (compute_5[cse_var_78] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 12)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1024)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_79: int32 = (cse_var_1 + 77)
+                  compute_5[cse_var_79] = (compute_5[cse_var_79] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 13)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1024)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_80: int32 = (cse_var_1 + 78)
+                  compute_5[cse_var_80] = (compute_5[cse_var_80] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 14)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1024)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_81: int32 = (cse_var_1 + 79)
+                  compute_5[cse_var_81] = (compute_5[cse_var_81] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 15)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1024)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_82: int32 = (cse_var_1 + 80)
+                  compute_5[cse_var_82] = (compute_5[cse_var_82] + (placeholder_1[((placeholder_3[cse_var_2]*16) + (elem_idx*16))]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1280)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_83: int32 = (cse_var_1 + 81)
+                  compute_5[cse_var_83] = (compute_5[cse_var_83] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 1)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1280)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_84: int32 = (cse_var_1 + 82)
+                  compute_5[cse_var_84] = (compute_5[cse_var_84] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 2)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1280)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_85: int32 = (cse_var_1 + 83)
+                  compute_5[cse_var_85] = (compute_5[cse_var_85] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 3)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1280)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_86: int32 = (cse_var_1 + 84)
+                  compute_5[cse_var_86] = (compute_5[cse_var_86] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 4)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1280)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_87: int32 = (cse_var_1 + 85)
+                  compute_5[cse_var_87] = (compute_5[cse_var_87] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 5)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1280)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_88: int32 = (cse_var_1 + 86)
+                  compute_5[cse_var_88] = (compute_5[cse_var_88] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 6)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1280)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_89: int32 = (cse_var_1 + 87)
+                  compute_5[cse_var_89] = (compute_5[cse_var_89] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 7)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1280)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_90: int32 = (cse_var_1 + 88)
+                  compute_5[cse_var_90] = (compute_5[cse_var_90] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 8)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1280)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_91: int32 = (cse_var_1 + 89)
+                  compute_5[cse_var_91] = (compute_5[cse_var_91] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 9)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1280)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_92: int32 = (cse_var_1 + 90)
+                  compute_5[cse_var_92] = (compute_5[cse_var_92] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 10)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1280)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_93: int32 = (cse_var_1 + 91)
+                  compute_5[cse_var_93] = (compute_5[cse_var_93] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 11)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1280)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_94: int32 = (cse_var_1 + 92)
+                  compute_5[cse_var_94] = (compute_5[cse_var_94] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 12)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1280)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_95: int32 = (cse_var_1 + 93)
+                  compute_5[cse_var_95] = (compute_5[cse_var_95] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 13)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1280)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_96: int32 = (cse_var_1 + 94)
+                  compute_5[cse_var_96] = (compute_5[cse_var_96] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 14)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1280)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_97: int32 = (cse_var_1 + 95)
+                  compute_5[cse_var_97] = (compute_5[cse_var_97] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 15)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1280)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_98: int32 = (cse_var_1 + 96)
+                  compute_5[cse_var_98] = (compute_5[cse_var_98] + (placeholder_1[((placeholder_3[cse_var_2]*16) + (elem_idx*16))]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1536)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_99: int32 = (cse_var_1 + 97)
+                  compute_5[cse_var_99] = (compute_5[cse_var_99] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 1)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1536)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_100: int32 = (cse_var_1 + 98)
+                  compute_5[cse_var_100] = (compute_5[cse_var_100] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 2)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1536)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_101: int32 = (cse_var_1 + 99)
+                  compute_5[cse_var_101] = (compute_5[cse_var_101] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 3)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1536)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_102: int32 = (cse_var_1 + 100)
+                  compute_5[cse_var_102] = (compute_5[cse_var_102] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 4)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1536)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_103: int32 = (cse_var_1 + 101)
+                  compute_5[cse_var_103] = (compute_5[cse_var_103] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 5)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1536)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_104: int32 = (cse_var_1 + 102)
+                  compute_5[cse_var_104] = (compute_5[cse_var_104] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 6)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1536)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_105: int32 = (cse_var_1 + 103)
+                  compute_5[cse_var_105] = (compute_5[cse_var_105] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 7)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1536)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_106: int32 = (cse_var_1 + 104)
+                  compute_5[cse_var_106] = (compute_5[cse_var_106] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 8)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1536)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_107: int32 = (cse_var_1 + 105)
+                  compute_5[cse_var_107] = (compute_5[cse_var_107] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 9)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1536)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_108: int32 = (cse_var_1 + 106)
+                  compute_5[cse_var_108] = (compute_5[cse_var_108] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 10)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1536)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_109: int32 = (cse_var_1 + 107)
+                  compute_5[cse_var_109] = (compute_5[cse_var_109] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 11)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1536)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_110: int32 = (cse_var_1 + 108)
+                  compute_5[cse_var_110] = (compute_5[cse_var_110] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 12)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1536)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_111: int32 = (cse_var_1 + 109)
+                  compute_5[cse_var_111] = (compute_5[cse_var_111] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 13)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1536)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_112: int32 = (cse_var_1 + 110)
+                  compute_5[cse_var_112] = (compute_5[cse_var_112] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 14)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1536)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_113: int32 = (cse_var_1 + 111)
+                  compute_5[cse_var_113] = (compute_5[cse_var_113] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 15)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1536)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_114: int32 = (cse_var_1 + 112)
+                  compute_5[cse_var_114] = (compute_5[cse_var_114] + (placeholder_1[((placeholder_3[cse_var_2]*16) + (elem_idx*16))]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1792)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_115: int32 = (cse_var_1 + 113)
+                  compute_5[cse_var_115] = (compute_5[cse_var_115] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 1)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1792)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_116: int32 = (cse_var_1 + 114)
+                  compute_5[cse_var_116] = (compute_5[cse_var_116] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 2)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1792)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_117: int32 = (cse_var_1 + 115)
+                  compute_5[cse_var_117] = (compute_5[cse_var_117] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 3)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1792)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_118: int32 = (cse_var_1 + 116)
+                  compute_5[cse_var_118] = (compute_5[cse_var_118] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 4)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1792)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_119: int32 = (cse_var_1 + 117)
+                  compute_5[cse_var_119] = (compute_5[cse_var_119] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 5)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1792)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_120: int32 = (cse_var_1 + 118)
+                  compute_5[cse_var_120] = (compute_5[cse_var_120] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 6)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1792)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_121: int32 = (cse_var_1 + 119)
+                  compute_5[cse_var_121] = (compute_5[cse_var_121] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 7)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1792)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_122: int32 = (cse_var_1 + 120)
+                  compute_5[cse_var_122] = (compute_5[cse_var_122] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 8)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1792)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_123: int32 = (cse_var_1 + 121)
+                  compute_5[cse_var_123] = (compute_5[cse_var_123] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 9)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1792)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_124: int32 = (cse_var_1 + 122)
+                  compute_5[cse_var_124] = (compute_5[cse_var_124] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 10)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1792)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_125: int32 = (cse_var_1 + 123)
+                  compute_5[cse_var_125] = (compute_5[cse_var_125] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 11)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1792)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_126: int32 = (cse_var_1 + 124)
+                  compute_5[cse_var_126] = (compute_5[cse_var_126] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 12)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1792)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_127: int32 = (cse_var_1 + 125)
+                  compute_5[cse_var_127] = (compute_5[cse_var_127] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 13)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1792)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_128: int32 = (cse_var_1 + 126)
+                  compute_5[cse_var_128] = (compute_5[cse_var_128] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 14)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1792)], 0f32)))
+                }
+                if @tir.likely((elem_idx < (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+                  let cse_var_129: int32 = (cse_var_1 + 127)
+                  compute_5[cse_var_129] = (compute_5[cse_var_129] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 15)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1792)], 0f32)))
                 }
               }
             }
           }
-          for (i0.inner: int32, 0, 128) {
-            for (i1.inner: int32, 0, 32) {
-              let cse_var_4: int32 = (((i0.inner*512) + (i0.outer.i1.outer.fused*32)) + i1.inner)
-              compute[cse_var_4] = max((compute_5[((i0.inner*32) + i1.inner)] + placeholder_4[cse_var_4]), 0f32)
-            }
+          for (i0.inner: int32, 0, 32) {
+            let cse_var_130: int32 = (((floordiv(i0.outer.i1.outer.fused, 32)*16384) + (i0.inner*512)) + (floormod(i0.outer.i1.outer.fused, 32)*16))
+            compute[ramp(cse_var_130, 1, 16)] = max((compute_5[ramp((i0.inner*16), 1, 16)] + placeholder_4[ramp(cse_var_130, 1, 16)]), broadcast(0f32, 16))
           }
         }
       }
@@ -474,7 +1105,7 @@ We build the binary and check its correctness and performance.
 
  .. code-block:: none
 
-    Execution time of this operator: 2.279 ms
+    Execution time of this operator: 2.721 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 5b6b1c830..0136fb84c 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
 =================
-**00:45.604** total execution time for **how_to_tune_with_autotvm** files:
+**00:45.336** total execution time for **how_to_tune_with_autotvm** files:
 
 +--------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autotvm_tune_conv2d_cuda.py` (``tune_conv2d_cuda.py``)           | 00:45.569 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autotvm_tune_conv2d_cuda.py` (``tune_conv2d_cuda.py``)           | 00:45.301 | 0.0 MB |
 +--------------------------------------------------------------------------------------------------+-----------+--------+
 | :ref:`sphx_glr_how_to_tune_with_autotvm_tune_relay_x86.py` (``tune_relay_x86.py``)               | 00:00.020 | 0.0 MB |
 +--------------------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/how_to/tune_with_autotvm/tune_conv2d_cuda.rst.txt b/docs/_sources/how_to/tune_with_autotvm/tune_conv2d_cuda.rst.txt
index 0e0d157a4..5f9c7b0bb 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
@@ -1156,8 +1156,8 @@ for this template
     TimeoutError
 
             [('tile_f', [-1, 2, 1, 64]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 1, 4]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,4909501
-    No: 9   GFLOPS: 120.19/120.19   result: MeasureResult(costs=(0.0019261711071428573,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.9952189922332764, timestamp=1658794515.151304)       [('tile_f', [-1, 1, 4, 8]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 2, 2]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,5072689
-    No: 10  GFLOPS: 0.00/120.19     result: Traceback (most recent call last):
+    No: 9   GFLOPS: 80.74/80.74     result: MeasureResult(costs=(0.002867357257142857,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.8408534526824951, timestamp=1658794692.6519725)       [('tile_f', [-1, 1, 4, 8]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 2, 2]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,5072689
+    No: 10  GFLOPS: 0.00/80.74      result: Traceback (most recent call last):
       File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 588, in __call__
         func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
       File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 540, in _build_func_common
@@ -1280,8 +1280,8 @@ for this template
       File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 871, in verify_pass
         raise InstantiationError("Skipped because of invalid gpu kernel")
     tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel        [('tile_f', [-1, 4, 4, 8]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 64, 2]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,5092711
-    No: 11  GFLOPS: 261.18/261.18   result: MeasureResult(costs=(0.0008863828011049723,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.4688620567321777, timestamp=1658794516.0824132)      [('tile_f', [-1, 8, 2, 1]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 2, 1]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,4264713
-    No: 12  GFLOPS: 0.00/261.18     result: Traceback (most recent call last):
+    No: 11  GFLOPS: 259.82/259.82   result: MeasureResult(costs=(0.0008910134088397789,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.4664301872253418, timestamp=1658794693.5753386)      [('tile_f', [-1, 8, 2, 1]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 2, 1]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,4264713
+    No: 12  GFLOPS: 0.00/259.82     result: Traceback (most recent call last):
       File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 588, in __call__
         func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
       File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 540, in _build_func_common
@@ -1404,7 +1404,7 @@ for this template
       File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 871, in verify_pass
         raise InstantiationError("Skipped because of invalid gpu kernel")
     tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel        [('tile_f', [-1, 128, 1, 2]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 1, 256]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,183542
-    No: 13  GFLOPS: 0.00/261.18     result: Traceback (most recent call last):
+    No: 13  GFLOPS: 0.00/259.82     result: Traceback (most recent call last):
       File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 588, in __call__
         func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
       File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 540, in _build_func_common
@@ -1527,7 +1527,7 @@ for this template
       File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 871, in verify_pass
         raise InstantiationError("Skipped because of invalid gpu kernel")
     tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel        [('tile_f', [-1, 4, 8, 8]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 1, 64]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,2482196
-    No: 14  GFLOPS: 0.00/261.18     result: Traceback (most recent call last):
+    No: 14  GFLOPS: 0.00/259.82     result: Traceback (most recent call last):
       File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 588, in __call__
         func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
       File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 540, in _build_func_common
@@ -1650,9 +1650,9 @@ for this template
       File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 871, in verify_pass
         raise InstantiationError("Skipped because of invalid gpu kernel")
     tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel        [('tile_f', [-1, 64, 1, 4]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 4, 2]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,10306226
-    No: 15  GFLOPS: 5.29/261.18     result: MeasureResult(costs=(0.043745694,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.8309924602508545, timestamp=1658794520.5886626)        [('tile_f', [-1, 2, 2, 8]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 4, 8]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,5330964
-    No: 16  GFLOPS: 3.34/261.18     result: MeasureResult(costs=(0.06941130225,), error_no=MeasureErrorNo.NO_ERROR, all_cost=4.529698848724365, timestamp=1658794521.8308659)       [('tile_f', [-1, 8, 4, 4]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 4, 1]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,2140058
-    No: 17  GFLOPS: 0.00/261.18     result: Traceback (most recent call last):
+    No: 15  GFLOPS: 5.29/259.82     result: MeasureResult(costs=(0.0437411905,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.825692892074585, timestamp=1658794698.0947344)        [('tile_f', [-1, 2, 2, 8]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 4, 8]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,5330964
+    No: 16  GFLOPS: 3.34/259.82     result: MeasureResult(costs=(0.0693275815,), error_no=MeasureErrorNo.NO_ERROR, all_cost=4.529665470123291, timestamp=1658794699.329122) [('tile_f', [-1, 8, 4, 4]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 4, 1]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,2140058
+    No: 17  GFLOPS: 0.00/259.82     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
@@ -1670,8 +1670,8 @@ for this template
     TimeoutError
 
             [('tile_f', [-1, 2, 2, 1]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 4, 16]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,10195251
-    No: 18  GFLOPS: 27.94/261.18    result: MeasureResult(costs=(0.008284331857142857,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.269599199295044, timestamp=1658794532.8574402)        [('tile_f', [-1, 4, 8, 4]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 1, 4]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,6068603
-    No: 19  GFLOPS: 0.00/261.18     result: Traceback (most recent call last):
+    No: 18  GFLOPS: 28.14/259.82    result: MeasureResult(costs=(0.008227899714285714,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.2651729583740234, timestamp=1658794710.3470612)       [('tile_f', [-1, 4, 8, 4]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 1, 4]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,6068603
+    No: 19  GFLOPS: 0.00/259.82     result: Traceback (most recent call last):
       File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 588, in __call__
         func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
       File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 540, in _build_func_common
@@ -1794,7 +1794,7 @@ for this template
       File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 871, in verify_pass
         raise InstantiationError("Skipped because of invalid gpu kernel")
     tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel        [('tile_f', [-1, 16, 4, 8]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 4, 128]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,6956993
-    No: 20  GFLOPS: 0.00/261.18     result: Traceback (most recent call last):
+    No: 20  GFLOPS: 0.00/259.82     result: Traceback (most recent call last):
       File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 588, in __call__
         func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
       File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 540, in _build_func_common
@@ -1973,7 +1973,7 @@ and measure running time.
     Best config:
     [('tile_f', [-1, 8, 2, 1]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 2, 1]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,4264713
     Finish loading 20 records
-    Time cost of this operator: 0.001283
+    Time cost of this operator: 0.001279
 
 
 
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 7261f847e..c7cb739e3 100644
--- a/docs/_sources/how_to/work_with_microtvm/micro_autotune.rst.txt
+++ b/docs/_sources/how_to/work_with_microtvm/micro_autotune.rst.txt
@@ -329,10 +329,10 @@ Timing the untuned program
     ########## Build without Autotuning ##########
     Node Name                                     Ops                                           Time(us)  Time(%)  Shape              Inputs  Outputs  Measurements(us)  
     ---------                                     ---                                           --------  -------  -----              ------  -------  ----------------  
-    tvmgen_default_fused_nn_contrib_conv2d_NCHWc  tvmgen_default_fused_nn_contrib_conv2d_NCHWc  309.9     98.71    (1, 2, 10, 10, 3)  2       1        [309.9]           
-    tvmgen_default_fused_layout_transform_1       tvmgen_default_fused_layout_transform_1       3.078     0.98     (1, 6, 10, 10)     1       1        [3.078]           
-    tvmgen_default_fused_layout_transform         tvmgen_default_fused_layout_transform         0.97      0.309    (1, 1, 10, 10, 3)  1       1        [0.97]            
-    Total_time                                    -                                             313.948   -        -                  -       -        -                 
+    tvmgen_default_fused_nn_contrib_conv2d_NCHWc  tvmgen_default_fused_nn_contrib_conv2d_NCHWc  311.4     98.726   (1, 2, 10, 10, 3)  2       1        [311.4]           
+    tvmgen_default_fused_layout_transform_1       tvmgen_default_fused_layout_transform_1       3.06      0.97     (1, 6, 10, 10)     1       1        [3.06]            
+    tvmgen_default_fused_layout_transform         tvmgen_default_fused_layout_transform         0.957     0.303    (1, 1, 10, 10, 3)  1       1        [0.957]           
+    Total_time                                    -                                             315.417   -        -                  -       -        -                 
 
 
 
@@ -398,10 +398,10 @@ Timing the tuned program
     ########## Build with Autotuning ##########
     Node Name                                     Ops                                           Time(us)  Time(%)  Shape              Inputs  Outputs  Measurements(us)  
     ---------                                     ---                                           --------  -------  -----              ------  -------  ----------------  
-    tvmgen_default_fused_nn_contrib_conv2d_NCHWc  tvmgen_default_fused_nn_contrib_conv2d_NCHWc  80.438    96.718   (1, 6, 10, 10, 1)  2       1        [80.438]          
-    tvmgen_default_fused_layout_transform_1       tvmgen_default_fused_layout_transform_1       1.766     2.123    (1, 6, 10, 10)     1       1        [1.766]           
-    tvmgen_default_fused_layout_transform         tvmgen_default_fused_layout_transform         0.964     1.159    (1, 1, 10, 10, 3)  1       1        [0.964]           
-    Total_time                                    -                                             83.167    -        -                  -       -        -                 
+    tvmgen_default_fused_nn_contrib_conv2d_NCHWc  tvmgen_default_fused_nn_contrib_conv2d_NCHWc  225.2     98.614   (1, 1, 10, 10, 6)  2       1        [225.2]           
+    tvmgen_default_fused_layout_transform_1       tvmgen_default_fused_layout_transform_1       2.186     0.957    (1, 6, 10, 10)     1       1        [2.186]           
+    tvmgen_default_fused_layout_transform         tvmgen_default_fused_layout_transform         0.978     0.428    (1, 1, 10, 10, 3)  1       1        [0.978]           
+    Total_time                                    -                                             228.364   -        -                  -       -        -                 
 
 
 
diff --git a/docs/_sources/how_to/work_with_microtvm/micro_train.rst.txt b/docs/_sources/how_to/work_with_microtvm/micro_train.rst.txt
index 6822d709d..bd1441cee 100644
--- a/docs/_sources/how_to/work_with_microtvm/micro_train.rst.txt
+++ b/docs/_sources/how_to/work_with_microtvm/micro_train.rst.txt
@@ -225,7 +225,7 @@ take about **2 minutes** to download the Stanford Cars, while COCO 2017 validati
  .. code-block:: none
 
 
-    '/tmp/tmpp0m4t4w1/images/random'
+    '/tmp/tmp8hc9tzj6/images/random'
 
 
 
@@ -325,8 +325,8 @@ objects to other stuff? We can display some examples from our datasets using ``m
 
  .. code-block:: none
 
-    /tmp/tmpp0m4t4w1/images/target contains 8144 images
-    /tmp/tmpp0m4t4w1/images/random contains 5000 images
+    /tmp/tmp8hc9tzj6/images/target contains 8144 images
+    /tmp/tmp8hc9tzj6/images/random contains 5000 images
 
 
 
@@ -501,13 +501,13 @@ the time on our validation set).
  .. code-block:: none
 
     Epoch 1/3
-    328/328 - 55s - loss: 0.2405 - accuracy: 0.9187 - val_loss: 0.1651 - val_accuracy: 0.9468
+    328/328 - 55s - loss: 0.2231 - accuracy: 0.9217 - val_loss: 0.1474 - val_accuracy: 0.9524
     Epoch 2/3
-    328/328 - 52s - loss: 0.1058 - accuracy: 0.9609 - val_loss: 0.1311 - val_accuracy: 0.9573
+    328/328 - 52s - loss: 0.1028 - accuracy: 0.9599 - val_loss: 0.1210 - val_accuracy: 0.9596
     Epoch 3/3
-    328/328 - 52s - loss: 0.0699 - accuracy: 0.9730 - val_loss: 0.1189 - val_accuracy: 0.9619
+    328/328 - 52s - loss: 0.0690 - accuracy: 0.9739 - val_loss: 0.1083 - val_accuracy: 0.9637
 
-    <keras.callbacks.History object at 0x7f855e3d0110>
+    <keras.callbacks.History object at 0x7fbc00a109d0>
 
 
 
@@ -864,7 +864,7 @@ Arduino tutorial for how to do that `on GitHub <https://github.com/guberti/tvm-a
 
 .. rst-class:: sphx-glr-timing
 
-   **Total running time of the script:** ( 5 minutes  9.984 seconds)
+   **Total running time of the script:** ( 5 minutes  8.039 seconds)
 
 
 .. _sphx_glr_download_how_to_work_with_microtvm_micro_train.py:
diff --git a/docs/_sources/how_to/work_with_microtvm/sg_execution_times.rst.txt b/docs/_sources/how_to/work_with_microtvm/sg_execution_times.rst.txt
index a335e0603..b1eabaa07 100644
--- a/docs/_sources/how_to/work_with_microtvm/sg_execution_times.rst.txt
+++ b/docs/_sources/how_to/work_with_microtvm/sg_execution_times.rst.txt
@@ -5,14 +5,14 @@
 
 Computation times
 =================
-**05:55.731** total execution time for **how_to_work_with_microtvm** files:
+**05:54.504** total execution time for **how_to_work_with_microtvm** files:
 
 +---------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_microtvm_micro_train.py` (``micro_train.py``)               | 05:09.984 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_microtvm_micro_train.py` (``micro_train.py``)               | 05:08.039 | 0.0 MB |
 +---------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_microtvm_micro_autotune.py` (``micro_autotune.py``)         | 00:42.489 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_microtvm_micro_autotune.py` (``micro_autotune.py``)         | 00:43.166 | 0.0 MB |
 +---------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_microtvm_micro_tflite.py` (``micro_tflite.py``)             | 00:03.257 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_microtvm_micro_tflite.py` (``micro_tflite.py``)             | 00:03.297 | 0.0 MB |
 +---------------------------------------------------------------------------------------------+-----------+--------+
 | :ref:`sphx_glr_how_to_work_with_microtvm_micro_ethosu.py` (``micro_ethosu.py``)             | 00:00.001 | 0.0 MB |
 +---------------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/how_to/work_with_relay/sg_execution_times.rst.txt b/docs/_sources/how_to/work_with_relay/sg_execution_times.rst.txt
index 306b4ed7c..ce1a2fa71 100644
--- a/docs/_sources/how_to/work_with_relay/sg_execution_times.rst.txt
+++ b/docs/_sources/how_to/work_with_relay/sg_execution_times.rst.txt
@@ -5,14 +5,14 @@
 
 Computation times
 =================
-**00:42.004** total execution time for **how_to_work_with_relay** files:
+**00:41.929** total execution time for **how_to_work_with_relay** files:
 
 +----------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_relay_using_pipeline_executor.py` (``using_pipeline_executor.py``) | 00:30.456 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_relay_using_pipeline_executor.py` (``using_pipeline_executor.py``) | 00:30.380 | 0.0 MB |
 +----------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_relay_using_external_lib.py` (``using_external_lib.py``)           | 00:09.916 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_relay_using_external_lib.py` (``using_external_lib.py``)           | 00:09.840 | 0.0 MB |
 +----------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_relay_build_gcn.py` (``build_gcn.py``)                             | 00:01.626 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_relay_build_gcn.py` (``build_gcn.py``)                             | 00:01.701 | 0.0 MB |
 +----------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_relay_using_relay_viz.py` (``using_relay_viz.py``)                 | 00:00.007 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_relay_using_relay_viz.py` (``using_relay_viz.py``)                 | 00:00.008 | 0.0 MB |
 +----------------------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/how_to/work_with_schedules/intrin_math.rst.txt b/docs/_sources/how_to/work_with_schedules/intrin_math.rst.txt
index 8d342e7fa..b78e9fe4b 100644
--- a/docs/_sources/how_to/work_with_schedules/intrin_math.rst.txt
+++ b/docs/_sources/how_to/work_with_schedules/intrin_math.rst.txt
@@ -261,7 +261,7 @@ The following example customizes CUDA lowering rule for :code:`exp`.
  .. code-block:: none
 
 
-    <function my_cuda_math_rule at 0x7f8556c6ce60>
+    <function my_cuda_math_rule at 0x7fbb85bdd050>
 
 
 
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 d4a2b0492..a69e132c6 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
@@ -5,20 +5,20 @@
 
 Computation times
 =================
-**00:04.042** total execution time for **how_to_work_with_schedules** files:
+**00:04.090** total execution time for **how_to_work_with_schedules** files:
 
 +------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_intrin_math.py` (``intrin_math.py``)                 | 00:01.893 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_intrin_math.py` (``intrin_math.py``)                 | 00:01.894 | 0.0 MB |
 +------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_tensorize.py` (``tensorize.py``)                     | 00:00.922 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_tensorize.py` (``tensorize.py``)                     | 00:00.966 | 0.0 MB |
 +------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_reduction.py` (``reduction.py``)                     | 00:00.528 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_reduction.py` (``reduction.py``)                     | 00:00.531 | 0.0 MB |
 +------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_scan.py` (``scan.py``)                               | 00:00.516 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_scan.py` (``scan.py``)                               | 00:00.514 | 0.0 MB |
 +------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_extern_op.py` (``extern_op.py``)                     | 00:00.100 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_extern_op.py` (``extern_op.py``)                     | 00:00.101 | 0.0 MB |
 +------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_schedule_primitives.py` (``schedule_primitives.py``) | 00:00.041 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_schedule_primitives.py` (``schedule_primitives.py``) | 00:00.042 | 0.0 MB |
 +------------------------------------------------------------------------------------------------+-----------+--------+
 | :ref:`sphx_glr_how_to_work_with_schedules_tedd.py` (``tedd.py``)                               | 00:00.027 | 0.0 MB |
 +------------------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/how_to/work_with_schedules/tensorize.rst.txt b/docs/_sources/how_to/work_with_schedules/tensorize.rst.txt
index 5f689e7ae..e4b3c50d2 100644
--- a/docs/_sources/how_to/work_with_schedules/tensorize.rst.txt
+++ b/docs/_sources/how_to/work_with_schedules/tensorize.rst.txt
@@ -347,7 +347,7 @@ The importing needs to happen before the tensorized GEMV being executed.
                  C: Buffer(C_2: Pointer(float32), float32, [524288], [])}
       buffer_map = {A_1: A, B_1: B, C_1: C}
       preflattened_buffer_map = {A_1: A_3: Buffer(A_2, float32, [1024, 64], []), B_1: B_3: Buffer(B_2, float32, [512, 64], []), C_1: C_3: Buffer(C_2, float32, [1024, 512], [])} {
-      attr [IterVar(i: int32, (nullptr), "DataPar", "")] "pragma_import_llvm" = "; ModuleID = '/tmp/tmp1q_31ci7/input0.cc'\nsource_filename = \"/tmp/tmp1q_31ci7/input0.cc\"\ntarget datalayout = \"e-m:e-i64:64-f80:128-n8:16:32:64-S128\"\ntarget triple = \"x86_64-pc-linux-gnu\"\n\n; Function Attrs: noinline nounwind optnone uwtable\ndefine dso_local i32 @gemv_update(float*, float*, float*, i32, i32, i32) #0 {\n  %7 = alloca float*, align 8\n  %8 = alloca float*, align 8\n  %9 = alloca floa [...]
+      attr [IterVar(i: int32, (nullptr), "DataPar", "")] "pragma_import_llvm" = "; ModuleID = '/tmp/tmp2orpuey9/input0.cc'\nsource_filename = \"/tmp/tmp2orpuey9/input0.cc\"\ntarget datalayout = \"e-m:e-i64:64-f80:128-n8:16:32:64-S128\"\ntarget triple = \"x86_64-pc-linux-gnu\"\n\n; Function Attrs: noinline nounwind optnone uwtable\ndefine dso_local i32 @gemv_update(float*, float*, float*, i32, i32, i32) #0 {\n  %7 = alloca float*, align 8\n  %8 = alloca float*, align 8\n  %9 = alloca floa [...]
       for (i, 0, 1024) {
         for (j.outer: int32, 0, 32) {
           @tir.call_extern("gemv_update", @tir.tvm_access_ptr(@tir.type_annotation(, dtype=float32), C_2, ((i*512) + (j.outer*16)), 16, 2, dtype=handle), @tir.tvm_access_ptr(@tir.type_annotation(, dtype=float32), A_2, (i*64), 64, 1, dtype=handle), @tir.tvm_access_ptr(@tir.type_annotation(, dtype=float32), B_2, (j.outer*1024), 1024, 1, dtype=handle), 16, 64, 64, dtype=int32)
diff --git a/docs/_sources/topic/vta/tutorials/autotvm/sg_execution_times.rst.txt b/docs/_sources/topic/vta/tutorials/autotvm/sg_execution_times.rst.txt
index ec6c1262f..1d77119e3 100644
--- a/docs/_sources/topic/vta/tutorials/autotvm/sg_execution_times.rst.txt
+++ b/docs/_sources/topic/vta/tutorials/autotvm/sg_execution_times.rst.txt
@@ -5,10 +5,10 @@
 
 Computation times
 =================
-**00:21.061** total execution time for **topic_vta_tutorials_autotvm** files:
+**00:21.301** total execution time for **topic_vta_tutorials_autotvm** files:
 
 +---------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_autotvm_tune_relay_vta.py` (``tune_relay_vta.py``) | 00:21.054 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_autotvm_tune_relay_vta.py` (``tune_relay_vta.py``) | 00:21.294 | 0.0 MB |
 +---------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_autotvm_tune_alu_vta.py` (``tune_alu_vta.py``)     | 00:00.006 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_autotvm_tune_alu_vta.py` (``tune_alu_vta.py``)     | 00:00.007 | 0.0 MB |
 +---------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/topic/vta/tutorials/frontend/deploy_classification.rst.txt b/docs/_sources/topic/vta/tutorials/frontend/deploy_classification.rst.txt
index 5b5ee93e9..be3451ac8 100644
--- a/docs/_sources/topic/vta/tutorials/frontend/deploy_classification.rst.txt
+++ b/docs/_sources/topic/vta/tutorials/frontend/deploy_classification.rst.txt
@@ -291,7 +291,7 @@ The compilation steps are:
       DeprecationWarning,
     /workspace/vta/tutorials/frontend/deploy_classification.py:213: DeprecationWarning: legacy graph executor behavior of producing json / lib / params will be removed in the next release. Please see documents of tvm.contrib.graph_executor.GraphModule for the  new recommended usage.
       relay_prog, target=tvm.target.Target(target, host=env.target_host), params=params
-    resnet18_v1 inference graph built in 22.58s!
+    resnet18_v1 inference graph built in 23.02s!
 
 
 
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 1b170eb7a..c956a106d 100644
--- a/docs/_sources/topic/vta/tutorials/frontend/deploy_detection.rst.txt
+++ b/docs/_sources/topic/vta/tutorials/frontend/deploy_detection.rst.txt
@@ -335,7 +335,7 @@ The compilation steps are:
       "target_host parameter is going to be deprecated. "
     /workspace/python/tvm/relay/build_module.py:411: DeprecationWarning: Please use input parameter mod (tvm.IRModule) instead of deprecated parameter mod (tvm.relay.function.Function)
       DeprecationWarning,
-    yolov3-tiny inference graph built in 15.76s!
+    yolov3-tiny inference graph built in 16.06s!
 
 
 
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 b0833e602..a21c7b838 100644
--- a/docs/_sources/topic/vta/tutorials/frontend/sg_execution_times.rst.txt
+++ b/docs/_sources/topic/vta/tutorials/frontend/sg_execution_times.rst.txt
@@ -5,10 +5,10 @@
 
 Computation times
 =================
-**01:31.547** total execution time for **topic_vta_tutorials_frontend** files:
+**01:32.213** total execution time for **topic_vta_tutorials_frontend** files:
 
 +------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_frontend_deploy_detection.py` (``deploy_detection.py``)           | 00:48.835 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_frontend_deploy_detection.py` (``deploy_detection.py``)           | 00:48.703 | 0.0 MB |
 +------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_frontend_deploy_classification.py` (``deploy_classification.py``) | 00:42.713 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_frontend_deploy_classification.py` (``deploy_classification.py``) | 00:43.510 | 0.0 MB |
 +------------------------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/topic/vta/tutorials/optimize/sg_execution_times.rst.txt b/docs/_sources/topic/vta/tutorials/optimize/sg_execution_times.rst.txt
index 570b8bb30..d0d3c7872 100644
--- a/docs/_sources/topic/vta/tutorials/optimize/sg_execution_times.rst.txt
+++ b/docs/_sources/topic/vta/tutorials/optimize/sg_execution_times.rst.txt
@@ -5,10 +5,10 @@
 
 Computation times
 =================
-**00:03.240** total execution time for **topic_vta_tutorials_optimize** files:
+**00:03.276** total execution time for **topic_vta_tutorials_optimize** files:
 
 +--------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_optimize_convolution_opt.py` (``convolution_opt.py``)         | 00:02.850 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_optimize_convolution_opt.py` (``convolution_opt.py``)         | 00:02.874 | 0.0 MB |
 +--------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_optimize_matrix_multiply_opt.py` (``matrix_multiply_opt.py``) | 00:00.390 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_optimize_matrix_multiply_opt.py` (``matrix_multiply_opt.py``) | 00:00.403 | 0.0 MB |
 +--------------------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/topic/vta/tutorials/sg_execution_times.rst.txt b/docs/_sources/topic/vta/tutorials/sg_execution_times.rst.txt
index 9acc024da..54942aea6 100644
--- a/docs/_sources/topic/vta/tutorials/sg_execution_times.rst.txt
+++ b/docs/_sources/topic/vta/tutorials/sg_execution_times.rst.txt
@@ -5,10 +5,10 @@
 
 Computation times
 =================
-**00:00.698** total execution time for **topic_vta_tutorials** files:
+**00:00.708** total execution time for **topic_vta_tutorials** files:
 
 +---------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_matrix_multiply.py` (``matrix_multiply.py``) | 00:00.370 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_matrix_multiply.py` (``matrix_multiply.py``) | 00:00.380 | 0.0 MB |
 +---------------------------------------------------------------------------------+-----------+--------+
 | :ref:`sphx_glr_topic_vta_tutorials_vta_get_started.py` (``vta_get_started.py``) | 00:00.328 | 0.0 MB |
 +---------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/tutorial/auto_scheduler_matmul_x86.rst.txt b/docs/_sources/tutorial/auto_scheduler_matmul_x86.rst.txt
index 5c3102814..797b8a7c0 100644
--- a/docs/_sources/tutorial/auto_scheduler_matmul_x86.rst.txt
+++ b/docs/_sources/tutorial/auto_scheduler_matmul_x86.rst.txt
@@ -328,7 +328,7 @@ We build the binary and check its correctness and performance.
 
  .. code-block:: none
 
-    Execution time of this operator: 93.635 ms
+    Execution time of this operator: 93.225 ms
 
 
 
diff --git a/docs/_sources/tutorial/autotvm_matmul_x86.rst.txt b/docs/_sources/tutorial/autotvm_matmul_x86.rst.txt
index e159de6dc..8c3a7ae41 100644
--- a/docs/_sources/tutorial/autotvm_matmul_x86.rst.txt
+++ b/docs/_sources/tutorial/autotvm_matmul_x86.rst.txt
@@ -462,16 +462,16 @@ reduce variance, we take 5 measurements and average them.
     waiting for device...
     device available
     Get devices for measurement successfully!
-    No: 1   GFLOPS: 9.82/9.82       result: MeasureResult(costs=(0.0273374958,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.573936939239502, timestamp=1658793316.72432)  [('tile_y', [-1, 1]), ('tile_x', [-1, 256])],None,80
-    No: 2   GFLOPS: 2.45/9.82       result: MeasureResult(costs=(0.10938282699999999,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.9021224975585938, timestamp=1658793318.6384602)        [('tile_y', [-1, 4]), ('tile_x', [-1, 8])],None,32
-    No: 3   GFLOPS: 11.81/11.81     result: MeasureResult(costs=(0.0227359086,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.5595288276672363, timestamp=1658793319.6902707)       [('tile_y', [-1, 64]), ('tile_x', [-1, 32])],None,56
-    No: 4   GFLOPS: 1.85/11.81      result: MeasureResult(costs=(0.14472830779999998,), error_no=MeasureErrorNo.NO_ERROR, all_cost=2.445192575454712, timestamp=1658793322.7089193) [('tile_y', [-1, 1]), ('tile_x', [-1, 4])],None,20
-    No: 5   GFLOPS: 3.64/11.81      result: MeasureResult(costs=(0.07374857180000001,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.313725233078003, timestamp=1658793324.1500275) [('tile_y', [-1, 256]), ('tile_x', [-1, 16])],None,48
-    No: 6   GFLOPS: 1.75/11.81      result: MeasureResult(costs=(0.15324395159999998,), error_no=MeasureErrorNo.NO_ERROR, all_cost=2.5731687545776367, timestamp=1658793327.2830582)        [('tile_y', [-1, 512]), ('tile_x', [-1, 4])],None,29
-    No: 7   GFLOPS: 0.81/11.81      result: MeasureResult(costs=(0.3316435378,), error_no=MeasureErrorNo.NO_ERROR, all_cost=5.410801649093628, timestamp=1658793332.7413354)        [('tile_y', [-1, 512]), ('tile_x', [-1, 2])],None,19
-    No: 8   GFLOPS: 10.07/11.81     result: MeasureResult(costs=(0.026644541800000004,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.5688803195953369, timestamp=1658793333.3296034)       [('tile_y', [-1, 4]), ('tile_x', [-1, 64])],None,62
-    No: 9   GFLOPS: 1.90/11.81      result: MeasureResult(costs=(0.1411162182,), error_no=MeasureErrorNo.NO_ERROR, all_cost=2.3581464290618896, timestamp=1658793335.8070061)       [('tile_y', [-1, 2]), ('tile_x', [-1, 2])],None,11
-    No: 10  GFLOPS: 2.49/11.81      result: MeasureResult(costs=(0.1076121892,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.829699993133545, timestamp=1658793337.6954505)        [('tile_y', [-1, 4]), ('tile_x', [-1, 4])],None,22
+    No: 1   GFLOPS: 9.18/9.18       result: MeasureResult(costs=(0.0292562832,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.604039192199707, timestamp=1658793502.5860975)        [('tile_y', [-1, 1]), ('tile_x', [-1, 256])],None,80
+    No: 2   GFLOPS: 2.52/9.18       result: MeasureResult(costs=(0.10647647099999999,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.8489928245544434, timestamp=1658793504.454151) [('tile_y', [-1, 4]), ('tile_x', [-1, 8])],None,32
+    No: 3   GFLOPS: 11.81/11.81     result: MeasureResult(costs=(0.0227294954,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.56583571434021, timestamp=1658793505.5104766) [('tile_y', [-1, 64]), ('tile_x', [-1, 32])],None,56
+    No: 4   GFLOPS: 1.65/11.81      result: MeasureResult(costs=(0.16275630140000003,), error_no=MeasureErrorNo.NO_ERROR, all_cost=2.728863477706909, timestamp=1658793508.2799802) [('tile_y', [-1, 1]), ('tile_x', [-1, 4])],None,20
+    No: 5   GFLOPS: 3.59/11.81      result: MeasureResult(costs=(0.0747437244,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.33174729347229, timestamp=1658793509.7422025) [('tile_y', [-1, 256]), ('tile_x', [-1, 16])],None,48
+    No: 6   GFLOPS: 1.72/11.81      result: MeasureResult(costs=(0.15638444680000002,), error_no=MeasureErrorNo.NO_ERROR, all_cost=2.6233057975769043, timestamp=1658793512.934545) [('tile_y', [-1, 512]), ('tile_x', [-1, 4])],None,29
+    No: 7   GFLOPS: 0.87/11.81      result: MeasureResult(costs=(0.3073094716,), error_no=MeasureErrorNo.NO_ERROR, all_cost=5.0365002155303955, timestamp=1658793518.5433748)       [('tile_y', [-1, 512]), ('tile_x', [-1, 2])],None,19
+    No: 8   GFLOPS: 10.64/11.81     result: MeasureResult(costs=(0.0252330394,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.5482997894287109, timestamp=1658793519.113209)        [('tile_y', [-1, 4]), ('tile_x', [-1, 64])],None,62
+    No: 9   GFLOPS: 1.91/11.81      result: MeasureResult(costs=(0.1404046786,), error_no=MeasureErrorNo.NO_ERROR, all_cost=2.357146978378296, timestamp=1658793521.5900376)        [('tile_y', [-1, 2]), ('tile_x', [-1, 2])],None,11
+    No: 10  GFLOPS: 2.65/11.81      result: MeasureResult(costs=(0.1012221128,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.724515438079834, timestamp=1658793523.3732836)        [('tile_y', [-1, 4]), ('tile_x', [-1, 4])],None,22
 
 
 
diff --git a/docs/_sources/tutorial/autotvm_relay_x86.rst.txt b/docs/_sources/tutorial/autotvm_relay_x86.rst.txt
index 63eb7b4e5..5e6f7e27c 100644
--- a/docs/_sources/tutorial/autotvm_relay_x86.rst.txt
+++ b/docs/_sources/tutorial/autotvm_relay_x86.rst.txt
@@ -327,7 +327,7 @@ standard deviation.
 
  .. code-block:: none
 
-    {'mean': 493.9443639300134, 'median': 494.06449910002266, 'std': 1.6148426721759408}
+    {'mean': 495.75926721000326, 'median': 495.88527695000266, 'std': 1.0175382126682093}
 
 
 
@@ -563,31 +563,30 @@ the tuning data to.
 
     /workspace/python/tvm/driver/build_module.py:268: UserWarning: target_host parameter is going to be deprecated. Please pass in tvm.target.Target(target, host=target_host) instead.
       "target_host parameter is going to be deprecated. "
-
    [Task  1/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task  1/25]  Current/Best:   17.53/  17.53 GFLOPS | Progress: (4/20) | 6.21 s
    [Task  1/25]  Current/Best:    6.05/  17.53 GFLOPS | Progress: (8/20) | 9.14 s
    [Task  1/25]  Current/Best:   11.55/  22.65 GFLOPS | Progress: (12/20) | 11.63 s
    [Task  1/25]  Current/Best:   16.92/  22.87 GFLOPS | Progress: (16/20) | 13.32 s
    [Task  1/25]  Current/Best:   11.61/  23.91 GFLOPS | Progress: (20/20) | 15.05 s Done.
-
    [Task  2/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task  2/25]  Current/Best:   12.30/  12.88 GFLOPS | Progress: (4/20) | 3.74 s
    [Task  2/25]  Current/Best:   14.11/  17.65 GFLOPS | Progress: (8/20) | 5.06 s
    [Task  2/25]  Current/Best:   21.11/  21.11 GFLOPS | Progress: (12/20) | 6.37 s
    [Task  2/25]  Current/Best:   12.34/  21.11 GFLOPS | Progress: (16/20) | 7.63 s
    [Task  2/25]  Current/Best:   19.90/  21.11 GFLOPS | Progress: (20/20) | 9.23 s Done.
-
    [Task  3/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task  3/25]  Current/Best:    1.63/  10.60 GFLOPS | Progress: (4/20) | 5.86 s
    [Task  3/25]  Current/Best:   15.62/  16.91 GFLOPS | Progress: (8/20) | 7.77 s
    [Task  3/25]  Current/Best:   14.90/  16.91 GFLOPS | Progress: (12/20) | 9.48 s
    [Task  3/25]  Current/Best:    7.22/  23.73 GFLOPS | Progress: (16/20) | 11.41 s
    [Task  3/25]  Current/Best:   12.86/  23.73 GFLOPS | Progress: (20/20) | 15.95 s Done.
-
    [Task  4/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task  4/25]  Current/Best:    9.68/  20.68 GFLOPS | Progress: (4/20) | 2.35 s
    [Task  4/25]  Current/Best:    6.76/  20.68 GFLOPS | Progress: (8/20) | 7.13 s
    [Task  4/25]  Current/Best:   22.03/  22.03 GFLOPS | Progress: (12/20) | 12.10 s
    [Task  4/25]  Current/Best:   17.38/  22.03 GFLOPS | Progress: (16/20) | 14.52 s
    [Task  4/25]  Current/Best:   13.53/  22.03 GFLOPS | Progress: (20/20) | 16.48 s Done.
-
    [Task  5/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task  5/25]  Current/Best:    9.57/  10.31 GFLOPS | Progress: (4/20) | 2.56 s
    [Task  5/25]  Current/Best:   11.82/  12.69 GFLOPS | Progress: (8/20) | 4.63 s
    [Task  5/25]  Current/Best:   11.92/  18.08 GFLOPS | Progress: (12/20) | 7.65 s
    [Task  5/25]  Current/Best:   11.79/  22.88 GFLOPS | Progress: (16/20) | 9.06 s
    [Task  5/25]  Current/Best:   12.16/  22.88 GFLOPS | Progress: (20/20) | 10.94 s Done.
-
    [Task  6/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task  6/25]  Current/Best:   12.37/  20.90 GFLOPS | Progress: (4/20) | 4.07 s
    [Task  6/25]  Current/Best:   19.10/  20.90 GFLOPS | Progress: (8/20) | 5.84 s
    [Task  6/25]  Current/Best:   13.43/  20.90 GFLOPS | Progress: (12/20) | 7.79 s
    [Task  6/25]  Current/Best:   20.06/  20.90 GFLOPS | Progress: (16/20) | 10.06 s
    [Task  6/25]  Current/Best:    3.70/  20.90 GFLOPS | Progress: (20/20) | 12.62 s Done.
-
    [Task  7/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task  7/25]  Current/Best:   11.22/  12.88 GFLOPS | Progress: (4/20) | 3.63 s
    [Task  7/25]  Current/Best:   20.29/  21.14 GFLOPS | Progress: (8/20) | 5.15 s
    [Task  7/25]  Current/Best:   16.19/  21.14 GFLOPS | Progress: (12/20) | 7.06 s
    [Task  7/25]  Current/Best:   12.31/  21.14 GFLOPS | Progress: (16/20) | 9.11 s
    [Task  7/25]  Current/Best:    6.31/  21.78 GFLOPS | Progress: (20/20) | 11.57 s Done.
-
    [Task  8/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task  8/25]  Current/Best:    9.85/  13.90 GFLOPS | Progress: (4/20) | 2.92 s
    [Task  8/25]  Current/Best:    9.51/  13.90 GFLOPS | Progress: (8/20) | 8.12 s
    [Task  8/25]  Current/Best:   12.51/  13.90 GFLOPS | Progress: (12/20) | 14.70 s
    [Task  8/25]  Current/Best:   18.79/  18.79 GFLOPS | Progress: (16/20) | 16.83 s
    [Task  8/25]  Current/Best:   20.14/  20.14 GFLOPS | Progress: (20/20) | 23.99 s Done.
-
    [Task  9/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task  9/25]  Current/Best:   14.31/  15.82 GFLOPS | Progress: (4/20) | 11.95 s
    [Task  9/25]  Current/Best:   23.40/  23.40 GFLOPS | Progress: (8/20) | 13.74 s
    [Task  9/25]  Current/Best:    8.28/  23.40 GFLOPS | Progress: (12/20) | 16.33 s
    [Task  9/25]  Current/Best:   17.86/  23.40 GFLOPS | Progress: (16/20) | 19.23 s
    [Task  9/25]  Current/Best:    9.06/  23.40 GFLOPS | Progress: (20/20) | 27.88 s
    [Task 10/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task 10/25]  Current/Best:   18.35/  18.35 GFLOPS | Progress: (4/20) | 2.60 s
    [Task 10/25]  Current/Best:   15.50/  18.35 GFLOPS | Progress: (8/20) | 4.27 s
    [Task 10/25]  Current/Best:   12.64/  18.66 GFLOPS | Progress: (12/20) | 5.82 s
    [Task 10/25]  Current/Best:   19.21/  20.29 GFLOPS | Progress: (16/20) | 6.93 s
    [Task 10/25]  Current/Best:    8.92/  20.29 GFLOPS | Progress: (20/20
 ) | 8.48 s Done.
-
    [Task 11/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task 11/25]  Current/Best:   12.17/  18.13 GFLOPS | Progress: (4/20) | 3.35 s
    [Task 11/25]  Current/Best:   16.97/  18.13 GFLOPS | Progress: (8/20) | 6.17 s
    [Task 11/25]  Current/Best:   18.14/  18.14 GFLOPS | Progress: (12/20) | 8.25 s
    [Task 11/25]  Current/Best:   13.41/  21.22 GFLOPS | Progress: (16/20) | 11.13 s
    [Task 11/25]  Current/Best:   19.46/  21.66 GFLOPS | Progress: (20/20) | 13.21 s Done.
-
    [Task 12/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task 12/25]  Current/Best:    7.84/  18.07 GFLOPS | Progress: (4/20) | 5.69 s
    [Task 12/25]  Current/Best:    5.20/  18.07 GFLOPS | Progress: (8/20) | 9.67 s
    [Task 12/25]  Current/Best:   18.90/  18.90 GFLOPS | Progress: (12/20) | 11.66 s
    [Task 12/25]  Current/Best:   15.50/  18.90 GFLOPS | Progress: (16/20) | 14.59 s
    [Task 12/25]  Current/Best:   15.10/  18.90 GFLOPS | Progress: (20/20) | 16.51 s Done.
-
    [Task 13/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task 13/25]  Current/Best:    8.69/  17.36 GFLOPS | Progress: (4/20) | 3.76 s
    [Task 13/25]  Current/Best:   15.69/  20.98 GFLOPS | Progress: (8/20) | 6.37 s
    [Task 13/25]  Current/Best:   19.59/  21.83 GFLOPS | Progress: (12/20) | 9.42 s
    [Task 13/25]  Current/Best:   12.24/  21.83 GFLOPS | Progress: (16/20) | 12.88 s
    [Task 13/25]  Current/Best:   18.70/  21.83 GFLOPS | Progress: (20/20) | 15.24 s Done.
-
    [Task 14/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task 14/25]  Current/Best:   12.28/  13.25 GFLOPS | Progress: (4/20) | 3.43 s
    [Task 14/25]  Current/Best:    6.12/  13.34 GFLOPS | Progress: (8/20) | 5.63 s
    [Task 14/25]  Current/Best:   20.32/  20.32 GFLOPS | Progress: (12/20) | 8.35 s
    [Task 14/25]  Current/Best:   16.75/  20.32 GFLOPS | Progress: (16/20) | 10.01 s Done.
-
    [Task 14/25]  Current/Best:   16.98/  20.32 GFLOPS | Progress: (20/20) | 11.75 s
    [Task 15/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task 15/25]  Current/Best:   16.15/  17.62 GFLOPS | Progress: (4/20) | 2.75 s
    [Task 15/25]  Current/Best:   14.51/  17.98 GFLOPS | Progress: (8/20) | 4.10 s
    [Task 15/25]  Current/Best:   10.38/  22.34 GFLOPS | Progress: (12/20) | 6.36 s
    [Task 15/25]  Current/Best:   20.45/  22.34 GFLOPS | Progress: (16/20) | 10.05 s
    [Task 15/25]  Current/Best:    9.70/  22.34 GFLOPS | Progress: (20/20) | 11.08 s
    [Task 16/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task 16/25]  Current/Best:   20.88/  20.88 GFLOPS | Progress: (4/20) | 2.96 s
    [Task 16/25]  Current/Best:    3.04/  20.88 GFLOPS | Progress: (8/20) | 4.57 s
    [Task 16/25]  Current/Best:   19.81/  20.88 GFLOPS | Progress: (12/20) | 5.77 s
    [Task 16/25]  Current/Best:   18.17/  20.88 GFLOPS | Progress: (16/20) 
 | 7.13 s
    [Task 16/25]  Current/Best:   10.01/  22.44 GFLOPS | Progress: (20/20) | 9.30 s Done.
-
    [Task 17/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task 17/25]  Current/Best:   13.27/  18.76 GFLOPS | Progress: (4/20) | 4.83 s
    [Task 17/25]  Current/Best:   14.36/  23.28 GFLOPS | Progress: (8/20) | 7.73 s
    [Task 17/25]  Current/Best:   17.05/  23.28 GFLOPS | Progress: (12/20) | 9.79 s
    [Task 17/25]  Current/Best:   16.52/  23.28 GFLOPS | Progress: (16/20) | 12.06 s
    [Task 17/25]  Current/Best:   10.05/  23.28 GFLOPS | Progress: (20/20) | 14.22 s Done.
-
    [Task 18/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task 18/25]  Current/Best:   11.44/  17.91 GFLOPS | Progress: (4/20) | 3.82 s
    [Task 18/25]  Current/Best:   10.60/  17.91 GFLOPS | Progress: (8/20) | 7.51 s
    [Task 18/25]  Current/Best:   19.24/  19.24 GFLOPS | Progress: (12/20) | 9.45 s
    [Task 18/25]  Current/Best:   10.00/  19.24 GFLOPS | Progress: (16/20) | 13.36 s
    [Task 18/25]  Current/Best:   20.57/  20.57 GFLOPS | Progress: (20/20) | 14.87 s Done.
-
    [Task 19/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task 19/25]  Current/Best:    7.17/  20.23 GFLOPS | Progress: (4/20) | 6.07 s
    [Task 19/25]  Current/Best:    2.61/  20.23 GFLOPS | Progress: (8/20) | 9.44 s
    [Task 19/25]  Current/Best:   19.93/  21.69 GFLOPS | Progress: (12/20) | 12.42 s
    [Task 19/25]  Current/Best:   14.23/  21.69 GFLOPS | Progress: (16/20) | 15.50 s
    [Task 19/25]  Current/Best:    2.70/  23.77 GFLOPS | Progress: (20/20) | 18.32 s Done.
-
    [Task 20/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task 20/25]  Current/Best:    8.76/  14.91 GFLOPS | Progress: (4/20) | 3.35 s Done.
+
    [Task  1/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task  1/25]  Current/Best:   17.43/  17.43 GFLOPS | Progress: (4/20) | 6.30 s
    [Task  1/25]  Current/Best:    6.16/  17.43 GFLOPS | Progress: (8/20) | 9.31 s
    [Task  1/25]  Current/Best:   11.52/  22.68 GFLOPS | Progress: (12/20) | 11.81 s
    [Task  1/25]  Current/Best:   16.70/  22.68 GFLOPS | Progress: (16/20) | 13.51 s
    [Task  1/25]  Current/Best:   11.60/  23.88 GFLOPS | Progress: (20/20) | 15.26 s Done.
+
    [Task  2/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task  2/25]  Current/Best:   12.27/  13.24 GFLOPS | Progress: (4/20) | 3.92 s
    [Task  2/25]  Current/Best:   14.08/  17.82 GFLOPS | Progress: (8/20) | 5.23 s
    [Task  2/25]  Current/Best:   20.62/  20.62 GFLOPS | Progress: (12/20) | 6.58 s
    [Task  2/25]  Current/Best:   12.98/  20.62 GFLOPS | Progress: (16/20) | 7.84 s
    [Task  2/25]  Current/Best:   20.35/  20.62 GFLOPS | Progress: (20/20) | 9.44 s Done.
+
    [Task  3/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task  3/25]  Current/Best:    1.63/  10.55 GFLOPS | Progress: (4/20) | 5.87 s
    [Task  3/25]  Current/Best:   15.55/  16.59 GFLOPS | Progress: (8/20) | 7.80 s
    [Task  3/25]  Current/Best:   14.91/  16.59 GFLOPS | Progress: (12/20) | 9.51 s
    [Task  3/25]  Current/Best:    7.16/  23.81 GFLOPS | Progress: (16/20) | 11.43 s
    [Task  3/25]  Current/Best:   12.52/  23.81 GFLOPS | Progress: (20/20) | 16.04 s Done.
+
    [Task  4/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task  4/25]  Current/Best:    9.53/  20.15 GFLOPS | Progress: (4/20) | 2.41 s
    [Task  4/25]  Current/Best:    6.72/  20.15 GFLOPS | Progress: (8/20) | 7.21 s
    [Task  4/25]  Current/Best:   22.22/  22.22 GFLOPS | Progress: (12/20) | 12.11 s
    [Task  4/25]  Current/Best:   16.78/  22.22 GFLOPS | Progress: (16/20) | 14.57 s
    [Task  4/25]  Current/Best:   13.35/  22.22 GFLOPS | Progress: (20/20) | 16.55 s Done.
+
    [Task  5/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task  5/25]  Current/Best:    9.89/  10.50 GFLOPS | Progress: (4/20) | 2.58 s
    [Task  5/25]  Current/Best:   11.86/  12.95 GFLOPS | Progress: (8/20) | 4.63 s
    [Task  5/25]  Current/Best:   10.28/  18.03 GFLOPS | Progress: (12/20) | 7.70 s
    [Task  5/25]  Current/Best:   11.89/  22.41 GFLOPS | Progress: (16/20) | 9.12 s
    [Task  5/25]  Current/Best:   11.98/  22.41 GFLOPS | Progress: (20/20) | 11.01 s Done.
+
    [Task  6/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task  6/25]  Current/Best:   12.26/  20.62 GFLOPS | Progress: (4/20) | 4.16 s
    [Task  6/25]  Current/Best:   18.90/  20.62 GFLOPS | Progress: (8/20) | 5.94 s
    [Task  6/25]  Current/Best:   13.26/  20.62 GFLOPS | Progress: (12/20) | 7.89 s
    [Task  6/25]  Current/Best:   20.02/  20.62 GFLOPS | Progress: (16/20) | 10.14 s
    [Task  6/25]  Current/Best:    3.73/  20.62 GFLOPS | Progress: (20/20) | 12.69 s Done.
+
    [Task  7/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task  7/25]  Current/Best:   11.26/  12.98 GFLOPS | Progress: (4/20) | 3.64 s
    [Task  7/25]  Current/Best:   20.24/  21.21 GFLOPS | Progress: (8/20) | 5.15 s
    [Task  7/25]  Current/Best:   15.63/  21.21 GFLOPS | Progress: (12/20) | 7.08 s
    [Task  7/25]  Current/Best:   12.29/  21.21 GFLOPS | Progress: (16/20) | 9.12 s
    [Task  7/25]  Current/Best:    6.30/  21.72 GFLOPS | Progress: (20/20) | 11.60 s Done.
+
    [Task  8/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task  8/25]  Current/Best:   10.40/  14.04 GFLOPS | Progress: (4/20) | 2.91 s
    [Task  8/25]  Current/Best:    9.71/  14.04 GFLOPS | Progress: (8/20) | 8.02 s
    [Task  8/25]  Current/Best:   12.77/  14.04 GFLOPS | Progress: (12/20) | 14.58 s
    [Task  8/25]  Current/Best:   18.96/  18.96 GFLOPS | Progress: (16/20) | 16.66 s
    [Task  8/25]  Current/Best:   20.35/  20.35 GFLOPS | Progress: (20/20) | 23.84 s Done.
+
    [Task  9/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task  9/25]  Current/Best:   14.35/  15.54 GFLOPS | Progress: (4/20) | 11.97 s
    [Task  9/25]  Current/Best:   23.31/  23.31 GFLOPS | Progress: (8/20) | 13.74 s
    [Task  9/25]  Current/Best:    8.16/  23.31 GFLOPS | Progress: (12/20) | 16.27 s
    [Task  9/25]  Current/Best:   17.96/  23.31 GFLOPS | Progress: (16/20) | 19.17 s
    [Task  9/25]  Current/Best:    9.15/  23.31 GFLOPS | Progress: (20/20) | 27.84 s
    [Task 10/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task 10/25]  Current/Best:   18.21/  18.21 GFLOPS | Progress: (4/20) | 2.58 s
    [Task 10/25]  Current/Best:   15.53/  18.21 GFLOPS | Progress: (8/20) | 4.22 s
    [Task 10/25]  Current/Best:   12.68/  19.32 GFLOPS | Progress: (12/20) | 5.77 s
    [Task 10/25]  Current/Best:   15.92/  20.35 GFLOPS | Progress: (16/20) | 6.90 s
    [Task 10/25]  Current/Best:    8.92/  20.35 GFLOPS | Progress: (20/20
 ) | 8.43 s Done.
+
    [Task 11/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task 11/25]  Current/Best:   11.34/  18.14 GFLOPS | Progress: (4/20) | 3.42 s
    [Task 11/25]  Current/Best:   16.90/  18.14 GFLOPS | Progress: (8/20) | 6.28 s
    [Task 11/25]  Current/Best:   18.17/  18.17 GFLOPS | Progress: (12/20) | 8.37 s
    [Task 11/25]  Current/Best:   13.41/  21.12 GFLOPS | Progress: (16/20) | 11.34 s
    [Task 11/25]  Current/Best:   19.49/  21.50 GFLOPS | Progress: (20/20) | 13.47 s Done.
+
    [Task 12/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task 12/25]  Current/Best:    7.79/  18.26 GFLOPS | Progress: (4/20) | 5.76 s
    [Task 12/25]  Current/Best:    5.31/  18.26 GFLOPS | Progress: (8/20) | 9.73 s
    [Task 12/25]  Current/Best:   19.10/  19.10 GFLOPS | Progress: (12/20) | 11.71 s
    [Task 12/25]  Current/Best:   15.13/  19.10 GFLOPS | Progress: (16/20) | 14.68 s
    [Task 12/25]  Current/Best:   15.16/  19.23 GFLOPS | Progress: (20/20) | 16.60 s Done.
+
    [Task 13/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task 13/25]  Current/Best:    8.79/  17.34 GFLOPS | Progress: (4/20) | 3.76 s
    [Task 13/25]  Current/Best:   15.60/  20.94 GFLOPS | Progress: (8/20) | 6.39 s
    [Task 13/25]  Current/Best:   19.52/  21.72 GFLOPS | Progress: (12/20) | 9.40 s
    [Task 13/25]  Current/Best:   12.25/  21.72 GFLOPS | Progress: (16/20) | 12.88 s
    [Task 13/25]  Current/Best:   18.79/  21.72 GFLOPS | Progress: (20/20) | 15.19 s Done.
+
    [Task 14/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task 14/25]  Current/Best:   13.44/  13.44 GFLOPS | Progress: (4/20) | 3.45 s
    [Task 14/25]  Current/Best:    6.09/  13.44 GFLOPS | Progress: (8/20) | 5.66 s
    [Task 14/25]  Current/Best:   20.30/  20.30 GFLOPS | Progress: (12/20) | 8.35 s
    [Task 14/25]  Current/Best:   16.87/  20.30 GFLOPS | Progress: (16/20) | 10.02 s Done.
+
    [Task 14/25]  Current/Best:   17.43/  20.30 GFLOPS | Progress: (20/20) | 11.78 s
    [Task 15/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task 15/25]  Current/Best:   16.11/  17.63 GFLOPS | Progress: (4/20) | 2.73 s
    [Task 15/25]  Current/Best:   14.21/  17.97 GFLOPS | Progress: (8/20) | 4.07 s
    [Task 15/25]  Current/Best:   10.38/  22.32 GFLOPS | Progress: (12/20) | 6.33 s
    [Task 15/25]  Current/Best:   20.38/  22.32 GFLOPS | Progress: (16/20) | 9.38 s
    [Task 15/25]  Current/Best:    9.60/  22.32 GFLOPS | Progress: (20/20) | 10.40 s
    [Task 16/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task 16/25]  Current/Best:   20.71/  20.71 GFLOPS | Progress: (4/20) | 2.96 s
    [Task 16/25]  Current/Best:    3.01/  20.71 GFLOPS | Progress: (8/20) | 4.58 s
    [Task 16/25]  Current/Best:   18.92/  20.71 GFLOPS | Progress: (12/20) | 5.79 s
    [Task 16/25]  Current/Best:   17.89/  20.71 GFLOPS | Progress: (16/20) |
  7.15 s
    [Task 16/25]  Current/Best:   10.00/  20.71 GFLOPS | Progress: (20/20) | 9.34 s Done.
+
    [Task 17/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task 17/25]  Current/Best:   13.41/  18.76 GFLOPS | Progress: (4/20) | 4.81 s
    [Task 17/25]  Current/Best:   14.48/  23.18 GFLOPS | Progress: (8/20) | 7.74 s
    [Task 17/25]  Current/Best:   16.92/  23.18 GFLOPS | Progress: (12/20) | 9.81 s
    [Task 17/25]  Current/Best:   16.46/  23.18 GFLOPS | Progress: (16/20) | 12.04 s
    [Task 17/25]  Current/Best:   10.03/  23.18 GFLOPS | Progress: (20/20) | 14.22 s Done.
+
    [Task 18/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task 18/25]  Current/Best:   11.36/  18.13 GFLOPS | Progress: (4/20) | 3.84 s
    [Task 18/25]  Current/Best:   10.62/  19.85 GFLOPS | Progress: (8/20) | 7.55 s
    [Task 18/25]  Current/Best:   18.84/  19.85 GFLOPS | Progress: (12/20) | 9.52 s
    [Task 18/25]  Current/Best:    9.97/  19.85 GFLOPS | Progress: (16/20) | 13.39 s
    [Task 18/25]  Current/Best:   20.65/  20.65 GFLOPS | Progress: (20/20) | 14.91 s Done.
+
    [Task 19/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task 19/25]  Current/Best:    7.10/  20.41 GFLOPS | Progress: (4/20) | 6.14 s
    [Task 19/25]  Current/Best:    2.60/  20.41 GFLOPS | Progress: (8/20) | 9.50 s
    [Task 19/25]  Current/Best:   19.31/  20.72 GFLOPS | Progress: (12/20) | 12.48 s
    [Task 19/25]  Current/Best:   15.33/  20.79 GFLOPS | Progress: (16/20) | 15.49 s
    [Task 19/25]  Current/Best:    2.69/  23.31 GFLOPS | Progress: (20/20) | 18.29 s Done.
+
    [Task 20/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task 20/25]  Current/Best:    8.38/  15.20 GFLOPS | Progress: (4/20) | 3.34 s Done.
      Done.
-
    [Task 20/25]  Current/Best:    9.89/  14.91 GFLOPS | Progress: (8/20) | 6.75 s
    [Task 20/25]  Current/Best:    2.32/  16.42 GFLOPS | Progress: (12/20) | 10.63 s
    [Task 20/25]  Current/Best:   12.35/  16.42 GFLOPS | Progress: (16/20) | 14.35 s
    [Task 20/25]  Current/Best:   12.93/  22.30 GFLOPS | Progress: (20/20) | 16.46 s
    [Task 21/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task 21/25]  Current/Best:    6.40/  17.73 GFLOPS | Progress: (4/20) | 3.28 s
    [Task 21/25]  Current/Best:   14.62/  17.73 GFLOPS | Progress: (8/20) | 4.89 s
    [Task 21/25]  Current/Best:    1.61/  17.73 GFLOPS | Progress: (12/20) | 7.04 s
    [Task 21/25]  Current/Best:   18.05/  18.05 GFLOPS | Progress: (16/20) | 10.57 s
    [Task 21/25]  Current/Best:    4.48/  18.05 GFLOPS | Progress: (20/20) | 17.87 s
    [Task 22/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task 22/25]  Current/Best:    2.70/  17.04 GFLOPS | Progress: (4/20
 ) | 2.67 s
    [Task 22/25]  Current/Best:    8.62/  22.30 GFLOPS | Progress: (8/20) | 4.71 s
    [Task 22/25]  Current/Best:   20.34/  22.30 GFLOPS | Progress: (12/20) | 7.11 s
    [Task 22/25]  Current/Best:   15.56/  22.30 GFLOPS | Progress: (16/20) | 9.21 s
    [Task 22/25]  Current/Best:   14.30/  22.30 GFLOPS | Progress: (20/20) | 10.93 s Done.
-
    [Task 23/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task 23/25]  Current/Best:   17.82/  21.13 GFLOPS | Progress: (4/20) | 3.20 s
    [Task 23/25]  Current/Best:   14.25/  21.13 GFLOPS | Progress: (8/20) | 6.57 s
    [Task 23/25]  Current/Best:   21.22/  22.12 GFLOPS | Progress: (12/20) | 8.40 s
    [Task 23/25]  Current/Best:    6.52/  22.12 GFLOPS | Progress: (16/20) | 15.48 s
    [Task 23/25]  Current/Best:    7.81/  22.12 GFLOPS | Progress: (20/20) | 19.66 s Done.
-
    [Task 24/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task 24/25]  Current/Best:    8.29/   8.29 GFLOPS | Progress: (4/20) | 11.78 s
    [Task 24/25]  Current/Best:    3.57/   8.29 GFLOPS | Progress: (8/20) | 23.00 s
    [Task 24/25]  Current/Best:    4.38/   8.29 GFLOPS | Progress: (12/20) | 33.71 s Done.
-     Done.
-
    [Task 24/25]  Current/Best:    5.35/   8.62 GFLOPS | Progress: (16/20) | 39.46 s
    [Task 24/25]  Current/Best:    3.39/   8.62 GFLOPS | Progress: (20/20) | 45.43 s Done.
-
    [Task 25/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task 25/25]  Current/Best:    1.55/   2.79 GFLOPS | Progress: (4/20) | 11.59 s
    [Task 25/25]  Current/Best:    5.67/   8.01 GFLOPS | Progress: (8/20) | 22.83 s
    [Task 25/25]  Current/Best:    5.92/   8.01 GFLOPS | Progress: (12/20) | 34.34 s
    [Task 25/25]  Current/Best:    5.80/   8.17 GFLOPS | Progress: (16/20) | 36.08 s
    [Task 25/25]  Current/Best:    2.83/   9.32 GFLOPS | Progress: (20/20) | 46.75 s
+
    [Task 20/25]  Current/Best:    9.80/  15.20 GFLOPS | Progress: (8/20) | 6.87 s
    [Task 20/25]  Current/Best:    2.32/  16.49 GFLOPS | Progress: (12/20) | 10.84 s
    [Task 20/25]  Current/Best:   12.41/  16.49 GFLOPS | Progress: (16/20) | 14.75 s
    [Task 20/25]  Current/Best:   12.65/  22.05 GFLOPS | Progress: (20/20) | 16.85 s
    [Task 21/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task 21/25]  Current/Best:    6.40/  17.66 GFLOPS | Progress: (4/20) | 3.29 s
    [Task 21/25]  Current/Best:   14.42/  17.66 GFLOPS | Progress: (8/20) | 4.90 s
    [Task 21/25]  Current/Best:    1.61/  17.66 GFLOPS | Progress: (12/20) | 7.06 s
    [Task 21/25]  Current/Best:   17.12/  17.66 GFLOPS | Progress: (16/20) | 10.62 s
    [Task 21/25]  Current/Best:    4.47/  17.66 GFLOPS | Progress: (20/20) | 18.01 s
    [Task 22/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task 22/25]  Current/Best:    2.70/  16.93 GFLOPS | Progress: (4/20
 ) | 2.69 s
    [Task 22/25]  Current/Best:    8.81/  21.94 GFLOPS | Progress: (8/20) | 4.72 s
    [Task 22/25]  Current/Best:   20.08/  21.94 GFLOPS | Progress: (12/20) | 7.10 s
    [Task 22/25]  Current/Best:   15.26/  21.94 GFLOPS | Progress: (16/20) | 9.21 s
    [Task 22/25]  Current/Best:   15.28/  21.94 GFLOPS | Progress: (20/20) | 10.97 s Done.
+
    [Task 23/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task 23/25]  Current/Best:   17.43/  20.22 GFLOPS | Progress: (4/20) | 3.27 s
    [Task 23/25]  Current/Best:   14.73/  20.22 GFLOPS | Progress: (8/20) | 6.73 s
    [Task 23/25]  Current/Best:   20.96/  21.30 GFLOPS | Progress: (12/20) | 8.60 s
    [Task 23/25]  Current/Best:    6.36/  21.30 GFLOPS | Progress: (16/20) | 15.57 s
    [Task 23/25]  Current/Best:    7.79/  21.30 GFLOPS | Progress: (20/20) | 19.81 s Done.
+
    [Task 24/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task 24/25]  Current/Best:    8.43/   8.43 GFLOPS | Progress: (4/20) | 11.79 s
    [Task 24/25]  Current/Best:    2.03/   8.43 GFLOPS | Progress: (8/20) | 22.79 s
    [Task 24/25]  Current/Best:    4.63/   8.43 GFLOPS | Progress: (12/20) | 34.36 s Done.
+
    [Task 24/25]  Current/Best:    7.06/   8.83 GFLOPS | Progress: (16/20) | 40.07 s
    [Task 24/25]  Current/Best:    3.32/   8.83 GFLOPS | Progress: (20/20) | 46.07 s Done.
+
    [Task 25/25]  Current/Best:    0.00/   0.00 GFLOPS | Progress: (0/20) | 0.00 s
    [Task 25/25]  Current/Best:    1.55/   2.81 GFLOPS | Progress: (4/20) | 11.61 s
    [Task 25/25]  Current/Best:    5.81/   8.25 GFLOPS | Progress: (8/20) | 22.88 s
    [Task 25/25]  Current/Best:    5.88/   8.25 GFLOPS | Progress: (12/20) | 34.36 s
    [Task 25/25]  Current/Best:    5.70/   9.26 GFLOPS | Progress: (16/20) | 36.24 s
    [Task 25/25]  Current/Best:    2.89/   9.26 GFLOPS | Progress: (20/20) | 46.93 s
 
 
 
@@ -655,6 +654,7 @@ model using optimized operators to speed up our computations.
 
  .. code-block:: none
 
+     Done.
     /workspace/python/tvm/driver/build_module.py:268: UserWarning: target_host parameter is going to be deprecated. Please pass in tvm.target.Target(target, host=target_host) instead.
       "target_host parameter is going to be deprecated. "
 
@@ -748,8 +748,8 @@ improvement in comparing the optimized model to the unoptimized model.
 
  .. code-block:: none
 
-    optimized: {'mean': 410.19464660001177, 'median': 409.9865518500337, 'std': 1.479414808980326}
-    unoptimized: {'mean': 493.9443639300134, 'median': 494.06449910002266, 'std': 1.6148426721759408}
+    optimized: {'mean': 408.73262078999915, 'median': 408.7684094499991, 'std': 0.6236709452687695}
+    unoptimized: {'mean': 495.75926721000326, 'median': 495.88527695000266, 'std': 1.0175382126682093}
 
 
 
@@ -772,7 +772,7 @@ profiling/benchmarking.
 
 .. rst-class:: sphx-glr-timing
 
-   **Total running time of the script:** ( 10 minutes  25.169 seconds)
+   **Total running time of the script:** ( 10 minutes  26.266 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 aecaa266e..2bd7e56d2 100644
--- a/docs/_sources/tutorial/cross_compilation_and_rpc.rst.txt
+++ b/docs/_sources/tutorial/cross_compilation_and_rpc.rst.txt
@@ -282,7 +282,7 @@ device and returns the measured cost. Network overhead is excluded.
 
  .. code-block:: none
 
-    1.284e-07 secs/op
+    1.239e-07 secs/op
 
 
 
diff --git a/docs/_sources/tutorial/intro_topi.rst.txt b/docs/_sources/tutorial/intro_topi.rst.txt
index 9bea6e0ab..813c0afe1 100644
--- a/docs/_sources/tutorial/intro_topi.rst.txt
+++ b/docs/_sources/tutorial/intro_topi.rst.txt
@@ -263,7 +263,7 @@ As you can see, scheduled stages of computation have been accumulated and we can
 
  .. code-block:: none
 
-    [stage(a, placeholder(a, 0x13e630d0)), stage(b, placeholder(b, 0xd1d8140)), stage(T_add, compute(T_add, body=[(a[ax0, ax1, ax2] + b[ax1, ax2])], axis=[iter_var(ax0, range(min=0, ext=100)), iter_var(ax1, range(min=0, ext=10)), iter_var(ax2, range(min=0, ext=10))], reduce_axis=[], tag=broadcast, attrs={})), stage(T_multiply, compute(T_multiply, body=[(a[ax0, ax1, ax2]*b[ax1, ax2])], axis=[iter_var(ax0, range(min=0, ext=100)), iter_var(ax1, range(min=0, ext=10)), iter_var(ax2, range(min [...]
+    [stage(a, placeholder(a, 0x202d84c0)), stage(b, placeholder(b, 0xace4620)), 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 d30f8d0c9..6e043dc89 100644
--- a/docs/_sources/tutorial/sg_execution_times.rst.txt
+++ b/docs/_sources/tutorial/sg_execution_times.rst.txt
@@ -5,30 +5,30 @@
 
 Computation times
 =================
-**13:17.160** total execution time for **tutorial** files:
+**13:13.704** total execution time for **tutorial** files:
 
 +------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_autotvm_relay_x86.py` (``autotvm_relay_x86.py``)                 | 10:25.169 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_autotvm_relay_x86.py` (``autotvm_relay_x86.py``)                 | 10:26.266 | 0.0 MB |
 +------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_tensor_expr_get_started.py` (``tensor_expr_get_started.py``)     | 01:01.661 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_tensor_expr_get_started.py` (``tensor_expr_get_started.py``)     | 00:58.703 | 0.0 MB |
 +------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_auto_scheduler_matmul_x86.py` (``auto_scheduler_matmul_x86.py``) | 00:53.974 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_auto_scheduler_matmul_x86.py` (``auto_scheduler_matmul_x86.py``) | 00:51.926 | 0.0 MB |
 +------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_relay_quick_start.py` (``relay_quick_start.py``)                 | 00:30.451 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_relay_quick_start.py` (``relay_quick_start.py``)                 | 00:30.415 | 0.0 MB |
 +------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_autotvm_matmul_x86.py` (``autotvm_matmul_x86.py``)               | 00:24.524 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_autotvm_matmul_x86.py` (``autotvm_matmul_x86.py``)               | 00:24.318 | 0.0 MB |
 +------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_intro_topi.py` (``intro_topi.py``)                               | 00:00.710 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_tensor_ir_blitz_course.py` (``tensor_ir_blitz_course.py``)       | 00:01.196 | 0.0 MB |
 +------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_tensor_ir_blitz_course.py` (``tensor_ir_blitz_course.py``)       | 00:00.512 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_intro_topi.py` (``intro_topi.py``)                               | 00:00.718 | 0.0 MB |
 +------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_cross_compilation_and_rpc.py` (``cross_compilation_and_rpc.py``) | 00:00.152 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_cross_compilation_and_rpc.py` (``cross_compilation_and_rpc.py``) | 00:00.155 | 0.0 MB |
 +------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_introduction.py` (``introduction.py``)                           | 00:00.004 | 0.0 MB |
-+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_tvmc_command_line_driver.py` (``tvmc_command_line_driver.py``)   | 00:00.001 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_introduction.py` (``introduction.py``)                           | 00:00.005 | 0.0 MB |
 +------------------------------------------------------------------------------------------+-----------+--------+
 | :ref:`sphx_glr_tutorial_tvmc_python.py` (``tvmc_python.py``)                             | 00:00.001 | 0.0 MB |
 +------------------------------------------------------------------------------------------+-----------+--------+
 | :ref:`sphx_glr_tutorial_install.py` (``install.py``)                                     | 00:00.001 | 0.0 MB |
 +------------------------------------------------------------------------------------------+-----------+--------+
+| :ref:`sphx_glr_tutorial_tvmc_command_line_driver.py` (``tvmc_command_line_driver.py``)   | 00:00.001 | 0.0 MB |
++------------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/tutorial/tensor_expr_get_started.rst.txt b/docs/_sources/tutorial/tensor_expr_get_started.rst.txt
index f23b9944f..f7117ef67 100644
--- a/docs/_sources/tutorial/tensor_expr_get_started.rst.txt
+++ b/docs/_sources/tutorial/tensor_expr_get_started.rst.txt
@@ -301,7 +301,7 @@ helper function to run a profile of the TVM generated code.
 
  .. code-block:: none
 
-    Numpy running time: 0.000009
+    Numpy running time: 0.000008
     naive: 0.000008
 
 
@@ -403,7 +403,7 @@ compile and run this new schedule with the parallel operation applied:
 
     /workspace/python/tvm/driver/build_module.py:268: UserWarning: target_host parameter is going to be deprecated. Please pass in tvm.target.Target(target, host=target_host) instead.
       "target_host parameter is going to be deprecated. "
-    parallel: 0.000006
+    parallel: 0.000007
 
 
 
@@ -512,10 +512,10 @@ We can now compare the different schedules
  .. code-block:: none
 
                 Operator                  Timing             Performance
-                   numpy    8.705270001883036e-06                    1.0
-                   naive              7.9467e-06      0.9128608300812093
-                parallel    6.057599999999999e-06      0.695854350145335
-                  vector    2.4561400000000002e-05     2.821440345295106
+                   numpy    7.822519999081123e-06                    1.0
+                   naive    7.5761999999999996e-06    0.9685114260992548
+                parallel    7.075399999999999e-06     0.9044911359550523
+                  vector             2.46329e-05       3.148972454259435
 
 
 
@@ -936,7 +936,7 @@ matrix multiplication.
 
  .. code-block:: none
 
-    Numpy running time: 0.018048
+    Numpy running time: 0.018528
 
 
 
@@ -996,7 +996,7 @@ optimizations.
 
     /workspace/python/tvm/driver/build_module.py:268: UserWarning: target_host parameter is going to be deprecated. Please pass in tvm.target.Target(target, host=target_host) instead.
       "target_host parameter is going to be deprecated. "
-    none: 3.460671
+    none: 3.230218
 
 
 
@@ -1101,7 +1101,7 @@ schedule.
 
     /workspace/python/tvm/driver/build_module.py:268: UserWarning: target_host parameter is going to be deprecated. Please pass in tvm.target.Target(target, host=target_host) instead.
       "target_host parameter is going to be deprecated. "
-    blocking: 0.304582
+    blocking: 0.306433
 
 
 
@@ -1199,7 +1199,7 @@ already cache friendly from our previous optimizations.
 
     /workspace/python/tvm/driver/build_module.py:268: UserWarning: target_host parameter is going to be deprecated. Please pass in tvm.target.Target(target, host=target_host) instead.
       "target_host parameter is going to be deprecated. "
-    vectorization: 0.341599
+    vectorization: 0.340673
     @main = primfn(A_1: handle, B_1: handle, C_1: handle) -> ()
       attr = {"from_legacy_te_schedule": True, "global_symbol": "main", "tir.noalias": True}
       buffers = {A: Buffer(A_2: Pointer(float32), float32, [1048576], []),
@@ -1275,7 +1275,7 @@ more cache friendly.
 
     /workspace/python/tvm/driver/build_module.py:268: UserWarning: target_host parameter is going to be deprecated. Please pass in tvm.target.Target(target, host=target_host) instead.
       "target_host parameter is going to be deprecated. "
-    loop permutation: 0.125728
+    loop permutation: 0.120267
     @main = primfn(A_1: handle, B_1: handle, C_1: handle) -> ()
       attr = {"from_legacy_te_schedule": True, "global_symbol": "main", "tir.noalias": True}
       buffers = {A: Buffer(A_2: Pointer(float32), float32, [1048576], []),
@@ -1376,7 +1376,7 @@ optimized schedule.
 
     /workspace/python/tvm/driver/build_module.py:268: UserWarning: target_host parameter is going to be deprecated. Please pass in tvm.target.Target(target, host=target_host) instead.
       "target_host parameter is going to be deprecated. "
-    array packing: 0.109040
+    array packing: 0.110449
     @main = primfn(A_1: handle, B_1: handle, C_1: handle) -> ()
       attr = {"from_legacy_te_schedule": True, "global_symbol": "main", "tir.noalias": True}
       buffers = {A: Buffer(A_2: Pointer(float32), float32, [1048576], []),
@@ -1471,7 +1471,7 @@ to `C` when all the block results are ready.
 
     /workspace/python/tvm/driver/build_module.py:268: UserWarning: target_host parameter is going to be deprecated. Please pass in tvm.target.Target(target, host=target_host) instead.
       "target_host parameter is going to be deprecated. "
-    block caching: 0.110982
+    block caching: 0.110847
     @main = primfn(A_1: handle, B_1: handle, C_1: handle) -> ()
       attr = {"from_legacy_te_schedule": True, "global_symbol": "main", "tir.noalias": True}
       buffers = {A: Buffer(A_2: Pointer(float32), float32, [1048576], []),
@@ -1559,7 +1559,7 @@ of thread-level parallelization.
 
     /workspace/python/tvm/driver/build_module.py:268: UserWarning: target_host parameter is going to be deprecated. Please pass in tvm.target.Target(target, host=target_host) instead.
       "target_host parameter is going to be deprecated. "
-    parallelization: 0.145363
+    parallelization: 0.144869
     @main = primfn(A_1: handle, B_1: handle, C_1: handle) -> ()
       attr = {"from_legacy_te_schedule": True, "global_symbol": "main", "tir.noalias": True}
       buffers = {A: Buffer(A_2: Pointer(float32), float32, [1048576], []),
@@ -1640,13 +1640,13 @@ working, we can compare the results.
  .. code-block:: none
 
                 Operator                  Timing             Performance
-                    none      3.4606707964999996                     1.0
-                blocking            0.3045817251      0.0880123372058513
-           vectorization            0.3415989042     0.09870887012583834
-        loop permutation     0.12572795650000002    0.036330516218750664
-           array packing            0.1090397093     0.03150825828630649
-           block caching     0.11098193730000001     0.03206948705211811
-         parallelization            0.1453633983    0.042004399391879575
+                    none      3.2302176320999996                     1.0
+                blocking            0.3064332693     0.09486458938705761
+           vectorization            0.3406729383     0.10546439190802287
+        loop permutation     0.12026720499999999     0.03723192016688144
+           array packing     0.11044913710000001     0.03419247545503484
+           block caching            0.1108467533    0.034315568151962975
+         parallelization            0.1448693847     0.04484818089665953
 
 
 
@@ -1686,11 +1686,6 @@ operations with tunable parameters that allows you to automatically optimize
 the computation for specific platforms.
 
 
-.. rst-class:: sphx-glr-timing
-
-   **Total running time of the script:** ( 1 minutes  1.661 seconds)
-
-
 .. _sphx_glr_download_tutorial_tensor_expr_get_started.py:
 
 .. only:: html
diff --git a/docs/commit_hash b/docs/commit_hash
index ab5126a16..23cd55bfb 100644
--- a/docs/commit_hash
+++ b/docs/commit_hash
@@ -1 +1 @@
-21d54f988056e7e84fdc6504aff683d5c6431266
+19e5ec65760e0edc3ae1c6e0a05cb9e78a139fd1
diff --git a/docs/how_to/compile_models/from_darknet.html b/docs/how_to/compile_models/from_darknet.html
index df4fa474c..7191c1ae1 100644
--- a/docs/how_to/compile_models/from_darknet.html
+++ b/docs/how_to/compile_models/from_darknet.html
@@ -569,7 +569,7 @@ class:[&#39;truck 0.9266&#39;] left:471 top:83 right:689 bottom:169
 class:[&#39;bicycle 0.9984&#39;] left:111 top:113 right:577 bottom:447
 </pre></div>
 </div>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes  3.722 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes  3.919 seconds)</p>
 <div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-compile-models-from-darknet-py">
 <div class="sphx-glr-download sphx-glr-download-python docutils container">
 <p><a class="reference download internal" download="" href="../../_downloads/7716f96385bd5abb6e822041e285be54/from_darknet.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">from_darknet.py</span></code></a></p>
diff --git a/docs/how_to/compile_models/from_mxnet.html b/docs/how_to/compile_models/from_mxnet.html
index b7dd72cc1..ccadf99b1 100644
--- a/docs/how_to/compile_models/from_mxnet.html
+++ b/docs/how_to/compile_models/from_mxnet.html
@@ -422,7 +422,7 @@ to download the full example code</p>
 <span class="nb">print</span><span class="p">(</span><span class="s2">&quot;x&quot;</span><span class="p">,</span> <a href="https://docs.python.org/3/library/stdtypes.html#tuple" title="builtins.tuple" class="sphx-glr-backref-module-builtins sphx-glr-backref-type-py-class sphx-glr-backref-instance"><span class="n">x</span><span class="o">.</span><span class="n">shape</span></a><span class="p">)</span>
 </pre></div>
 </div>
-<img src="../../_images/sphx_glr_from_mxnet_001.png" srcset="../../_images/sphx_glr_from_mxnet_001.png" alt="from mxnet" class = "sphx-glr-single-img"/><div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Downloading /workspace/.mxnet/models/resnet18_v1-a0666292.zip8263c9fb-cde9-4e2b-85c1-086c075c285c from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/resnet18_v1-a0666292.zip...
+<img src="../../_images/sphx_glr_from_mxnet_001.png" srcset="../../_images/sphx_glr_from_mxnet_001.png" alt="from mxnet" class = "sphx-glr-single-img"/><div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Downloading /workspace/.mxnet/models/resnet18_v1-a0666292.zipd59d0974-3f8a-4c88-80a5-58e86eaa5467 from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/resnet18_v1-a0666292.zip...
 x (1, 3, 224, 224)
 </pre></div>
 </div>
diff --git a/docs/how_to/compile_models/from_oneflow.html b/docs/how_to/compile_models/from_oneflow.html
index 6c91853e3..2df37e273 100644
--- a/docs/how_to/compile_models/from_oneflow.html
+++ b/docs/how_to/compile_models/from_oneflow.html
@@ -427,12 +427,14 @@ python3 -m pip install -f https://release.oneflow.info <span class="nv">oneflow<
 <div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Downloading: &quot;https://oneflow-public.oss-cn-beijing.aliyuncs.com/model_zoo/flowvision/classification/ResNet/resnet18.zip&quot; to /workspace/.oneflow/flowvision_cache/resnet18.zip
 
   0%|          | 0.00/41.5M [00:00&lt;?, ?B/s]
- 19%|#9        | 7.99M/41.5M [00:00&lt;00:00, 55.4MB/s]
- 35%|###4      | 14.3M/41.5M [00:00&lt;00:00, 52.8MB/s]
- 47%|####6     | 19.4M/41.5M [00:00&lt;00:00, 45.0MB/s]
- 58%|#####7    | 24.0M/41.5M [00:00&lt;00:00, 41.0MB/s]
- 82%|########2 | 34.1M/41.5M [00:00&lt;00:00, 51.3MB/s]
-100%|##########| 41.5M/41.5M [00:00&lt;00:00, 54.4MB/s]
+ 19%|#9        | 7.99M/41.5M [00:00&lt;00:00, 44.4MB/s]
+ 35%|###4      | 14.3M/41.5M [00:00&lt;00:00, 48.0MB/s]
+ 46%|####5     | 19.0M/41.5M [00:00&lt;00:00, 41.0MB/s]
+ 55%|#####5    | 22.9M/41.5M [00:00&lt;00:00, 35.5MB/s]
+ 63%|######3   | 26.3M/41.5M [00:00&lt;00:00, 34.0MB/s]
+ 77%|#######7  | 32.0M/41.5M [00:00&lt;00:00, 40.2MB/s]
+ 92%|#########2| 38.3M/41.5M [00:01&lt;00:00, 39.9MB/s]
+100%|##########| 41.5M/41.5M [00:01&lt;00:00, 39.7MB/s]
 </pre></div>
 </div>
 </div>
diff --git a/docs/how_to/compile_models/from_pytorch.html b/docs/how_to/compile_models/from_pytorch.html
index 033d965c3..a3e0dfcf6 100644
--- a/docs/how_to/compile_models/from_pytorch.html
+++ b/docs/how_to/compile_models/from_pytorch.html
@@ -409,8 +409,9 @@ be unstable.</p>
 <div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Downloading: &quot;https://download.pytorch.org/models/resnet18-f37072fd.pth&quot; to /workspace/.cache/torch/hub/checkpoints/resnet18-f37072fd.pth
 
   0%|          | 0.00/44.7M [00:00&lt;?, ?B/s]
- 49%|####9     | 22.0M/44.7M [00:00&lt;00:00, 230MB/s]
-100%|##########| 44.7M/44.7M [00:00&lt;00:00, 262MB/s]
+ 46%|####6     | 20.7M/44.7M [00:00&lt;00:00, 217MB/s]
+ 98%|#########8| 43.9M/44.7M [00:00&lt;00:00, 232MB/s]
+100%|##########| 44.7M/44.7M [00:00&lt;00:00, 230MB/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 f91f8e217..652681e31 100644
--- a/docs/how_to/compile_models/from_tensorflow.html
+++ b/docs/how_to/compile_models/from_tensorflow.html
@@ -631,7 +631,7 @@ banana (score = 0.00022)
 desk (score = 0.00019)
 </pre></div>
 </div>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes  2.917 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes  3.019 seconds)</p>
 <div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-compile-models-from-tensorflow-py">
 <div class="sphx-glr-download sphx-glr-download-python docutils container">
 <p><a class="reference download internal" download="" href="../../_downloads/7f1d3d1b878694c201c614c807cdebc8/from_tensorflow.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">from_tensorflow.py</span></code></a></p>
diff --git a/docs/how_to/compile_models/sg_execution_times.html b/docs/how_to/compile_models/sg_execution_times.html
index ed3d2e766..1a65ba7e0 100644
--- a/docs/how_to/compile_models/sg_execution_times.html
+++ b/docs/how_to/compile_models/sg_execution_times.html
@@ -322,7 +322,7 @@
             
   <div class="section" id="computation-times">
 <span id="sphx-glr-how-to-compile-models-sg-execution-times"></span><h1>Computation times<a class="headerlink" href="#computation-times" title="Permalink to this headline">¶</a></h1>
-<p><strong>05:03.235</strong> total execution time for <strong>how_to_compile_models</strong> files:</p>
+<p><strong>05:01.855</strong> total execution time for <strong>how_to_compile_models</strong> files:</p>
 <table class="docutils align-default">
 <colgroup>
 <col style="width: 81%" />
@@ -331,43 +331,43 @@
 </colgroup>
 <tbody>
 <tr class="row-odd"><td><p><a class="reference internal" href="from_darknet.html#sphx-glr-how-to-compile-models-from-darknet-py"><span class="std std-ref">Compile YOLO-V2 and YOLO-V3 in DarkNet Models</span></a> (<code class="docutils literal notranslate"><span class="pre">from_darknet.py</span></code>)</p></td>
-<td><p>01:03.722</p></td>
+<td><p>01:03.919</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-even"><td><p><a class="reference internal" href="from_tensorflow.html#sphx-glr-how-to-compile-models-from-tensorflow-py"><span class="std std-ref">Compile Tensorflow Models</span></a> (<code class="docutils literal notranslate"><span class="pre">from_tensorflow.py</span></code>)</p></td>
-<td><p>01:02.917</p></td>
+<td><p>01:03.019</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-odd"><td><p><a class="reference internal" href="from_paddle.html#sphx-glr-how-to-compile-models-from-paddle-py"><span class="std std-ref">Compile PaddlePaddle Models</span></a> (<code class="docutils literal notranslate"><span class="pre">from_paddle.py</span></code>)</p></td>
-<td><p>00:40.483</p></td>
+<td><p>00:39.316</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-even"><td><p><a class="reference internal" href="from_oneflow.html#sphx-glr-how-to-compile-models-from-oneflow-py"><span class="std std-ref">Compile OneFlow Models</span></a> (<code class="docutils literal notranslate"><span class="pre">from_oneflow.py</span></code>)</p></td>
-<td><p>00:27.777</p></td>
+<td><p>00:27.537</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-odd"><td><p><a class="reference internal" href="from_mxnet.html#sphx-glr-how-to-compile-models-from-mxnet-py"><span class="std std-ref">Compile MXNet Models</span></a> (<code class="docutils literal notranslate"><span class="pre">from_mxnet.py</span></code>)</p></td>
-<td><p>00:24.542</p></td>
+<td><p>00:24.488</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-even"><td><p><a class="reference internal" href="from_tflite.html#sphx-glr-how-to-compile-models-from-tflite-py"><span class="std std-ref">Compile TFLite Models</span></a> (<code class="docutils literal notranslate"><span class="pre">from_tflite.py</span></code>)</p></td>
-<td><p>00:24.308</p></td>
+<td><p>00:24.201</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-odd"><td><p><a class="reference internal" href="from_coreml.html#sphx-glr-how-to-compile-models-from-coreml-py"><span class="std std-ref">Compile CoreML Models</span></a> (<code class="docutils literal notranslate"><span class="pre">from_coreml.py</span></code>)</p></td>
-<td><p>00:22.364</p></td>
+<td><p>00:22.408</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-even"><td><p><a class="reference internal" href="from_pytorch.html#sphx-glr-how-to-compile-models-from-pytorch-py"><span class="std std-ref">Compile PyTorch Models</span></a> (<code class="docutils literal notranslate"><span class="pre">from_pytorch.py</span></code>)</p></td>
-<td><p>00:20.139</p></td>
+<td><p>00:19.540</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-odd"><td><p><a class="reference internal" href="from_keras.html#sphx-glr-how-to-compile-models-from-keras-py"><span class="std std-ref">Compile Keras Models</span></a> (<code class="docutils literal notranslate"><span class="pre">from_keras.py</span></code>)</p></td>
-<td><p>00:14.587</p></td>
+<td><p>00:15.148</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-even"><td><p><a class="reference internal" href="from_onnx.html#sphx-glr-how-to-compile-models-from-onnx-py"><span class="std std-ref">Compile ONNX Models</span></a> (<code class="docutils literal notranslate"><span class="pre">from_onnx.py</span></code>)</p></td>
-<td><p>00:02.396</p></td>
+<td><p>00:02.279</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 </tbody>
diff --git a/docs/how_to/deploy_models/deploy_model_on_android.html b/docs/how_to/deploy_models/deploy_model_on_android.html
index 900e486c3..f8b803e8e 100644
--- a/docs/how_to/deploy_models/deploy_model_on_android.html
+++ b/docs/how_to/deploy_models/deploy_model_on_android.html
@@ -648,7 +648,7 @@ to the remote android device.</p>
 Evaluate inference time cost...
 Execution time summary:
  mean (ms)   median (ms)    max (ms)     min (ms)     std (ms)
-  16.0766      16.0816      16.2853      15.8805       0.1172
+  16.2731      16.1871      16.7614      15.8627       0.3202
 </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 b09259fd0..a7d134931 100644
--- a/docs/how_to/deploy_models/deploy_object_detection_pytorch.html
+++ b/docs/how_to/deploy_models/deploy_object_detection_pytorch.html
@@ -431,14 +431,14 @@ be unstable.</p>
 <div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Downloading: &quot;https://download.pytorch.org/models/maskrcnn_resnet50_fpn_coco-bf2d0c1e.pth&quot; to /workspace/.cache/torch/hub/checkpoints/maskrcnn_resnet50_fpn_coco-bf2d0c1e.pth
 
   0%|          | 0.00/170M [00:00&lt;?, ?B/s]
-  8%|8         | 13.9M/170M [00:00&lt;00:01, 143MB/s]
- 16%|#6        | 27.6M/170M [00:00&lt;00:01, 126MB/s]
- 30%|###       | 51.2M/170M [00:00&lt;00:00, 178MB/s]
- 45%|####4     | 76.2M/170M [00:00&lt;00:00, 209MB/s]
- 58%|#####7    | 98.3M/170M [00:00&lt;00:00, 217MB/s]
- 72%|#######1  | 121M/170M [00:00&lt;00:00, 226MB/s]
- 86%|########6 | 147M/170M [00:00&lt;00:00, 238MB/s]
-100%|##########| 170M/170M [00:00&lt;00:00, 219MB/s]
+  8%|8         | 14.4M/170M [00:00&lt;00:01, 151MB/s]
+ 21%|##        | 35.2M/170M [00:00&lt;00:00, 190MB/s]
+ 34%|###3      | 57.5M/170M [00:00&lt;00:00, 210MB/s]
+ 47%|####6     | 79.4M/170M [00:00&lt;00:00, 218MB/s]
+ 61%|######    | 103M/170M [00:00&lt;00:00, 230MB/s]
+ 74%|#######4  | 126M/170M [00:00&lt;00:00, 234MB/s]
+ 88%|########7 | 149M/170M [00:00&lt;00:00, 230MB/s]
+100%|##########| 170M/170M [00:00&lt;00:00, 224MB/s]
 /usr/local/lib/python3.7/dist-packages/torch/nn/functional.py:3878: UserWarning: To copy construct from a tensor, it is recommended to use sourceTensor.clone().detach() or sourceTensor.clone().detach().requires_grad_(True), rather than torch.tensor(sourceTensor).
   for i in range(dim)
 /usr/local/lib/python3.7/dist-packages/torchvision/models/detection/anchor_utils.py:127: UserWarning: __floordiv__ is deprecated, and its behavior will change in a future version of pytorch. It currently rounds toward 0 (like the &#39;trunc&#39; function NOT &#39;floor&#39;). This results in incorrect rounding for negative values. To keep the current behavior, use torch.div(a, b, rounding_mode=&#39;trunc&#39;), or for actual floor division, use torch.div(a, b, rounding_mode=&#39;floor&#39;).
@@ -533,7 +533,7 @@ 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  2.841 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 2 minutes  57.665 seconds)</p>
 <div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-deploy-models-deploy-object-detection-pytorch-py">
 <div class="sphx-glr-download sphx-glr-download-python docutils container">
 <p><a class="reference download internal" download="" href="../../_downloads/7795da4b258c8feff986668b95ef57ad/deploy_object_detection_pytorch.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">deploy_object_detection_pytorch.py</span></code></a></p>
diff --git a/docs/how_to/deploy_models/deploy_prequantized.html b/docs/how_to/deploy_models/deploy_prequantized.html
index 79f87bee4..8a98ce4ee 100644
--- a/docs/how_to/deploy_models/deploy_prequantized.html
+++ b/docs/how_to/deploy_models/deploy_prequantized.html
@@ -475,22 +475,7 @@ training. Other models require a full post training calibration.</p>
 <div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Downloading: &quot;https://download.pytorch.org/models/mobilenet_v2-b0353104.pth&quot; to /workspace/.cache/torch/hub/checkpoints/mobilenet_v2-b0353104.pth
 
   0%|          | 0.00/13.6M [00:00&lt;?, ?B/s]
-  0%|          | 40.0k/13.6M [00:00&lt;00:48, 290kB/s]
-  1%|          | 88.0k/13.6M [00:00&lt;00:37, 373kB/s]
-  2%|1         | 216k/13.6M [00:00&lt;00:20, 687kB/s]
-  3%|3         | 464k/13.6M [00:00&lt;00:10, 1.28MB/s]
-  7%|6         | 968k/13.6M [00:00&lt;00:05, 2.48MB/s]
- 13%|#2        | 1.71M/13.6M [00:00&lt;00:02, 4.20MB/s]
- 21%|##        | 2.82M/13.6M [00:00&lt;00:01, 6.48MB/s]
- 28%|##7       | 3.78M/13.6M [00:00&lt;00:01, 7.57MB/s]
- 37%|###6      | 4.97M/13.6M [00:01&lt;00:01, 8.84MB/s]
- 47%|####7     | 6.39M/13.6M [00:01&lt;00:00, 10.3MB/s]
- 58%|#####7    | 7.82M/13.6M [00:01&lt;00:00, 11.5MB/s]
- 67%|######7   | 9.13M/13.6M [00:01&lt;00:00, 12.2MB/s]
- 76%|#######6  | 10.3M/13.6M [00:01&lt;00:00, 11.9MB/s]
- 84%|########4 | 11.5M/13.6M [00:01&lt;00:00, 11.7MB/s]
- 93%|#########2| 12.6M/13.6M [00:01&lt;00:00, 11.4MB/s]
-100%|##########| 13.6M/13.6M [00:01&lt;00:00, 8.40MB/s]
+100%|##########| 13.6M/13.6M [00:00&lt;00:00, 174MB/s]
 </pre></div>
 </div>
 </div>
@@ -579,7 +564,7 @@ output values are identical out of 1000 outputs from mobilenet v2.</p>
 </div>
 <div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Execution time summary:
  mean (ms)   median (ms)    max (ms)     min (ms)     std (ms)
-  90.3297      90.2381      93.5955      90.1025       0.3959
+  90.2922      90.2391      92.0132      90.0752       0.2443
 </pre></div>
 </div>
 <div class="admonition note">
@@ -618,7 +603,7 @@ This includes support for the VNNI 8 bit dot product instruction (CascadeLake or
 <div class="section" id="deploy-a-quantized-tflite-model">
 <h2>Deploy a quantized TFLite Model<a class="headerlink" href="#deploy-a-quantized-tflite-model" title="Permalink to this headline">¶</a></h2>
 <p>TODO</p>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes  11.268 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes  9.087 seconds)</p>
 <div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-deploy-models-deploy-prequantized-py">
 <div class="sphx-glr-download sphx-glr-download-python docutils container">
 <p><a class="reference download internal" download="" href="../../_downloads/fb8217c13f4351224c6cf3aacf1a87fc/deploy_prequantized.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">deploy_prequantized.py</span></code></a></p>
diff --git a/docs/how_to/deploy_models/deploy_prequantized_tflite.html b/docs/how_to/deploy_models/deploy_prequantized_tflite.html
index 6863dc545..415f21107 100644
--- a/docs/how_to/deploy_models/deploy_prequantized_tflite.html
+++ b/docs/how_to/deploy_models/deploy_prequantized_tflite.html
@@ -568,7 +568,7 @@ TFLite Top-5 labels: [387 102 386 341 349]
 </div>
 <div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Execution time summary:
  mean (ms)   median (ms)    max (ms)     min (ms)     std (ms)
-  120.7101     120.6807     121.4579     120.0139      0.3205
+  119.9342     119.9348     120.8328     119.1156      0.3515
 </pre></div>
 </div>
 <div class="admonition note">
@@ -596,7 +596,7 @@ network for ARM CPU</span></a>.</p></li>
 </ul>
 </div></blockquote>
 </div>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes  59.003 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes  58.670 seconds)</p>
 <div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-deploy-models-deploy-prequantized-tflite-py">
 <div class="sphx-glr-download sphx-glr-download-python docutils container">
 <p><a class="reference download internal" download="" href="../../_downloads/56691c7a27d45da61d112276334640d3/deploy_prequantized_tflite.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">deploy_prequantized_tflite.py</span></code></a></p>
diff --git a/docs/how_to/deploy_models/deploy_quantized.html b/docs/how_to/deploy_models/deploy_quantized.html
index 2194a0155..7b4e6fab7 100644
--- a/docs/how_to/deploy_models/deploy_quantized.html
+++ b/docs/how_to/deploy_models/deploy_quantized.html
@@ -504,7 +504,7 @@ for calibration. But the accuracy might be impacted.</p>
   DeprecationWarning,
 </pre></div>
 </div>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes  32.419 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes  35.140 seconds)</p>
 <div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-deploy-models-deploy-quantized-py">
 <div class="sphx-glr-download sphx-glr-download-python docutils container">
 <p><a class="reference download internal" download="" href="../../_downloads/7810ecf51bfc05f7d5e8a400ac3e815d/deploy_quantized.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">deploy_quantized.py</span></code></a></p>
diff --git a/docs/how_to/deploy_models/deploy_ssd_gluoncv.html b/docs/how_to/deploy_models/deploy_ssd_gluoncv.html
index c1f16c2f9..993032a17 100644
--- a/docs/how_to/deploy_models/deploy_ssd_gluoncv.html
+++ b/docs/how_to/deploy_models/deploy_ssd_gluoncv.html
@@ -436,24 +436,23 @@ 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&lt;?, ?KB/s]
-  5%|4         | 6047/132723 [00:00&lt;00:02, 60458.67KB/s]
- 10%|#         | 13851/132723 [00:00&lt;00:01, 70789.25KB/s]
- 16%|#5        | 20930/132723 [00:00&lt;00:02, 38392.67KB/s]
- 22%|##1       | 28746/132723 [00:00&lt;00:02, 48780.93KB/s]
- 26%|##6       | 34846/132723 [00:00&lt;00:01, 49733.35KB/s]
- 32%|###2      | 42737/132723 [00:00&lt;00:01, 57621.30KB/s]
- 38%|###8      | 50567/132723 [00:00&lt;00:01, 63381.71KB/s]
- 44%|####4     | 58432/132723 [00:01&lt;00:01, 67726.62KB/s]
- 49%|####9     | 65663/132723 [00:01&lt;00:01, 66884.73KB/s]
- 55%|#####5    | 73245/132723 [00:01&lt;00:00, 69435.86KB/s]
- 61%|######1   | 81281/132723 [00:01&lt;00:00, 72601.27KB/s]
- 67%|######7   | 89135/132723 [00:01&lt;00:00, 74339.15KB/s]
- 73%|#######3  | 97001/132723 [00:01&lt;00:00, 75609.88KB/s]
- 79%|#######9  | 104914/132723 [00:01&lt;00:00, 76650.79KB/s]
- 85%|########4 | 112774/132723 [00:01&lt;00:00, 77227.12KB/s]
- 91%|######### | 120546/132723 [00:01&lt;00:00, 75506.82KB/s]
- 97%|#########7| 129011/132723 [00:01&lt;00:00, 78189.14KB/s]
-100%|##########| 132723/132723 [00:01&lt;00:00, 67463.49KB/s]
+  4%|4         | 5693/132723 [00:00&lt;00:02, 56919.33KB/s]
+ 11%|#         | 14125/132723 [00:00&lt;00:01, 73029.72KB/s]
+ 16%|#6        | 21428/132723 [00:00&lt;00:02, 45469.78KB/s]
+ 22%|##2       | 29856/132723 [00:00&lt;00:01, 56677.01KB/s]
+ 29%|##8       | 38277/132723 [00:00&lt;00:01, 64713.77KB/s]
+ 35%|###5      | 46823/132723 [00:00&lt;00:01, 70818.86KB/s]
+ 41%|####1     | 54516/132723 [00:00&lt;00:01, 65459.78KB/s]
+ 47%|####7     | 63013/132723 [00:00&lt;00:00, 70843.29KB/s]
+ 54%|#####3    | 71250/132723 [00:01&lt;00:00, 74101.28KB/s]
+ 60%|######    | 79758/132723 [00:01&lt;00:00, 77257.66KB/s]
+ 66%|######6   | 87722/132723 [00:01&lt;00:00, 73035.84KB/s]
+ 73%|#######2  | 96231/132723 [00:01&lt;00:00, 76412.74KB/s]
+ 78%|#######8  | 104042/132723 [00:01&lt;00:00, 60262.14KB/s]
+ 85%|########4 | 112624/132723 [00:01&lt;00:00, 66461.17KB/s]
+ 90%|######### | 119857/132723 [00:01&lt;00:00, 63444.31KB/s]
+ 97%|#########6| 128272/132723 [00:01&lt;00:00, 68732.98KB/s]
+100%|##########| 132723/132723 [00:01&lt;00:00, 67320.76KB/s]
 </pre></div>
 </div>
 <p>Create TVM runtime and do inference
@@ -496,7 +495,7 @@ Downloading /workspace/.mxnet/models/ssd_512_resnet50_v1_voc-9c8b225a.zip from h
 <span class="n">plt</span><span class="o">.</span><span class="n">show</span><span class="p">()</span>
 </pre></div>
 </div>
-<img src="../../_images/sphx_glr_deploy_ssd_gluoncv_001.png" srcset="../../_images/sphx_glr_deploy_ssd_gluoncv_001.png" alt="deploy ssd gluoncv" class = "sphx-glr-single-img"/><p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 2 minutes  32.705 seconds)</p>
+<img src="../../_images/sphx_glr_deploy_ssd_gluoncv_001.png" srcset="../../_images/sphx_glr_deploy_ssd_gluoncv_001.png" alt="deploy ssd gluoncv" class = "sphx-glr-single-img"/><p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 2 minutes  34.549 seconds)</p>
 <div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-deploy-models-deploy-ssd-gluoncv-py">
 <div class="sphx-glr-download sphx-glr-download-python docutils container">
 <p><a class="reference download internal" download="" href="../../_downloads/cccb17d28e5e8b2e94ea8cd5ec59f6ed/deploy_ssd_gluoncv.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">deploy_ssd_gluoncv.py</span></code></a></p>
diff --git a/docs/how_to/deploy_models/sg_execution_times.html b/docs/how_to/deploy_models/sg_execution_times.html
index af93c675c..43c57bd30 100644
--- a/docs/how_to/deploy_models/sg_execution_times.html
+++ b/docs/how_to/deploy_models/sg_execution_times.html
@@ -322,7 +322,7 @@
             
   <div class="section" id="computation-times">
 <span id="sphx-glr-how-to-deploy-models-sg-execution-times"></span><h1>Computation times<a class="headerlink" href="#computation-times" title="Permalink to this headline">¶</a></h1>
-<p><strong>11:12.097</strong> total execution time for <strong>how_to_deploy_models</strong> files:</p>
+<p><strong>11:07.409</strong> total execution time for <strong>how_to_deploy_models</strong> files:</p>
 <table class="docutils align-default">
 <colgroup>
 <col style="width: 86%" />
@@ -331,31 +331,31 @@
 </colgroup>
 <tbody>
 <tr class="row-odd"><td><p><a class="reference internal" href="deploy_object_detection_pytorch.html#sphx-glr-how-to-deploy-models-deploy-object-detection-pytorch-py"><span class="std std-ref">Compile PyTorch Object Detection Models</span></a> (<code class="docutils literal notranslate"><span class="pre">deploy_object_detection_pytorch.py</span></code>)</p></td>
-<td><p>03:02.841</p></td>
+<td><p>02:57.665</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-even"><td><p><a class="reference internal" href="deploy_ssd_gluoncv.html#sphx-glr-how-to-deploy-models-deploy-ssd-gluoncv-py"><span class="std std-ref">Deploy Single Shot Multibox Detector(SSD) model</span></a> (<code class="docutils literal notranslate"><span class="pre">deploy_ssd_gluoncv.py</span></code>)</p></td>
-<td><p>02:32.705</p></td>
+<td><p>02:34.549</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-odd"><td><p><a class="reference internal" href="deploy_prequantized_tflite.html#sphx-glr-how-to-deploy-models-deploy-prequantized-tflite-py"><span class="std std-ref">Deploy a Framework-prequantized Model with TVM - Part 3 (TFLite)</span></a> (<code class="docutils literal notranslate"><span class="pre">deploy_prequantized_tflite.py</span></code>)</p></td>
-<td><p>01:59.003</p></td>
+<td><p>01:58.670</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-even"><td><p><a class="reference internal" href="deploy_quantized.html#sphx-glr-how-to-deploy-models-deploy-quantized-py"><span class="std std-ref">Deploy a Quantized Model on Cuda</span></a> (<code class="docutils literal notranslate"><span class="pre">deploy_quantized.py</span></code>)</p></td>
-<td><p>01:32.419</p></td>
+<td><p>01:35.140</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-odd"><td><p><a class="reference internal" href="deploy_prequantized.html#sphx-glr-how-to-deploy-models-deploy-prequantized-py"><span class="std std-ref">Deploy a Framework-prequantized Model with TVM</span></a> (<code class="docutils literal notranslate"><span class="pre">deploy_prequantized.py</span></code>)</p></td>
-<td><p>01:11.268</p></td>
+<td><p>01:09.087</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-even"><td><p><a class="reference internal" href="deploy_model_on_android.html#sphx-glr-how-to-deploy-models-deploy-model-on-android-py"><span class="std std-ref">Deploy the Pretrained Model on Android</span></a> (<code class="docutils literal notranslate"><span class="pre">deploy_model_on_android.py</span></code>)</p></td>
-<td><p>00:30.519</p></td>
+<td><p>00:29.455</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-odd"><td><p><a class="reference internal" href="deploy_model_on_rasp.html#sphx-glr-how-to-deploy-models-deploy-model-on-rasp-py"><span class="std std-ref">Deploy the Pretrained Model on Raspberry Pi</span></a> (<code class="docutils literal notranslate"><span class="pre">deploy_model_on_rasp.py</span></code>)</p></td>
-<td><p>00:23.335</p></td>
+<td><p>00:22.838</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-even"><td><p><a class="reference internal" href="deploy_sparse.html#sphx-glr-how-to-deploy-models-deploy-sparse-py"><span class="std std-ref">Deploy a Hugging Face Pruned Model on CPU</span></a> (<code class="docutils literal notranslate"><span class="pre">deploy_sparse.py</span></code>)</p></td>
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 eccf23867..e18a78868 100644
--- a/docs/how_to/extend_tvm/bring_your_own_datatypes.html
+++ b/docs/how_to/extend_tvm/bring_your_own_datatypes.html
@@ -607,7 +607,7 @@ In this alpha state of the Bring Your Own Datatypes framework, we have not imple
 <span class="n">module</span><span class="p">,</span> <a href="https://docs.python.org/3/library/stdtypes.html#dict" title="builtins.dict" class="sphx-glr-backref-module-builtins sphx-glr-backref-type-py-class sphx-glr-backref-instance"><span class="n">params</span></a> <span class="o">=</span> <span class="n">get_mobilenet</span><span class="p">()</span>
 </pre></div>
 </div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Downloading /workspace/.mxnet/models/mobilenet0.25-9f83e440.zipcc65caf3-e88d-4049-b12f-b94bc5ee1995 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.zip34156840-1b6c-4cb3-ab83-81270d9ed63d 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 3a17cda81..08ea4d269 100644
--- a/docs/how_to/extend_tvm/sg_execution_times.html
+++ b/docs/how_to/extend_tvm/sg_execution_times.html
@@ -322,7 +322,7 @@
             
   <div class="section" id="computation-times">
 <span id="sphx-glr-how-to-extend-tvm-sg-execution-times"></span><h1>Computation times<a class="headerlink" href="#computation-times" title="Permalink to this headline">¶</a></h1>
-<p><strong>00:40.059</strong> total execution time for <strong>how_to_extend_tvm</strong> files:</p>
+<p><strong>00:40.453</strong> total execution time for <strong>how_to_extend_tvm</strong> files:</p>
 <table class="docutils align-default">
 <colgroup>
 <col style="width: 84%" />
@@ -331,15 +331,15 @@
 </colgroup>
 <tbody>
 <tr class="row-odd"><td><p><a class="reference internal" href="bring_your_own_datatypes.html#sphx-glr-how-to-extend-tvm-bring-your-own-datatypes-py"><span class="std std-ref">Bring Your Own Datatypes to TVM</span></a> (<code class="docutils literal notranslate"><span class="pre">bring_your_own_datatypes.py</span></code>)</p></td>
-<td><p>00:36.917</p></td>
+<td><p>00:37.260</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-even"><td><p><a class="reference internal" href="use_pass_instrument.html#sphx-glr-how-to-extend-tvm-use-pass-instrument-py"><span class="std std-ref">How to Use TVM Pass Instrument</span></a> (<code class="docutils literal notranslate"><span class="pre">use_pass_instrument.py</span></code>)</p></td>
-<td><p>00:02.218</p></td>
+<td><p>00:02.250</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-odd"><td><p><a class="reference internal" href="use_pass_infra.html#sphx-glr-how-to-extend-tvm-use-pass-infra-py"><span class="std std-ref">How to Use TVM Pass Infra</span></a> (<code class="docutils literal notranslate"><span class="pre">use_pass_infra.py</span></code>)</p></td>
-<td><p>00:00.916</p></td>
+<td><p>00:00.934</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-even"><td><p><a class="reference internal" href="low_level_custom_pass.html#sphx-glr-how-to-extend-tvm-low-level-custom-pass-py"><span class="std std-ref">Writing a Customized Pass</span></a> (<code class="docutils literal notranslate"><span class="pre">low_level_custom_pass.py</span></code>)</p></td>
diff --git a/docs/how_to/extend_tvm/use_pass_instrument.html b/docs/how_to/extend_tvm/use_pass_instrument.html
index 9bd331e87..649859bc0 100644
--- a/docs/how_to/extend_tvm/use_pass_instrument.html
+++ b/docs/how_to/extend_tvm/use_pass_instrument.html
@@ -507,10 +507,10 @@ profile the execution time of each passes.</p>
 </pre></div>
 </div>
 <div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Printing results of timing profile...
-InferType: 6695us [6695us] (45.79%; 45.79%)
-FoldScaleAxis: 7925us [5us] (54.21%; 54.21%)
-        FoldConstant: 7920us [1593us] (54.17%; 99.93%)
-                InferType: 6327us [6327us] (43.28%; 79.89%)
+InferType: 6504us [6504us] (45.81%; 45.81%)
+FoldScaleAxis: 7694us [5us] (54.19%; 54.19%)
+        FoldConstant: 7688us [1584us] (54.15%; 99.93%)
+                InferType: 6104us [6104us] (42.99%; 79.39%)
 </pre></div>
 </div>
 </div>
@@ -532,10 +532,10 @@ Refer to following sections and <a class="reference internal" href="../../refere
 </pre></div>
 </div>
 <div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Printing results of timing profile...
-InferType: 6308us [6308us] (45.02%; 45.02%)
-FoldScaleAxis: 7705us [5us] (54.98%; 54.98%)
-        FoldConstant: 7700us [1587us] (54.95%; 99.94%)
-                InferType: 6113us [6113us] (43.62%; 79.39%)
+InferType: 6124us [6124us] (44.68%; 44.68%)
+FoldScaleAxis: 7584us [4us] (55.32%; 55.32%)
+        FoldConstant: 7579us [1566us] (55.29%; 99.94%)
+                InferType: 6013us [6013us] (43.87%; 79.33%)
 </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 3275ff160..64ed6952b 100644
--- a/docs/how_to/optimize_operators/opt_conv_cuda.html
+++ b/docs/how_to/optimize_operators/opt_conv_cuda.html
@@ -559,7 +559,7 @@ latency of convolution.</p>
 <span class="nb">print</span><span class="p">(</span><span class="s2">&quot;Convolution: </span><span class="si">%f</span><span class="s2"> ms&quot;</span> <span class="o">%</span> <span class="p">(</span><span class="n">evaluator</span><span class="p">(</span><span class="n">a</span><span class="p">,</span> <span class="n">w</span><span class="p">,</span> <span class="n">b</span><span class="p">)</span><span class="o">.</span><span class="n">mean</span> <span class="o">*</span> <span cl [...]
 </pre></div>
 </div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Convolution: 54.160442 ms
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Convolution: 49.836182 ms
 </pre></div>
 </div>
 <div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-optimize-operators-opt-conv-cuda-py">
diff --git a/docs/how_to/optimize_operators/opt_conv_tensorcore.html b/docs/how_to/optimize_operators/opt_conv_tensorcore.html
index 2ee23a6c0..892071679 100644
--- a/docs/how_to/optimize_operators/opt_conv_tensorcore.html
+++ b/docs/how_to/optimize_operators/opt_conv_tensorcore.html
@@ -901,7 +901,7 @@ be able to run on our build server</p>
     <span class="nb">print</span><span class="p">(</span><span class="s2">&quot;conv2d with tensor core: </span><span class="si">%f</span><span class="s2"> ms&quot;</span> <span class="o">%</span> <span class="p">(</span><span class="n">evaluator</span><span class="p">(</span><span class="n">a</span><span class="p">,</span> <span class="n">w</span><span class="p">,</span> <span class="n">c</span><span class="p">)</span><span class="o">.</span><span class="n">mean</span> <span class="o">* [...]
 </pre></div>
 </div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>conv2d with tensor core: 9.090662 ms
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>conv2d with tensor core: 6.863631 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 c61a6909b..8724aeb08 100644
--- a/docs/how_to/optimize_operators/opt_gemm.html
+++ b/docs/how_to/optimize_operators/opt_gemm.html
@@ -456,8 +456,8 @@ Then we write a baseline implementation, the simplest way to write a matrix mult
 <span class="nb">print</span><span class="p">(</span><span class="s2">&quot;Baseline: </span><span class="si">%f</span><span class="s2">&quot;</span> <span class="o">%</span> <span class="n">evaluator</span><span class="p">(</span><span class="n">a</span><span class="p">,</span> <span class="n">b</span><span class="p">,</span> <span class="n">c</span><span class="p">)</span><span class="o">.</span><span class="n">mean</span><span class="p">)</span>
 </pre></div>
 </div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Numpy running time: 0.018934
-Baseline: 3.386234
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Numpy running time: 0.018806
+Baseline: 3.216682
 </pre></div>
 </div>
 <p>In TVM, we can always inspect lower level IR to debug or optimize our schedule.
@@ -517,7 +517,7 @@ fill 32 * 32 * sizeof(float) which is 4KB in the cache whose total size is 32KB
 <span class="nb">print</span><span class="p">(</span><span class="s2">&quot;Opt1: </span><span class="si">%f</span><span class="s2">&quot;</span> <span class="o">%</span> <span class="n">evaluator</span><span class="p">(</span><span class="n">a</span><span class="p">,</span> <span class="n">b</span><span class="p">,</span> <span class="n">c</span><span class="p">)</span><span class="o">.</span><span class="n">mean</span><span class="p">)</span>
 </pre></div>
 </div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt1: 0.289869
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt1: 0.299593
 </pre></div>
 </div>
 <p>Here is the generated IR after blocking.</p>
@@ -584,7 +584,7 @@ vastly.</p>
 <span class="nb">print</span><span class="p">(</span><span class="s2">&quot;Opt2: </span><span class="si">%f</span><span class="s2">&quot;</span> <span class="o">%</span> <span class="n">evaluator</span><span class="p">(</span><span class="n">a</span><span class="p">,</span> <span class="n">b</span><span class="p">,</span> <span class="n">c</span><span class="p">)</span><span class="o">.</span><span class="n">mean</span><span class="p">)</span>
 </pre></div>
 </div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt2: 0.341522
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt2: 0.343425
 </pre></div>
 </div>
 <p>Here is the generated IR after vectorization.</p>
@@ -645,7 +645,7 @@ the access pattern for A matrix is more cache friendly.</p>
 <span class="nb">print</span><span class="p">(</span><span class="s2">&quot;Opt3: </span><span class="si">%f</span><span class="s2">&quot;</span> <span class="o">%</span> <span class="n">evaluator</span><span class="p">(</span><span class="n">a</span><span class="p">,</span> <span class="n">b</span><span class="p">,</span> <span class="n">c</span><span class="p">)</span><span class="o">.</span><span class="n">mean</span><span class="p">)</span>
 </pre></div>
 </div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt3: 0.120605
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt3: 0.114249
 </pre></div>
 </div>
 <p>Here is the generated IR after loop permutation.</p>
@@ -728,7 +728,7 @@ flattening.</p>
 <span class="nb">print</span><span class="p">(</span><span class="s2">&quot;Opt4: </span><span class="si">%f</span><span class="s2">&quot;</span> <span class="o">%</span> <span class="n">evaluator</span><span class="p">(</span><span class="n">a</span><span class="p">,</span> <span class="n">b</span><span class="p">,</span> <span class="n">c</span><span class="p">)</span><span class="o">.</span><span class="n">mean</span><span class="p">)</span>
 </pre></div>
 </div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt4: 0.110875
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt4: 0.109088
 </pre></div>
 </div>
 <p>Here is the generated IR after array packing.</p>
@@ -814,7 +814,7 @@ write to C when all the block results are ready.</p>
 <span class="nb">print</span><span class="p">(</span><span class="s2">&quot;Opt5: </span><span class="si">%f</span><span class="s2">&quot;</span> <span class="o">%</span> <span class="n">evaluator</span><span class="p">(</span><span class="n">a</span><span class="p">,</span> <span class="n">b</span><span class="p">,</span> <span class="n">c</span><span class="p">)</span><span class="o">.</span><span class="n">mean</span><span class="p">)</span>
 </pre></div>
 </div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt5: 0.111340
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt5: 0.110632
 </pre></div>
 </div>
 <p>Here is the generated IR after blocking.</p>
@@ -904,7 +904,7 @@ write to C when all the block results are ready.</p>
 <span class="nb">print</span><span class="p">(</span><span class="s2">&quot;Opt6: </span><span class="si">%f</span><span class="s2">&quot;</span> <span class="o">%</span> <span class="n">opt6_time</span><span class="p">)</span>
 </pre></div>
 </div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt6: 0.144972
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt6: 0.144890
 </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 45bd0f09e..2a1779318 100644
--- a/docs/how_to/optimize_operators/sg_execution_times.html
+++ b/docs/how_to/optimize_operators/sg_execution_times.html
@@ -322,7 +322,7 @@
             
   <div class="section" id="computation-times">
 <span id="sphx-glr-how-to-optimize-operators-sg-execution-times"></span><h1>Computation times<a class="headerlink" href="#computation-times" title="Permalink to this headline">¶</a></h1>
-<p><strong>00:34.419</strong> total execution time for <strong>how_to_optimize_operators</strong> files:</p>
+<p><strong>00:33.887</strong> total execution time for <strong>how_to_optimize_operators</strong> files:</p>
 <table class="docutils align-default">
 <colgroup>
 <col style="width: 83%" />
@@ -331,15 +331,15 @@
 </colgroup>
 <tbody>
 <tr class="row-odd"><td><p><a class="reference internal" href="opt_gemm.html#sphx-glr-how-to-optimize-operators-opt-gemm-py"><span class="std std-ref">How to optimize GEMM on CPU</span></a> (<code class="docutils literal notranslate"><span class="pre">opt_gemm.py</span></code>)</p></td>
-<td><p>00:32.168</p></td>
+<td><p>00:31.654</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-even"><td><p><a class="reference internal" href="opt_conv_tensorcore.html#sphx-glr-how-to-optimize-operators-opt-conv-tensorcore-py"><span class="std std-ref">How to optimize convolution using TensorCores</span></a> (<code class="docutils literal notranslate"><span class="pre">opt_conv_tensorcore.py</span></code>)</p></td>
-<td><p>00:01.253</p></td>
+<td><p>00:01.183</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-odd"><td><p><a class="reference internal" href="opt_conv_cuda.html#sphx-glr-how-to-optimize-operators-opt-conv-cuda-py"><span class="std std-ref">How to optimize convolution on GPU</span></a> (<code class="docutils literal notranslate"><span class="pre">opt_conv_cuda.py</span></code>)</p></td>
-<td><p>00:00.997</p></td>
+<td><p>00:01.050</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 </tbody>
diff --git a/docs/how_to/tune_with_autoscheduler/sg_execution_times.html b/docs/how_to/tune_with_autoscheduler/sg_execution_times.html
index e29c8aa38..a957e6694 100644
--- a/docs/how_to/tune_with_autoscheduler/sg_execution_times.html
+++ b/docs/how_to/tune_with_autoscheduler/sg_execution_times.html
@@ -322,7 +322,7 @@
             
   <div class="section" id="computation-times">
 <span id="sphx-glr-how-to-tune-with-autoscheduler-sg-execution-times"></span><h1>Computation times<a class="headerlink" href="#computation-times" title="Permalink to this headline">¶</a></h1>
-<p><strong>05:54.523</strong> total execution time for <strong>how_to_tune_with_autoscheduler</strong> files:</p>
+<p><strong>05:56.604</strong> total execution time for <strong>how_to_tune_with_autoscheduler</strong> files:</p>
 <table class="docutils align-default">
 <colgroup>
 <col style="width: 85%" />
@@ -331,27 +331,27 @@
 </colgroup>
 <tbody>
 <tr class="row-odd"><td><p><a class="reference internal" href="tune_conv2d_layer_cuda.html#sphx-glr-how-to-tune-with-autoscheduler-tune-conv2d-layer-cuda-py"><span class="std std-ref">Auto-scheduling a Convolution Layer for GPU</span></a> (<code class="docutils literal notranslate"><span class="pre">tune_conv2d_layer_cuda.py</span></code>)</p></td>
-<td><p>03:11.157</p></td>
+<td><p>03:10.386</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-even"><td><p><a class="reference internal" href="tune_network_x86.html#sphx-glr-how-to-tune-with-autoscheduler-tune-network-x86-py"><span class="std std-ref">Auto-scheduling a Neural Network for x86 CPU</span></a> (<code class="docutils literal notranslate"><span class="pre">tune_network_x86.py</span></code>)</p></td>
-<td><p>01:21.967</p></td>
+<td><p>01:22.515</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-odd"><td><p><a class="reference internal" href="tune_network_cuda.html#sphx-glr-how-to-tune-with-autoscheduler-tune-network-cuda-py"><span class="std std-ref">Auto-scheduling a Neural Network for NVIDIA GPU</span></a> (<code class="docutils literal notranslate"><span class="pre">tune_network_cuda.py</span></code>)</p></td>
-<td><p>00:45.629</p></td>
+<td><p>00:45.653</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-even"><td><p><a class="reference internal" href="tune_sparse_x86.html#sphx-glr-how-to-tune-with-autoscheduler-tune-sparse-x86-py"><span class="std std-ref">Auto-scheduling Sparse Matrix Multiplication on CPU with Custom Sketch Rule</span></a> (<code class="docutils literal notranslate"><span class="pre">tune_sparse_x86.py</span></code>)</p></td>
-<td><p>00:18.431</p></td>
+<td><p>00:20.490</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-odd"><td><p><a class="reference internal" href="tune_network_mali.html#sphx-glr-how-to-tune-with-autoscheduler-tune-network-mali-py"><span class="std std-ref">Auto-scheduling a Neural Network for mali GPU</span></a> (<code class="docutils literal notranslate"><span class="pre">tune_network_mali.py</span></code>)</p></td>
-<td><p>00:08.821</p></td>
+<td><p>00:08.951</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-even"><td><p><a class="reference internal" href="tune_network_arm.html#sphx-glr-how-to-tune-with-autoscheduler-tune-network-arm-py"><span class="std std-ref">Auto-scheduling a Neural Network for ARM CPU</span></a> (<code class="docutils literal notranslate"><span class="pre">tune_network_arm.py</span></code>)</p></td>
-<td><p>00:08.518</p></td>
+<td><p>00:08.609</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 </tbody>
diff --git a/docs/how_to/tune_with_autoscheduler/tune_conv2d_layer_cuda.html b/docs/how_to/tune_with_autoscheduler/tune_conv2d_layer_cuda.html
index b00e28b01..b943d9488 100644
--- a/docs/how_to/tune_with_autoscheduler/tune_conv2d_layer_cuda.html
+++ b/docs/how_to/tune_with_autoscheduler/tune_conv2d_layer_cuda.html
@@ -486,699 +486,485 @@ cooperative fetching, unrolling and operator fusion.</p>
              compute: Buffer(compute_2: Pointer(float32), float32, [25088], [])}
   buffer_map = {data_1: data, kernel_1: kernel, bias_1: bias, compute_1: compute}
   preflattened_buffer_map = {data_1: data_3: Buffer(data_2, float32, [1, 512, 7, 7], []), kernel_1: kernel_3: Buffer(kernel_2, float32, [512, 512, 3, 3], []), bias_1: bias_3: Buffer(bias_2, float32, [1, 512, 1, 1], []), compute_1: compute_3: Buffer(compute_2, float32, [1, 512, 7, 7], [])} {
-  attr [IterVar(blockIdx.x: int32, (nullptr), &quot;ThreadIndex&quot;, &quot;blockIdx.x&quot;)] &quot;thread_extent&quot; = 112;
-  allocate(conv2d_nchw: Pointer(local float32), float32, [14]), storage_scope = local;
-  allocate(pad_temp.shared: Pointer(shared float32), float32, [108]), storage_scope = shared;
-  allocate(kernel.shared: Pointer(shared float32), float32, [1152]), storage_scope = shared;
-  attr [IterVar(threadIdx.x: int32, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16 {
-    conv2d_nchw_1: Buffer(conv2d_nchw, float32, [14], [], scope=&quot;local&quot;, align=32)[0] = 0f32
-    conv2d_nchw_1[1] = 0f32
+  attr [IterVar(blockIdx.x: int32, (nullptr), &quot;ThreadIndex&quot;, &quot;blockIdx.x&quot;)] &quot;thread_extent&quot; = 128;
+  allocate(conv2d_nchw: Pointer(local float32), float32, [28]), storage_scope = local;
+  allocate(pad_temp.shared: Pointer(shared float32), float32, [252]), storage_scope = shared;
+  allocate(kernel.shared: Pointer(shared float32), float32, [48]), storage_scope = shared;
+  attr [IterVar(threadIdx.x: int32, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7 {
+    conv2d_nchw_1: Buffer(conv2d_nchw, float32, [4], [], scope=&quot;local&quot;, align=8)[0] = 0f32
     conv2d_nchw_1[2] = 0f32
-    conv2d_nchw_1[3] = 0f32
     conv2d_nchw_1[4] = 0f32
-    conv2d_nchw_1[5] = 0f32
     conv2d_nchw_1[6] = 0f32
-    conv2d_nchw_1[7] = 0f32
     conv2d_nchw_1[8] = 0f32
-    conv2d_nchw_1[9] = 0f32
     conv2d_nchw_1[10] = 0f32
-    conv2d_nchw_1[11] = 0f32
     conv2d_nchw_1[12] = 0f32
+    conv2d_nchw_1[14] = 0f32
+    conv2d_nchw_1[16] = 0f32
+    conv2d_nchw_1[18] = 0f32
+    conv2d_nchw_1[20] = 0f32
+    conv2d_nchw_1[22] = 0f32
+    conv2d_nchw_1[24] = 0f32
+    conv2d_nchw_1[26] = 0f32
+    conv2d_nchw_1[1] = 0f32
+    conv2d_nchw_1[3] = 0f32
+    conv2d_nchw_1[5] = 0f32
+    conv2d_nchw_1[7] = 0f32
+    conv2d_nchw_1[9] = 0f32
+    conv2d_nchw_1[11] = 0f32
     conv2d_nchw_1[13] = 0f32
+    conv2d_nchw_1[15] = 0f32
+    conv2d_nchw_1[17] = 0f32
+    conv2d_nchw_1[19] = 0f32
+    conv2d_nchw_1[21] = 0f32
+    conv2d_nchw_1[23] = 0f32
+    conv2d_nchw_1[25] = 0f32
+    conv2d_nchw_1[27] = 0f32
     for (rc.outer.outer: int32, 0, 128) {
-      let cse_var_2: int32 = (rc.outer.outer*196)
-      let cse_var_1: int32 = (rc.outer.outer*36)
-       {
-        attr [IterVar(threadIdx.x_1: int32, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        pad_temp.shared_1: Buffer(pad_temp.shared, float32, [108], [], scope=&quot;shared&quot;)[threadIdx.x_1] = @tir.if_then_else((((3 &lt;= threadIdx.x_1) &amp;&amp; (1 &lt;= (floormod(blockIdx.x, 7) + floormod(threadIdx.x_1, 3)))) &amp;&amp; ((floormod(blockIdx.x, 7) + floormod(threadIdx.x_1, 3)) &lt; 8)), data[((((cse_var_2 + (floordiv(threadIdx.x_1, 3)*7)) + floormod(blockIdx.x, 7)) + floormod(threadIdx.x_1, 3)) - 8)], 0f32, dtype=float32)
-        attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        pad_temp.shared_1[(threadIdx.x_1 + 16)] = @tir.if_then_else(((((3 &lt;= floormod((threadIdx.x_1 + 16), 27)) &amp;&amp; (floormod((threadIdx.x_1 + 16), 27) &lt; 24)) &amp;&amp; (1 &lt;= (floormod(blockIdx.x, 7) + floormod((threadIdx.x_1 + 1), 3)))) &amp;&amp; ((floormod(blockIdx.x, 7) + floormod((threadIdx.x_1 + 1), 3)) &lt; 8)), data[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 16), 27)*49)) + (floordiv(floormod((threadIdx.x_1 + 16), 27), 3)*7)) + floormod(blockIdx.x, 7)) + floorm [...]
-        attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        pad_temp.shared_1[(threadIdx.x_1 + 32)] = @tir.if_then_else(((1 &lt;= (floormod(blockIdx.x, 7) + floormod((threadIdx.x_1 + 2), 3))) &amp;&amp; ((floormod(blockIdx.x, 7) + floormod((threadIdx.x_1 + 2), 3)) &lt; 8)), data[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 32), 27)*49)) + (floordiv(floormod((threadIdx.x_1 + 5), 27), 3)*7)) + floormod(blockIdx.x, 7)) + floormod((threadIdx.x_1 + 2), 3)) - 8)], 0f32, dtype=float32)
-        attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        pad_temp.shared_1[(threadIdx.x_1 + 48)] = @tir.if_then_else(((((1 &lt;= floormod((floordiv(threadIdx.x_1, 3) + 7), 9)) &amp;&amp; (floormod((threadIdx.x_1 + 21), 27) &lt; 24)) &amp;&amp; (1 &lt;= (floormod(blockIdx.x, 7) + floormod(threadIdx.x_1, 3)))) &amp;&amp; ((floormod(blockIdx.x, 7) + floormod(threadIdx.x_1, 3)) &lt; 8)), data[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 48), 27)*49)) + (floormod((floordiv(threadIdx.x_1, 3) + 7), 9)*7)) + floormod(blockIdx.x, 7)) + floormod( [...]
-        attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        pad_temp.shared_1[(threadIdx.x_1 + 64)] = @tir.if_then_else((((threadIdx.x_1 &lt; 14) &amp;&amp; (1 &lt;= (floormod(blockIdx.x, 7) + floormod((threadIdx.x_1 + 1), 3)))) &amp;&amp; ((floormod(blockIdx.x, 7) + floormod((threadIdx.x_1 + 1), 3)) &lt; 8)), data[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 64), 27)*49)) + (floordiv(floormod((threadIdx.x_1 + 10), 27), 3)*7)) + floormod(blockIdx.x, 7)) + floormod((threadIdx.x_1 + 1), 3)) - 8)], 0f32, dtype=float32)
-        attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        pad_temp.shared_1[(threadIdx.x_1 + 80)] = @tir.if_then_else(((((3 &lt;= floormod((threadIdx.x_1 + 26), 27)) &amp;&amp; (floormod((threadIdx.x_1 + 26), 27) &lt; 24)) &amp;&amp; (1 &lt;= (floormod(blockIdx.x, 7) + floormod((threadIdx.x_1 + 2), 3)))) &amp;&amp; ((floormod(blockIdx.x, 7) + floormod((threadIdx.x_1 + 2), 3)) &lt; 8)), data[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 80), 27)*49)) + (floordiv(floormod((threadIdx.x_1 + 26), 27), 3)*7)) + floormod(blockIdx.x, 7)) + floorm [...]
-        attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        if @tir.likely((threadIdx.x_1 &lt; 12), dtype=bool) {
-          pad_temp.shared_1[(threadIdx.x_1 + 96)] = @tir.if_then_else((((threadIdx.x_1 &lt; 9) &amp;&amp; (1 &lt;= (floormod(blockIdx.x, 7) + floormod(threadIdx.x_1, 3)))) &amp;&amp; ((floormod(blockIdx.x, 7) + floormod(threadIdx.x_1, 3)) &lt; 8)), data[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 96), 27)*49)) + ((floordiv(threadIdx.x_1, 3) + 5)*7)) + floormod(blockIdx.x, 7)) + floormod(threadIdx.x_1, 3)) - 8)], 0f32, dtype=float32)
+      for (rx.outer.outer: int32, 0, 3) {
+        let cse_var_1: int32 = (rc.outer.outer*36)
+         {
+          attr [IterVar(threadIdx.x_1: int32, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          pad_temp.shared_1: Buffer(pad_temp.shared, float32, [252], [], scope=&quot;shared&quot;)[threadIdx.x_1] = 0f32
+          attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          pad_temp.shared_1[(threadIdx.x_1 + 7)] = @tir.if_then_else(((1 &lt;= (threadIdx.x_1 + rx.outer.outer)) &amp;&amp; ((threadIdx.x_1 + rx.outer.outer) &lt; 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) - 1)], 0f32, dtype=float32)
+          attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          pad_temp.shared_1[(threadIdx.x_1 + 14)] = @tir.if_then_else(((1 &lt;= (threadIdx.x_1 + rx.outer.outer)) &amp;&amp; ((threadIdx.x_1 + rx.outer.outer) &lt; 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 6)], 0f32, dtype=float32)
+          attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          pad_temp.shared_1[(threadIdx.x_1 + 21)] = @tir.if_then_else(((1 &lt;= (threadIdx.x_1 + rx.outer.outer)) &amp;&amp; ((threadIdx.x_1 + rx.outer.outer) &lt; 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 13)], 0f32, dtype=float32)
+          attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          pad_temp.shared_1[(threadIdx.x_1 + 28)] = @tir.if_then_else(((1 &lt;= (threadIdx.x_1 + rx.outer.outer)) &amp;&amp; ((threadIdx.x_1 + rx.outer.outer) &lt; 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 20)], 0f32, dtype=float32)
+          attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          pad_temp.shared_1[(threadIdx.x_1 + 35)] = @tir.if_then_else(((1 &lt;= (threadIdx.x_1 + rx.outer.outer)) &amp;&amp; ((threadIdx.x_1 + rx.outer.outer) &lt; 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 27)], 0f32, dtype=float32)
+          attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          pad_temp.shared_1[(threadIdx.x_1 + 42)] = @tir.if_then_else(((1 &lt;= (threadIdx.x_1 + rx.outer.outer)) &amp;&amp; ((threadIdx.x_1 + rx.outer.outer) &lt; 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 34)], 0f32, dtype=float32)
+          attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          pad_temp.shared_1[(threadIdx.x_1 + 49)] = @tir.if_then_else(((1 &lt;= (threadIdx.x_1 + rx.outer.outer)) &amp;&amp; ((threadIdx.x_1 + rx.outer.outer) &lt; 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 41)], 0f32, dtype=float32)
+          attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          pad_temp.shared_1[(threadIdx.x_1 + 56)] = 0f32
+          attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          pad_temp.shared_1[(threadIdx.x_1 + 63)] = 0f32
+          attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          pad_temp.shared_1[(threadIdx.x_1 + 70)] = @tir.if_then_else(((1 &lt;= (threadIdx.x_1 + rx.outer.outer)) &amp;&amp; ((threadIdx.x_1 + rx.outer.outer) &lt; 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 48)], 0f32, dtype=float32)
+          attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          pad_temp.shared_1[(threadIdx.x_1 + 77)] = @tir.if_then_else(((1 &lt;= (threadIdx.x_1 + rx.outer.outer)) &amp;&amp; ((threadIdx.x_1 + rx.outer.outer) &lt; 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 55)], 0f32, dtype=float32)
+          attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          pad_temp.shared_1[(threadIdx.x_1 + 84)] = @tir.if_then_else(((1 &lt;= (threadIdx.x_1 + rx.outer.outer)) &amp;&amp; ((threadIdx.x_1 + rx.outer.outer) &lt; 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 62)], 0f32, dtype=float32)
+          attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          pad_temp.shared_1[(threadIdx.x_1 + 91)] = @tir.if_then_else(((1 &lt;= (threadIdx.x_1 + rx.outer.outer)) &amp;&amp; ((threadIdx.x_1 + rx.outer.outer) &lt; 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 69)], 0f32, dtype=float32)
+          attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          pad_temp.shared_1[(threadIdx.x_1 + 98)] = @tir.if_then_else(((1 &lt;= (threadIdx.x_1 + rx.outer.outer)) &amp;&amp; ((threadIdx.x_1 + rx.outer.outer) &lt; 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 76)], 0f32, dtype=float32)
+          attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          pad_temp.shared_1[(threadIdx.x_1 + 105)] = @tir.if_then_else(((1 &lt;= (threadIdx.x_1 + rx.outer.outer)) &amp;&amp; ((threadIdx.x_1 + rx.outer.outer) &lt; 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 83)], 0f32, dtype=float32)
+          attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          pad_temp.shared_1[(threadIdx.x_1 + 112)] = @tir.if_then_else(((1 &lt;= (threadIdx.x_1 + rx.outer.outer)) &amp;&amp; ((threadIdx.x_1 + rx.outer.outer) &lt; 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 90)], 0f32, dtype=float32)
+          attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          pad_temp.shared_1[(threadIdx.x_1 + 119)] = 0f32
+          attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          pad_temp.shared_1[(threadIdx.x_1 + 126)] = 0f32
+          attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          pad_temp.shared_1[(threadIdx.x_1 + 133)] = @tir.if_then_else(((1 &lt;= (threadIdx.x_1 + rx.outer.outer)) &amp;&amp; ((threadIdx.x_1 + rx.outer.outer) &lt; 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 97)], 0f32, dtype=float32)
+          attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          pad_temp.shared_1[(threadIdx.x_1 + 140)] = @tir.if_then_else(((1 &lt;= (threadIdx.x_1 + rx.outer.outer)) &amp;&amp; ((threadIdx.x_1 + rx.outer.outer) &lt; 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 104)], 0f32, dtype=float32)
+          attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          pad_temp.shared_1[(threadIdx.x_1 + 147)] = @tir.if_then_else(((1 &lt;= (threadIdx.x_1 + rx.outer.outer)) &amp;&amp; ((threadIdx.x_1 + rx.outer.outer) &lt; 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 111)], 0f32, dtype=float32)
+          attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          pad_temp.shared_1[(threadIdx.x_1 + 154)] = @tir.if_then_else(((1 &lt;= (threadIdx.x_1 + rx.outer.outer)) &amp;&amp; ((threadIdx.x_1 + rx.outer.outer) &lt; 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 118)], 0f32, dtype=float32)
+          attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          pad_temp.shared_1[(threadIdx.x_1 + 161)] = @tir.if_then_else(((1 &lt;= (threadIdx.x_1 + rx.outer.outer)) &amp;&amp; ((threadIdx.x_1 + rx.outer.outer) &lt; 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 125)], 0f32, dtype=float32)
+          attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          pad_temp.shared_1[(threadIdx.x_1 + 168)] = @tir.if_then_else(((1 &lt;= (threadIdx.x_1 + rx.outer.outer)) &amp;&amp; ((threadIdx.x_1 + rx.outer.outer) &lt; 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 132)], 0f32, dtype=float32)
+          attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          pad_temp.shared_1[(threadIdx.x_1 + 175)] = @tir.if_then_else(((1 &lt;= (threadIdx.x_1 + rx.outer.outer)) &amp;&amp; ((threadIdx.x_1 + rx.outer.outer) &lt; 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 139)], 0f32, dtype=float32)
+          attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          pad_temp.shared_1[(threadIdx.x_1 + 182)] = 0f32
+          attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          pad_temp.shared_1[(threadIdx.x_1 + 189)] = 0f32
+          attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          pad_temp.shared_1[(threadIdx.x_1 + 196)] = @tir.if_then_else(((1 &lt;= (threadIdx.x_1 + rx.outer.outer)) &amp;&amp; ((threadIdx.x_1 + rx.outer.outer) &lt; 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 146)], 0f32, dtype=float32)
+          attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          pad_temp.shared_1[(threadIdx.x_1 + 203)] = @tir.if_then_else(((1 &lt;= (threadIdx.x_1 + rx.outer.outer)) &amp;&amp; ((threadIdx.x_1 + rx.outer.outer) &lt; 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 153)], 0f32, dtype=float32)
+          attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          pad_temp.shared_1[(threadIdx.x_1 + 210)] = @tir.if_then_else(((1 &lt;= (threadIdx.x_1 + rx.outer.outer)) &amp;&amp; ((threadIdx.x_1 + rx.outer.outer) &lt; 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 160)], 0f32, dtype=float32)
+          attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          pad_temp.shared_1[(threadIdx.x_1 + 217)] = @tir.if_then_else(((1 &lt;= (threadIdx.x_1 + rx.outer.outer)) &amp;&amp; ((threadIdx.x_1 + rx.outer.outer) &lt; 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 167)], 0f32, dtype=float32)
+          attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          pad_temp.shared_1[(threadIdx.x_1 + 224)] = @tir.if_then_else(((1 &lt;= (threadIdx.x_1 + rx.outer.outer)) &amp;&amp; ((threadIdx.x_1 + rx.outer.outer) &lt; 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 174)], 0f32, dtype=float32)
+          attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          pad_temp.shared_1[(threadIdx.x_1 + 231)] = @tir.if_then_else(((1 &lt;= (threadIdx.x_1 + rx.outer.outer)) &amp;&amp; ((threadIdx.x_1 + rx.outer.outer) &lt; 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 181)], 0f32, dtype=float32)
+          attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          pad_temp.shared_1[(threadIdx.x_1 + 238)] = @tir.if_then_else(((1 &lt;= (threadIdx.x_1 + rx.outer.outer)) &amp;&amp; ((threadIdx.x_1 + rx.outer.outer) &lt; 8)), data[((((rc.outer.outer*196) + rx.outer.outer) + threadIdx.x_1) + 188)], 0f32, dtype=float32)
+          attr [IterVar(threadIdx.x_1, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          pad_temp.shared_1[(threadIdx.x_1 + 245)] = 0f32
+          attr [IterVar(threadIdx.x_2: int32, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          kernel.shared_1: Buffer(kernel.shared, float32, [48], [], scope=&quot;shared&quot;)[threadIdx.x_2] = kernel[((((blockIdx.x*18432) + cse_var_1) + (threadIdx.x_2*3)) + rx.outer.outer)]
+          attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          kernel.shared_1[(threadIdx.x_2 + 7)] = kernel[((((((blockIdx.x*18432) + (floordiv((threadIdx.x_2 + 7), 12)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 7), 12), 3)*9)) + (floormod((threadIdx.x_2 + 1), 3)*3)) + rx.outer.outer)]
+          attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          kernel.shared_1[(threadIdx.x_2 + 14)] = kernel[(((((blockIdx.x*18432) + (floordiv((threadIdx.x_2 + 14), 12)*4608)) + cse_var_1) + (floormod((threadIdx.x_2 + 2), 12)*3)) + rx.outer.outer)]
+          attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          kernel.shared_1[(threadIdx.x_2 + 21)] = kernel[((((((blockIdx.x*18432) + (floordiv((threadIdx.x_2 + 21), 12)*4608)) + cse_var_1) + (floormod((floordiv(threadIdx.x_2, 3) + 3), 4)*9)) + (floormod(threadIdx.x_2, 3)*3)) + rx.outer.outer)]
+          attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          kernel.shared_1[(threadIdx.x_2 + 28)] = kernel[((((((blockIdx.x*18432) + (floordiv((threadIdx.x_2 + 28), 12)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 4), 12), 3)*9)) + (floormod((threadIdx.x_2 + 1), 3)*3)) + rx.outer.outer)]
+          attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          kernel.shared_1[(threadIdx.x_2 + 35)] = kernel[((((((blockIdx.x*18432) + (floordiv((threadIdx.x_2 + 35), 12)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 11), 12), 3)*9)) + (floormod((threadIdx.x_2 + 2), 3)*3)) + rx.outer.outer)]
+          attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 7;
+          if @tir.likely((threadIdx.x_2 &lt; 6), dtype=bool) {
+            kernel.shared_1[(threadIdx.x_2 + 42)] = kernel[((((((blockIdx.x*18432) + (floordiv((threadIdx.x_2 + 42), 12)*4608)) + cse_var_1) + ((floordiv(threadIdx.x_2, 3) + 2)*9)) + (floormod(threadIdx.x_2, 3)*3)) + rx.outer.outer)]
+          }
+          conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[threadIdx.x]*kernel.shared_1[0]))
+          conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(threadIdx.x + 7)]*kernel.shared_1[0]))
+          conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(threadIdx.x + 14)]*kernel.shared_1[0]))
+          conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(threadIdx.x + 21)]*kernel.shared_1[0]))
+          conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(threadIdx.x + 28)]*kernel.shared_1[0]))
+          conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(threadIdx.x + 35)]*kernel.shared_1[0]))
+          conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(threadIdx.x + 42)]*kernel.shared_1[0]))
+          conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[threadIdx.x]*kernel.shared_1[24]))
+          conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(threadIdx.x + 7)]*kernel.shared_1[24]))
+          conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(threadIdx.x + 14)]*kernel.shared_1[24]))
+          conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(threadIdx.x + 21)]*kernel.shared_1[24]))
+          conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(threadIdx.x + 28)]*kernel.shared_1[24]))
+          conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(threadIdx.x + 35)]*kernel.shared_1[24]))
+          conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(threadIdx.x + 42)]*kernel.shared_1[24]))
+          conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[threadIdx.x]*kernel.shared_1[12]))
+          conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(threadIdx.x + 7)]*kernel.shared_1[12]))
+          conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(threadIdx.x + 14)]*kernel.shared_1[12]))
+          conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(threadIdx.x + 21)]*kernel.shared_1[12]))
+          conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(threadIdx.x + 28)]*kernel.shared_1[12]))
+          conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(threadIdx.x + 35)]*kernel.shared_1[12]))
+          conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(threadIdx.x + 42)]*kernel.shared_1[12]))
+          conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[threadIdx.x]*kernel.shared_1[36]))
+          conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(threadIdx.x + 7)]*kernel.shared_1[36]))
+          conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(threadIdx.x + 14)]*kernel.shared_1[36]))
+          conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(threadIdx.x + 21)]*kernel.shared_1[36]))
+          conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(threadIdx.x + 28)]*kernel.shared_1[36]))
+          conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(threadIdx.x + 35)]*kernel.shared_1[36]))
+          conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(threadIdx.x + 42)]*kernel.shared_1[36]))
+          conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(threadIdx.x + 7)]*kernel.shared_1[1]))
+          conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(threadIdx.x + 14)]*kernel.shared_1[1]))
+          conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(threadIdx.x + 21)]*kernel.shared_1[1]))
+          conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(threadIdx.x + 28)]*kernel.shared_1[1]))
+          conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(threadIdx.x + 35)]*kernel.shared_1[1]))
+          conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(threadIdx.x + 42)]*kernel.shared_1[1]))
+          conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(threadIdx.x + 49)]*kernel.shared_1[1]))
+          conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(threadIdx.x + 7)]*kernel.shared_1[25]))
+          conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(threadIdx.x + 14)]*kernel.shared_1[25]))
+          conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(threadIdx.x + 21)]*kernel.shared_1[25]))
+          conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(threadIdx.x + 28)]*kernel.shared_1[25]))
+          conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(threadIdx.x + 35)]*kernel.shared_1[25]))
+          conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(threadIdx.x + 42)]*kernel.shared_1[25]))
+          conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(threadIdx.x + 49)]*kernel.shared_1[25]))
+          conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(threadIdx.x + 7)]*kernel.shared_1[13]))
+          conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(threadIdx.x + 14)]*kernel.shared_1[13]))
+          conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(threadIdx.x + 21)]*kernel.shared_1[13]))
+          conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(threadIdx.x + 28)]*kernel.shared_1[13]))
+          conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(threadIdx.x + 35)]*kernel.shared_1[13]))
+          conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(threadIdx.x + 42)]*kernel.shared_1[13]))
+          conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(threadIdx.x + 49)]*kernel.shared_1[13]))
+          conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(threadIdx.x + 7)]*kernel.shared_1[37]))
+          conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(threadIdx.x + 14)]*kernel.shared_1[37]))
+          conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(threadIdx.x + 21)]*kernel.shared_1[37]))
+          conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(threadIdx.x + 28)]*kernel.shared_1[37]))
+          conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(threadIdx.x + 35)]*kernel.shared_1[37]))
+          conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(threadIdx.x + 42)]*kernel.shared_1[37]))
+          conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(threadIdx.x + 49)]*kernel.shared_1[37]))
+          conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(threadIdx.x + 14)]*kernel.shared_1[2]))
+          conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(threadIdx.x + 21)]*kernel.shared_1[2]))
+          conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(threadIdx.x + 28)]*kernel.shared_1[2]))
+          conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(threadIdx.x + 35)]*kernel.shared_1[2]))
+          conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(threadIdx.x + 42)]*kernel.shared_1[2]))
+          conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(threadIdx.x + 49)]*kernel.shared_1[2]))
+          conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(threadIdx.x + 56)]*kernel.shared_1[2]))
+          conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(threadIdx.x + 14)]*kernel.shared_1[26]))
+          conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(threadIdx.x + 21)]*kernel.shared_1[26]))
+          conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(threadIdx.x + 28)]*kernel.shared_1[26]))
+          conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(threadIdx.x + 35)]*kernel.shared_1[26]))
+          conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(threadIdx.x + 42)]*kernel.shared_1[26]))
+          conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(threadIdx.x + 49)]*kernel.shared_1[26]))
+          conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(threadIdx.x + 56)]*kernel.shared_1[26]))
+          conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(threadIdx.x + 14)]*kernel.shared_1[14]))
+          conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(threadIdx.x + 21)]*kernel.shared_1[14]))
+          conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(threadIdx.x + 28)]*kernel.shared_1[14]))
+          conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(threadIdx.x + 35)]*kernel.shared_1[14]))
+          conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(threadIdx.x + 42)]*kernel.shared_1[14]))
+          conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(threadIdx.x + 49)]*kernel.shared_1[14]))
+          conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(threadIdx.x + 56)]*kernel.shared_1[14]))
+          conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(threadIdx.x + 14)]*kernel.shared_1[38]))
+          conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(threadIdx.x + 21)]*kernel.shared_1[38]))
+          conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(threadIdx.x + 28)]*kernel.shared_1[38]))
+          conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(threadIdx.x + 35)]*kernel.shared_1[38]))
+          conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(threadIdx.x + 42)]*kernel.shared_1[38]))
+          conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(threadIdx.x + 49)]*kernel.shared_1[38]))
+          conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(threadIdx.x + 56)]*kernel.shared_1[38]))
+          conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(threadIdx.x + 63)]*kernel.shared_1[3]))
+          conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(threadIdx.x + 70)]*kernel.shared_1[3]))
+          conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(threadIdx.x + 77)]*kernel.shared_1[3]))
+          conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(threadIdx.x + 84)]*kernel.shared_1[3]))
+          conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(threadIdx.x + 91)]*kernel.shared_1[3]))
+          conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(threadIdx.x + 98)]*kernel.shared_1[3]))
+          conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(threadIdx.x + 105)]*kernel.shared_1[3]))
+          conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(threadIdx.x + 63)]*kernel.shared_1[27]))
+          conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(threadIdx.x + 70)]*kernel.shared_1[27]))
+          conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(threadIdx.x + 77)]*kernel.shared_1[27]))
+          conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(threadIdx.x + 84)]*kernel.shared_1[27]))
+          conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(threadIdx.x + 91)]*kernel.shared_1[27]))
+          conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(threadIdx.x + 98)]*kernel.shared_1[27]))
+          conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(threadIdx.x + 105)]*kernel.shared_1[27]))
+          conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(threadIdx.x + 63)]*kernel.shared_1[15]))
+          conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(threadIdx.x + 70)]*kernel.shared_1[15]))
+          conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(threadIdx.x + 77)]*kernel.shared_1[15]))
+          conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(threadIdx.x + 84)]*kernel.shared_1[15]))
+          conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(threadIdx.x + 91)]*kernel.shared_1[15]))
+          conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(threadIdx.x + 98)]*kernel.shared_1[15]))
+          conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(threadIdx.x + 105)]*kernel.shared_1[15]))
+          conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(threadIdx.x + 63)]*kernel.shared_1[39]))
+          conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(threadIdx.x + 70)]*kernel.shared_1[39]))
+          conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(threadIdx.x + 77)]*kernel.shared_1[39]))
+          conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(threadIdx.x + 84)]*kernel.shared_1[39]))
+          conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(threadIdx.x + 91)]*kernel.shared_1[39]))
+          conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(threadIdx.x + 98)]*kernel.shared_1[39]))
+          conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(threadIdx.x + 105)]*kernel.shared_1[39]))
+          conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(threadIdx.x + 70)]*kernel.shared_1[4]))
+          conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(threadIdx.x + 77)]*kernel.shared_1[4]))
+          conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(threadIdx.x + 84)]*kernel.shared_1[4]))
+          conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(threadIdx.x + 91)]*kernel.shared_1[4]))
+          conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(threadIdx.x + 98)]*kernel.shared_1[4]))
+          conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(threadIdx.x + 105)]*kernel.shared_1[4]))
+          conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(threadIdx.x + 112)]*kernel.shared_1[4]))
+          conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(threadIdx.x + 70)]*kernel.shared_1[28]))
+          conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(threadIdx.x + 77)]*kernel.shared_1[28]))
+          conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(threadIdx.x + 84)]*kernel.shared_1[28]))
+          conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(threadIdx.x + 91)]*kernel.shared_1[28]))
+          conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(threadIdx.x + 98)]*kernel.shared_1[28]))
+          conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(threadIdx.x + 105)]*kernel.shared_1[28]))
+          conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(threadIdx.x + 112)]*kernel.shared_1[28]))
+          conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(threadIdx.x + 70)]*kernel.shared_1[16]))
+          conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(threadIdx.x + 77)]*kernel.shared_1[16]))
+          conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(threadIdx.x + 84)]*kernel.shared_1[16]))
+          conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(threadIdx.x + 91)]*kernel.shared_1[16]))
+          conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(threadIdx.x + 98)]*kernel.shared_1[16]))
+          conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(threadIdx.x + 105)]*kernel.shared_1[16]))
+          conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(threadIdx.x + 112)]*kernel.shared_1[16]))
+          conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(threadIdx.x + 70)]*kernel.shared_1[40]))
+          conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(threadIdx.x + 77)]*kernel.shared_1[40]))
+          conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(threadIdx.x + 84)]*kernel.shared_1[40]))
+          conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(threadIdx.x + 91)]*kernel.shared_1[40]))
+          conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(threadIdx.x + 98)]*kernel.shared_1[40]))
+          conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(threadIdx.x + 105)]*kernel.shared_1[40]))
+          conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(threadIdx.x + 112)]*kernel.shared_1[40]))
+          conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(threadIdx.x + 77)]*kernel.shared_1[5]))
+          conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(threadIdx.x + 84)]*kernel.shared_1[5]))
+          conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(threadIdx.x + 91)]*kernel.shared_1[5]))
+          conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(threadIdx.x + 98)]*kernel.shared_1[5]))
+          conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(threadIdx.x + 105)]*kernel.shared_1[5]))
+          conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(threadIdx.x + 112)]*kernel.shared_1[5]))
+          conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(threadIdx.x + 119)]*kernel.shared_1[5]))
+          conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(threadIdx.x + 77)]*kernel.shared_1[29]))
+          conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(threadIdx.x + 84)]*kernel.shared_1[29]))
+          conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(threadIdx.x + 91)]*kernel.shared_1[29]))
+          conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(threadIdx.x + 98)]*kernel.shared_1[29]))
+          conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(threadIdx.x + 105)]*kernel.shared_1[29]))
+          conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(threadIdx.x + 112)]*kernel.shared_1[29]))
+          conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(threadIdx.x + 119)]*kernel.shared_1[29]))
+          conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(threadIdx.x + 77)]*kernel.shared_1[17]))
+          conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(threadIdx.x + 84)]*kernel.shared_1[17]))
+          conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(threadIdx.x + 91)]*kernel.shared_1[17]))
+          conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(threadIdx.x + 98)]*kernel.shared_1[17]))
+          conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(threadIdx.x + 105)]*kernel.shared_1[17]))
+          conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(threadIdx.x + 112)]*kernel.shared_1[17]))
+          conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(threadIdx.x + 119)]*kernel.shared_1[17]))
+          conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(threadIdx.x + 77)]*kernel.shared_1[41]))
+          conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(threadIdx.x + 84)]*kernel.shared_1[41]))
+          conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(threadIdx.x + 91)]*kernel.shared_1[41]))
+          conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(threadIdx.x + 98)]*kernel.shared_1[41]))
+          conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(threadIdx.x + 105)]*kernel.shared_1[41]))
+          conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(threadIdx.x + 112)]*kernel.shared_1[41]))
+          conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(threadIdx.x + 119)]*kernel.shared_1[41]))
+          conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(threadIdx.x + 126)]*kernel.shared_1[6]))
+          conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(threadIdx.x + 133)]*kernel.shared_1[6]))
+          conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(threadIdx.x + 140)]*kernel.shared_1[6]))
+          conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(threadIdx.x + 147)]*kernel.shared_1[6]))
+          conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(threadIdx.x + 154)]*kernel.shared_1[6]))
+          conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(threadIdx.x + 161)]*kernel.shared_1[6]))
+          conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(threadIdx.x + 168)]*kernel.shared_1[6]))
+          conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(threadIdx.x + 126)]*kernel.shared_1[30]))
+          conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(threadIdx.x + 133)]*kernel.shared_1[30]))
+          conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(threadIdx.x + 140)]*kernel.shared_1[30]))
+          conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(threadIdx.x + 147)]*kernel.shared_1[30]))
+          conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(threadIdx.x + 154)]*kernel.shared_1[30]))
+          conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(threadIdx.x + 161)]*kernel.shared_1[30]))
+          conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(threadIdx.x + 168)]*kernel.shared_1[30]))
+          conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(threadIdx.x + 126)]*kernel.shared_1[18]))
+          conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(threadIdx.x + 133)]*kernel.shared_1[18]))
+          conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(threadIdx.x + 140)]*kernel.shared_1[18]))
+          conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(threadIdx.x + 147)]*kernel.shared_1[18]))
+          conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(threadIdx.x + 154)]*kernel.shared_1[18]))
+          conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(threadIdx.x + 161)]*kernel.shared_1[18]))
+          conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(threadIdx.x + 168)]*kernel.shared_1[18]))
+          conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(threadIdx.x + 126)]*kernel.shared_1[42]))
+          conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(threadIdx.x + 133)]*kernel.shared_1[42]))
+          conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(threadIdx.x + 140)]*kernel.shared_1[42]))
+          conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(threadIdx.x + 147)]*kernel.shared_1[42]))
+          conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(threadIdx.x + 154)]*kernel.shared_1[42]))
+          conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(threadIdx.x + 161)]*kernel.shared_1[42]))
+          conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(threadIdx.x + 168)]*kernel.shared_1[42]))
+          conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(threadIdx.x + 133)]*kernel.shared_1[7]))
+          conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(threadIdx.x + 140)]*kernel.shared_1[7]))
+          conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(threadIdx.x + 147)]*kernel.shared_1[7]))
+          conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(threadIdx.x + 154)]*kernel.shared_1[7]))
+          conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(threadIdx.x + 161)]*kernel.shared_1[7]))
+          conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(threadIdx.x + 168)]*kernel.shared_1[7]))
+          conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(threadIdx.x + 175)]*kernel.shared_1[7]))
+          conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(threadIdx.x + 133)]*kernel.shared_1[31]))
+          conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(threadIdx.x + 140)]*kernel.shared_1[31]))
+          conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(threadIdx.x + 147)]*kernel.shared_1[31]))
+          conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(threadIdx.x + 154)]*kernel.shared_1[31]))
+          conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(threadIdx.x + 161)]*kernel.shared_1[31]))
+          conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(threadIdx.x + 168)]*kernel.shared_1[31]))
+          conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(threadIdx.x + 175)]*kernel.shared_1[31]))
+          conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(threadIdx.x + 133)]*kernel.shared_1[19]))
+          conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(threadIdx.x + 140)]*kernel.shared_1[19]))
+          conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(threadIdx.x + 147)]*kernel.shared_1[19]))
+          conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(threadIdx.x + 154)]*kernel.shared_1[19]))
+          conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(threadIdx.x + 161)]*kernel.shared_1[19]))
+          conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(threadIdx.x + 168)]*kernel.shared_1[19]))
+          conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(threadIdx.x + 175)]*kernel.shared_1[19]))
+          conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(threadIdx.x + 133)]*kernel.shared_1[43]))
+          conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(threadIdx.x + 140)]*kernel.shared_1[43]))
+          conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(threadIdx.x + 147)]*kernel.shared_1[43]))
+          conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(threadIdx.x + 154)]*kernel.shared_1[43]))
+          conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(threadIdx.x + 161)]*kernel.shared_1[43]))
+          conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(threadIdx.x + 168)]*kernel.shared_1[43]))
+          conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(threadIdx.x + 175)]*kernel.shared_1[43]))
+          conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(threadIdx.x + 140)]*kernel.shared_1[8]))
+          conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(threadIdx.x + 147)]*kernel.shared_1[8]))
+          conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(threadIdx.x + 154)]*kernel.shared_1[8]))
+          conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(threadIdx.x + 161)]*kernel.shared_1[8]))
+          conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(threadIdx.x + 168)]*kernel.shared_1[8]))
+          conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(threadIdx.x + 175)]*kernel.shared_1[8]))
+          conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(threadIdx.x + 182)]*kernel.shared_1[8]))
+          conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(threadIdx.x + 140)]*kernel.shared_1[32]))
+          conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(threadIdx.x + 147)]*kernel.shared_1[32]))
+          conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(threadIdx.x + 154)]*kernel.shared_1[32]))
+          conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(threadIdx.x + 161)]*kernel.shared_1[32]))
+          conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(threadIdx.x + 168)]*kernel.shared_1[32]))
+          conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(threadIdx.x + 175)]*kernel.shared_1[32]))
+          conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(threadIdx.x + 182)]*kernel.shared_1[32]))
+          conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(threadIdx.x + 140)]*kernel.shared_1[20]))
+          conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(threadIdx.x + 147)]*kernel.shared_1[20]))
+          conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(threadIdx.x + 154)]*kernel.shared_1[20]))
+          conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(threadIdx.x + 161)]*kernel.shared_1[20]))
+          conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(threadIdx.x + 168)]*kernel.shared_1[20]))
+          conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(threadIdx.x + 175)]*kernel.shared_1[20]))
+          conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(threadIdx.x + 182)]*kernel.shared_1[20]))
+          conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(threadIdx.x + 140)]*kernel.shared_1[44]))
+          conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(threadIdx.x + 147)]*kernel.shared_1[44]))
+          conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(threadIdx.x + 154)]*kernel.shared_1[44]))
+          conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(threadIdx.x + 161)]*kernel.shared_1[44]))
+          conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(threadIdx.x + 168)]*kernel.shared_1[44]))
+          conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(threadIdx.x + 175)]*kernel.shared_1[44]))
+          conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(threadIdx.x + 182)]*kernel.shared_1[44]))
+          conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(threadIdx.x + 189)]*kernel.shared_1[9]))
+          conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(threadIdx.x + 196)]*kernel.shared_1[9]))
+          conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(threadIdx.x + 203)]*kernel.shared_1[9]))
+          conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(threadIdx.x + 210)]*kernel.shared_1[9]))
+          conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(threadIdx.x + 217)]*kernel.shared_1[9]))
+          conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(threadIdx.x + 224)]*kernel.shared_1[9]))
+          conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(threadIdx.x + 231)]*kernel.shared_1[9]))
+          conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(threadIdx.x + 189)]*kernel.shared_1[33]))
+          conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(threadIdx.x + 196)]*kernel.shared_1[33]))
+          conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(threadIdx.x + 203)]*kernel.shared_1[33]))
+          conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(threadIdx.x + 210)]*kernel.shared_1[33]))
+          conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(threadIdx.x + 217)]*kernel.shared_1[33]))
+          conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(threadIdx.x + 224)]*kernel.shared_1[33]))
+          conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(threadIdx.x + 231)]*kernel.shared_1[33]))
+          conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(threadIdx.x + 189)]*kernel.shared_1[21]))
+          conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(threadIdx.x + 196)]*kernel.shared_1[21]))
+          conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(threadIdx.x + 203)]*kernel.shared_1[21]))
+          conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(threadIdx.x + 210)]*kernel.shared_1[21]))
+          conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(threadIdx.x + 217)]*kernel.shared_1[21]))
+          conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(threadIdx.x + 224)]*kernel.shared_1[21]))
+          conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(threadIdx.x + 231)]*kernel.shared_1[21]))
+          conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(threadIdx.x + 189)]*kernel.shared_1[45]))
+          conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(threadIdx.x + 196)]*kernel.shared_1[45]))
+          conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(threadIdx.x + 203)]*kernel.shared_1[45]))
+          conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(threadIdx.x + 210)]*kernel.shared_1[45]))
+          conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(threadIdx.x + 217)]*kernel.shared_1[45]))
+          conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(threadIdx.x + 224)]*kernel.shared_1[45]))
+          conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(threadIdx.x + 231)]*kernel.shared_1[45]))
+          conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(threadIdx.x + 196)]*kernel.shared_1[10]))
+          conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(threadIdx.x + 203)]*kernel.shared_1[10]))
+          conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(threadIdx.x + 210)]*kernel.shared_1[10]))
+          conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(threadIdx.x + 217)]*kernel.shared_1[10]))
+          conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(threadIdx.x + 224)]*kernel.shared_1[10]))
+          conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(threadIdx.x + 231)]*kernel.shared_1[10]))
+          conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(threadIdx.x + 238)]*kernel.shared_1[10]))
+          conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(threadIdx.x + 196)]*kernel.shared_1[34]))
+          conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(threadIdx.x + 203)]*kernel.shared_1[34]))
+          conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(threadIdx.x + 210)]*kernel.shared_1[34]))
+          conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(threadIdx.x + 217)]*kernel.shared_1[34]))
+          conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(threadIdx.x + 224)]*kernel.shared_1[34]))
+          conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(threadIdx.x + 231)]*kernel.shared_1[34]))
+          conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(threadIdx.x + 238)]*kernel.shared_1[34]))
+          conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(threadIdx.x + 196)]*kernel.shared_1[22]))
+          conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(threadIdx.x + 203)]*kernel.shared_1[22]))
+          conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(threadIdx.x + 210)]*kernel.shared_1[22]))
+          conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(threadIdx.x + 217)]*kernel.shared_1[22]))
+          conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(threadIdx.x + 224)]*kernel.shared_1[22]))
+          conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(threadIdx.x + 231)]*kernel.shared_1[22]))
+          conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(threadIdx.x + 238)]*kernel.shared_1[22]))
+          conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(threadIdx.x + 196)]*kernel.shared_1[46]))
+          conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(threadIdx.x + 203)]*kernel.shared_1[46]))
+          conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(threadIdx.x + 210)]*kernel.shared_1[46]))
+          conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(threadIdx.x + 217)]*kernel.shared_1[46]))
+          conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(threadIdx.x + 224)]*kernel.shared_1[46]))
+          conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(threadIdx.x + 231)]*kernel.shared_1[46]))
+          conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(threadIdx.x + 238)]*kernel.shared_1[46]))
+          conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(threadIdx.x + 203)]*kernel.shared_1[11]))
+          conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(threadIdx.x + 210)]*kernel.shared_1[11]))
+          conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(threadIdx.x + 217)]*kernel.shared_1[11]))
+          conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(threadIdx.x + 224)]*kernel.shared_1[11]))
+          conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[(threadIdx.x + 231)]*kernel.shared_1[11]))
+          conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[(threadIdx.x + 238)]*kernel.shared_1[11]))
+          conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[(threadIdx.x + 245)]*kernel.shared_1[11]))
+          conv2d_nchw_1[14] = (conv2d_nchw_1[14] + (pad_temp.shared_1[(threadIdx.x + 203)]*kernel.shared_1[35]))
+          conv2d_nchw_1[16] = (conv2d_nchw_1[16] + (pad_temp.shared_1[(threadIdx.x + 210)]*kernel.shared_1[35]))
+          conv2d_nchw_1[18] = (conv2d_nchw_1[18] + (pad_temp.shared_1[(threadIdx.x + 217)]*kernel.shared_1[35]))
+          conv2d_nchw_1[20] = (conv2d_nchw_1[20] + (pad_temp.shared_1[(threadIdx.x + 224)]*kernel.shared_1[35]))
+          conv2d_nchw_1[22] = (conv2d_nchw_1[22] + (pad_temp.shared_1[(threadIdx.x + 231)]*kernel.shared_1[35]))
+          conv2d_nchw_1[24] = (conv2d_nchw_1[24] + (pad_temp.shared_1[(threadIdx.x + 238)]*kernel.shared_1[35]))
+          conv2d_nchw_1[26] = (conv2d_nchw_1[26] + (pad_temp.shared_1[(threadIdx.x + 245)]*kernel.shared_1[35]))
+          conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(threadIdx.x + 203)]*kernel.shared_1[23]))
+          conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(threadIdx.x + 210)]*kernel.shared_1[23]))
+          conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(threadIdx.x + 217)]*kernel.shared_1[23]))
+          conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(threadIdx.x + 224)]*kernel.shared_1[23]))
+          conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[(threadIdx.x + 231)]*kernel.shared_1[23]))
+          conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[(threadIdx.x + 238)]*kernel.shared_1[23]))
+          conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[(threadIdx.x + 245)]*kernel.shared_1[23]))
+          conv2d_nchw_1[15] = (conv2d_nchw_1[15] + (pad_temp.shared_1[(threadIdx.x + 203)]*kernel.shared_1[47]))
+          conv2d_nchw_1[17] = (conv2d_nchw_1[17] + (pad_temp.shared_1[(threadIdx.x + 210)]*kernel.shared_1[47]))
+          conv2d_nchw_1[19] = (conv2d_nchw_1[19] + (pad_temp.shared_1[(threadIdx.x + 217)]*kernel.shared_1[47]))
+          conv2d_nchw_1[21] = (conv2d_nchw_1[21] + (pad_temp.shared_1[(threadIdx.x + 224)]*kernel.shared_1[47]))
+          conv2d_nchw_1[23] = (conv2d_nchw_1[23] + (pad_temp.shared_1[(threadIdx.x + 231)]*kernel.shared_1[47]))
+          conv2d_nchw_1[25] = (conv2d_nchw_1[25] + (pad_temp.shared_1[(threadIdx.x + 238)]*kernel.shared_1[47]))
+          conv2d_nchw_1[27] = (conv2d_nchw_1[27] + (pad_temp.shared_1[(threadIdx.x + 245)]*kernel.shared_1[47]))
         }
-        attr [IterVar(threadIdx.x_2: int32, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1: Buffer(kernel.shared, float32, [1152], [], scope=&quot;shared&quot;)[threadIdx.x_2] = kernel[(((floordiv(blockIdx.x, 7)*147456) + cse_var_1) + threadIdx.x_2)]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 16)] = kernel[((((floordiv(blockIdx.x, 7)*147456) + cse_var_1) + (floordiv((threadIdx.x_2 + 16), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 32)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 32), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 32), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 48)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 48), 36)*4608)) + cse_var_1) + ((floordiv(threadIdx.x_2, 3) + 4)*3)) + floormod(threadIdx.x_2, 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 64)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 64), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 28), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 80)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 80), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 8), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 96)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 96), 36)*4608)) + cse_var_1) + (floormod((floordiv(threadIdx.x_2, 3) + 8), 12)*3)) + floormod(threadIdx.x_2, 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 112)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 112), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 4), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 128)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 128), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 20), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 144)] = kernel[((((floordiv(blockIdx.x, 7)*147456) + cse_var_1) + threadIdx.x_2) + 18432)]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 160)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 160), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 16), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 176)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 176), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 32), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 192)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 192), 36)*4608)) + cse_var_1) + ((floordiv(threadIdx.x_2, 3) + 4)*3)) + floormod(threadIdx.x_2, 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 208)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 208), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 28), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 224)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 224), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 8), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 240)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 240), 36)*4608)) + cse_var_1) + (floormod((floordiv(threadIdx.x_2, 3) + 8), 12)*3)) + floormod(threadIdx.x_2, 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 256)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 256), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 4), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 272)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 272), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 20), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 288)] = kernel[((((floordiv(blockIdx.x, 7)*147456) + cse_var_1) + threadIdx.x_2) + 36864)]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 304)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 304), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 16), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 320)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 320), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 32), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 336)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 336), 36)*4608)) + cse_var_1) + ((floordiv(threadIdx.x_2, 3) + 4)*3)) + floormod(threadIdx.x_2, 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 352)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 352), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 28), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 368)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 368), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 8), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 384)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 384), 36)*4608)) + cse_var_1) + (floormod((floordiv(threadIdx.x_2, 3) + 8), 12)*3)) + floormod(threadIdx.x_2, 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 400)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 400), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 4), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 416)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 416), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 20), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 432)] = kernel[((((floordiv(blockIdx.x, 7)*147456) + cse_var_1) + threadIdx.x_2) + 55296)]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 448)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 448), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 16), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 464)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 464), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 32), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 480)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 480), 36)*4608)) + cse_var_1) + ((floordiv(threadIdx.x_2, 3) + 4)*3)) + floormod(threadIdx.x_2, 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 496)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 496), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 28), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 512)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 512), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 8), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 528)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 528), 36)*4608)) + cse_var_1) + (floormod((floordiv(threadIdx.x_2, 3) + 8), 12)*3)) + floormod(threadIdx.x_2, 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 544)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 544), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 4), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 560)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 560), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 20), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 576)] = kernel[((((floordiv(blockIdx.x, 7)*147456) + cse_var_1) + threadIdx.x_2) + 73728)]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 592)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 592), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 16), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 608)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 608), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 32), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 624)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 624), 36)*4608)) + cse_var_1) + ((floordiv(threadIdx.x_2, 3) + 4)*3)) + floormod(threadIdx.x_2, 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 640)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 640), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 28), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 656)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 656), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 8), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 672)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 672), 36)*4608)) + cse_var_1) + (floormod((floordiv(threadIdx.x_2, 3) + 8), 12)*3)) + floormod(threadIdx.x_2, 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 688)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 688), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 4), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 704)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 704), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 20), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 720)] = kernel[((((floordiv(blockIdx.x, 7)*147456) + cse_var_1) + threadIdx.x_2) + 92160)]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 736)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 736), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 16), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 752)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 752), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 32), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 768)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 768), 36)*4608)) + cse_var_1) + ((floordiv(threadIdx.x_2, 3) + 4)*3)) + floormod(threadIdx.x_2, 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 784)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 784), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 28), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 800)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 800), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 8), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 816)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 816), 36)*4608)) + cse_var_1) + (floormod((floordiv(threadIdx.x_2, 3) + 8), 12)*3)) + floormod(threadIdx.x_2, 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 832)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 832), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 4), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 848)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 848), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 20), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 864)] = kernel[((((floordiv(blockIdx.x, 7)*147456) + cse_var_1) + threadIdx.x_2) + 110592)]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 880)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 880), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 16), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 896)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 896), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 32), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 912)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 912), 36)*4608)) + cse_var_1) + ((floordiv(threadIdx.x_2, 3) + 4)*3)) + floormod(threadIdx.x_2, 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 928)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 928), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 28), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 944)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 944), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 8), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 960)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 960), 36)*4608)) + cse_var_1) + (floormod((floordiv(threadIdx.x_2, 3) + 8), 12)*3)) + floormod(threadIdx.x_2, 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 976)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 976), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 4), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 992)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 992), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 20), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 1008)] = kernel[((((floordiv(blockIdx.x, 7)*147456) + cse_var_1) + threadIdx.x_2) + 129024)]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 1024)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 1024), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 16), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 1040)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 1040), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 32), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 1056)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 1056), 36)*4608)) + cse_var_1) + ((floordiv(threadIdx.x_2, 3) + 4)*3)) + floormod(threadIdx.x_2, 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 1072)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 1072), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 28), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 1088)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 1088), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 8), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 1104)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 1104), 36)*4608)) + cse_var_1) + (floormod((floordiv(threadIdx.x_2, 3) + 8), 12)*3)) + floormod(threadIdx.x_2, 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 1120)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 1120), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 4), 36), 3)*3)) + floormod((threadIdx.x_2 + 1), 3))]
-        attr [IterVar(threadIdx.x_2, (nullptr), &quot;ThreadIndex&quot;, &quot;threadIdx.x&quot;)] &quot;thread_extent&quot; = 16;
-        kernel.shared_1[(threadIdx.x_2 + 1136)] = kernel[(((((floordiv(blockIdx.x, 7)*147456) + (floordiv((threadIdx.x_2 + 1136), 36)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 20), 36), 3)*3)) + floormod((threadIdx.x_2 + 2), 3))]
-        conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[0]*kernel.shared_1[(threadIdx.x*72)]))
-        conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[1]*kernel.shared_1[((threadIdx.x*72) + 1)]))
-        conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[2]*kernel.shared_1[((threadIdx.x*72) + 2)]))
-        conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[3]*kernel.shared_1[((threadIdx.x*72) + 3)]))
-        conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[4]*kernel.shared_1[((threadIdx.x*72) + 4)]))
-        conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[5]*kernel.shared_1[((threadIdx.x*72) + 5)]))
-        conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[6]*kernel.shared_1[((threadIdx.x*72) + 6)]))
-        conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[7]*kernel.shared_1[((threadIdx.x*72) + 7)]))
-        conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[8]*kernel.shared_1[((threadIdx.x*72) + 8)]))
-        conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[3]*kernel.shared_1[(threadIdx.x*72)]))
-        conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[4]*kernel.shared_1[((threadIdx.x*72) + 1)]))
-        conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[5]*kernel.shared_1[((threadIdx.x*72) + 2)]))
-        conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[6]*kernel.shared_1[((threadIdx.x*72) + 3)]))
-        conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[7]*kernel.shared_1[((threadIdx.x*72) + 4)]))
-        conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[8]*kernel.shared_1[((threadIdx.x*72) + 5)]))
-        conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[9]*kernel.shared_1[((threadIdx.x*72) + 6)]))
-        conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[10]*kernel.shared_1[((threadIdx.x*72) + 7)]))
-        conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[11]*kernel.shared_1[((threadIdx.x*72) + 8)]))
-        conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[6]*kernel.shared_1[(threadIdx.x*72)]))
-        conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[7]*kernel.shared_1[((threadIdx.x*72) + 1)]))
-        conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[8]*kernel.shared_1[((threadIdx.x*72) + 2)]))
-        conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[9]*kernel.shared_1[((threadIdx.x*72) + 3)]))
-        conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[10]*kernel.shared_1[((threadIdx.x*72) + 4)]))
-        conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[11]*kernel.shared_1[((threadIdx.x*72) + 5)]))
-        conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[12]*kernel.shared_1[((threadIdx.x*72) + 6)]))
-        conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[13]*kernel.shared_1[((threadIdx.x*72) + 7)]))
-        conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[14]*kernel.shared_1[((threadIdx.x*72) + 8)]))
-        conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[9]*kernel.shared_1[(threadIdx.x*72)]))
-        conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[10]*kernel.shared_1[((threadIdx.x*72) + 1)]))
-        conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[11]*kernel.shared_1[((threadIdx.x*72) + 2)]))
-        conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[12]*kernel.shared_1[((threadIdx.x*72) + 3)]))
-        conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[13]*kernel.shared_1[((threadIdx.x*72) + 4)]))
-        conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[14]*kernel.shared_1[((threadIdx.x*72) + 5)]))
-        conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[15]*kernel.shared_1[((threadIdx.x*72) + 6)]))
-        conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[16]*kernel.shared_1[((threadIdx.x*72) + 7)]))
-        conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[17]*kernel.shared_1[((threadIdx.x*72) + 8)]))
-        conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[12]*kernel.shared_1[(threadIdx.x*72)]))
-        conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[13]*kernel.shared_1[((threadIdx.x*72) + 1)]))
-        conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[14]*kernel.shared_1[((threadIdx.x*72) + 2)]))
-        conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[15]*kernel.shared_1[((threadIdx.x*72) + 3)]))
-        conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[16]*kernel.shared_1[((threadIdx.x*72) + 4)]))
-        conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[17]*kernel.shared_1[((threadIdx.x*72) + 5)]))
-        conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[18]*kernel.shared_1[((threadIdx.x*72) + 6)]))
-        conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[19]*kernel.shared_1[((threadIdx.x*72) + 7)]))
-        conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[20]*kernel.shared_1[((threadIdx.x*72) + 8)]))
-        conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[15]*kernel.shared_1[(threadIdx.x*72)]))
-        conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[16]*kernel.shared_1[((threadIdx.x*72) + 1)]))
-        conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[17]*kernel.shared_1[((threadIdx.x*72) + 2)]))
-        conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[18]*kernel.shared_1[((threadIdx.x*72) + 3)]))
-        conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[19]*kernel.shared_1[((threadIdx.x*72) + 4)]))
-        conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[20]*kernel.shared_1[((threadIdx.x*72) + 5)]))
-        conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[21]*kernel.shared_1[((threadIdx.x*72) + 6)]))
-        conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[22]*kernel.shared_1[((threadIdx.x*72) + 7)]))
-        conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[23]*kernel.shared_1[((threadIdx.x*72) + 8)]))
-        conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[18]*kernel.shared_1[(threadIdx.x*72)]))
-        conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[19]*kernel.shared_1[((threadIdx.x*72) + 1)]))
-        conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[20]*kernel.shared_1[((threadIdx.x*72) + 2)]))
-        conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[21]*kernel.shared_1[((threadIdx.x*72) + 3)]))
-        conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[22]*kernel.shared_1[((threadIdx.x*72) + 4)]))
-        conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[23]*kernel.shared_1[((threadIdx.x*72) + 5)]))
-        conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[24]*kernel.shared_1[((threadIdx.x*72) + 6)]))
-        conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[25]*kernel.shared_1[((threadIdx.x*72) + 7)]))
-        conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[26]*kernel.shared_1[((threadIdx.x*72) + 8)]))
-        conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[0]*kernel.shared_1[((threadIdx.x*72) + 36)]))
-        conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[1]*kernel.shared_1[((threadIdx.x*72) + 37)]))
-        conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[2]*kernel.shared_1[((threadIdx.x*72) + 38)]))
-        conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[3]*kernel.shared_1[((threadIdx.x*72) + 39)]))
-        conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[4]*kernel.shared_1[((threadIdx.x*72) + 40)]))
-        conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[5]*kernel.shared_1[((threadIdx.x*72) + 41)]))
-        conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[6]*kernel.shared_1[((threadIdx.x*72) + 42)]))
-        conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[7]*kernel.shared_1[((threadIdx.x*72) + 43)]))
-        conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[8]*kernel.shared_1[((threadIdx.x*72) + 44)]))
-        conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[3]*kernel.shared_1[((threadIdx.x*72) + 36)]))
-        conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[4]*kernel.shared_1[((threadIdx.x*72) + 37)]))
-        conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[5]*kernel.shared_1[((threadIdx.x*72) + 38)]))
-        conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[6]*kernel.shared_1[((threadIdx.x*72) + 39)]))
-        conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[7]*kernel.shared_1[((threadIdx.x*72) + 40)]))
-        conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[8]*kernel.shared_1[((threadIdx.x*72) + 41)]))
-        conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[9]*kernel.shared_1[((threadIdx.x*72) + 42)]))
-        conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[10]*kernel.shared_1[((threadIdx.x*72) + 43)]))
-        conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[11]*kernel.shared_1[((threadIdx.x*72) + 44)]))
-        conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[6]*kernel.shared_1[((threadIdx.x*72) + 36)]))
-        conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[7]*kernel.shared_1[((threadIdx.x*72) + 37)]))
-        conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[8]*kernel.shared_1[((threadIdx.x*72) + 38)]))
-        conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[9]*kernel.shared_1[((threadIdx.x*72) + 39)]))
-        conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[10]*kernel.shared_1[((threadIdx.x*72) + 40)]))
-        conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[11]*kernel.shared_1[((threadIdx.x*72) + 41)]))
-        conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[12]*kernel.shared_1[((threadIdx.x*72) + 42)]))
-        conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[13]*kernel.shared_1[((threadIdx.x*72) + 43)]))
-        conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[14]*kernel.shared_1[((threadIdx.x*72) + 44)]))
-        conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[9]*kernel.shared_1[((threadIdx.x*72) + 36)]))
-        conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[10]*kernel.shared_1[((threadIdx.x*72) + 37)]))
-        conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[11]*kernel.shared_1[((threadIdx.x*72) + 38)]))
-        conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[12]*kernel.shared_1[((threadIdx.x*72) + 39)]))
-        conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[13]*kernel.shared_1[((threadIdx.x*72) + 40)]))
-        conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[14]*kernel.shared_1[((threadIdx.x*72) + 41)]))
-        conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[15]*kernel.shared_1[((threadIdx.x*72) + 42)]))
-        conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[16]*kernel.shared_1[((threadIdx.x*72) + 43)]))
-        conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[17]*kernel.shared_1[((threadIdx.x*72) + 44)]))
-        conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[12]*kernel.shared_1[((threadIdx.x*72) + 36)]))
-        conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[13]*kernel.shared_1[((threadIdx.x*72) + 37)]))
-        conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[14]*kernel.shared_1[((threadIdx.x*72) + 38)]))
-        conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[15]*kernel.shared_1[((threadIdx.x*72) + 39)]))
-        conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[16]*kernel.shared_1[((threadIdx.x*72) + 40)]))
-        conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[17]*kernel.shared_1[((threadIdx.x*72) + 41)]))
-        conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[18]*kernel.shared_1[((threadIdx.x*72) + 42)]))
-        conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[19]*kernel.shared_1[((threadIdx.x*72) + 43)]))
-        conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[20]*kernel.shared_1[((threadIdx.x*72) + 44)]))
-        conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[15]*kernel.shared_1[((threadIdx.x*72) + 36)]))
-        conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[16]*kernel.shared_1[((threadIdx.x*72) + 37)]))
-        conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[17]*kernel.shared_1[((threadIdx.x*72) + 38)]))
-        conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[18]*kernel.shared_1[((threadIdx.x*72) + 39)]))
-        conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[19]*kernel.shared_1[((threadIdx.x*72) + 40)]))
-        conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[20]*kernel.shared_1[((threadIdx.x*72) + 41)]))
-        conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[21]*kernel.shared_1[((threadIdx.x*72) + 42)]))
-        conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[22]*kernel.shared_1[((threadIdx.x*72) + 43)]))
-        conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[23]*kernel.shared_1[((threadIdx.x*72) + 44)]))
-        conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[18]*kernel.shared_1[((threadIdx.x*72) + 36)]))
-        conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[19]*kernel.shared_1[((threadIdx.x*72) + 37)]))
-        conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[20]*kernel.shared_1[((threadIdx.x*72) + 38)]))
-        conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[21]*kernel.shared_1[((threadIdx.x*72) + 39)]))
-        conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[22]*kernel.shared_1[((threadIdx.x*72) + 40)]))
-        conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[23]*kernel.shared_1[((threadIdx.x*72) + 41)]))
-        conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[24]*kernel.shared_1[((threadIdx.x*72) + 42)]))
-        conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[25]*kernel.shared_1[((threadIdx.x*72) + 43)]))
-        conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[26]*kernel.shared_1[((threadIdx.x*72) + 44)]))
-        conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[27]*kernel.shared_1[((threadIdx.x*72) + 9)]))
-        conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[28]*kernel.shared_1[((threadIdx.x*72) + 10)]))
-        conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[29]*kernel.shared_1[((threadIdx.x*72) + 11)]))
-        conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[30]*kernel.shared_1[((threadIdx.x*72) + 12)]))
-        conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[31]*kernel.shared_1[((threadIdx.x*72) + 13)]))
-        conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[32]*kernel.shared_1[((threadIdx.x*72) + 14)]))
-        conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[33]*kernel.shared_1[((threadIdx.x*72) + 15)]))
-        conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[34]*kernel.shared_1[((threadIdx.x*72) + 16)]))
-        conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[35]*kernel.shared_1[((threadIdx.x*72) + 17)]))
-        conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[30]*kernel.shared_1[((threadIdx.x*72) + 9)]))
-        conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[31]*kernel.shared_1[((threadIdx.x*72) + 10)]))
-        conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[32]*kernel.shared_1[((threadIdx.x*72) + 11)]))
-        conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[33]*kernel.shared_1[((threadIdx.x*72) + 12)]))
-        conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[34]*kernel.shared_1[((threadIdx.x*72) + 13)]))
-        conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[35]*kernel.shared_1[((threadIdx.x*72) + 14)]))
-        conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[36]*kernel.shared_1[((threadIdx.x*72) + 15)]))
-        conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[37]*kernel.shared_1[((threadIdx.x*72) + 16)]))
-        conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[38]*kernel.shared_1[((threadIdx.x*72) + 17)]))
-        conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[33]*kernel.shared_1[((threadIdx.x*72) + 9)]))
-        conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[34]*kernel.shared_1[((threadIdx.x*72) + 10)]))
-        conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[35]*kernel.shared_1[((threadIdx.x*72) + 11)]))
-        conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[36]*kernel.shared_1[((threadIdx.x*72) + 12)]))
-        conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[37]*kernel.shared_1[((threadIdx.x*72) + 13)]))
-        conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[38]*kernel.shared_1[((threadIdx.x*72) + 14)]))
-        conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[39]*kernel.shared_1[((threadIdx.x*72) + 15)]))
-        conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[40]*kernel.shared_1[((threadIdx.x*72) + 16)]))
-        conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[41]*kernel.shared_1[((threadIdx.x*72) + 17)]))
-        conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[36]*kernel.shared_1[((threadIdx.x*72) + 9)]))
-        conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[37]*kernel.shared_1[((threadIdx.x*72) + 10)]))
-        conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[38]*kernel.shared_1[((threadIdx.x*72) + 11)]))
-        conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[39]*kernel.shared_1[((threadIdx.x*72) + 12)]))
-        conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[40]*kernel.shared_1[((threadIdx.x*72) + 13)]))
-        conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[41]*kernel.shared_1[((threadIdx.x*72) + 14)]))
-        conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[42]*kernel.shared_1[((threadIdx.x*72) + 15)]))
-        conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[43]*kernel.shared_1[((threadIdx.x*72) + 16)]))
-        conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[44]*kernel.shared_1[((threadIdx.x*72) + 17)]))
-        conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[39]*kernel.shared_1[((threadIdx.x*72) + 9)]))
-        conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[40]*kernel.shared_1[((threadIdx.x*72) + 10)]))
-        conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[41]*kernel.shared_1[((threadIdx.x*72) + 11)]))
-        conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[42]*kernel.shared_1[((threadIdx.x*72) + 12)]))
-        conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[43]*kernel.shared_1[((threadIdx.x*72) + 13)]))
-        conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[44]*kernel.shared_1[((threadIdx.x*72) + 14)]))
-        conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[45]*kernel.shared_1[((threadIdx.x*72) + 15)]))
-        conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[46]*kernel.shared_1[((threadIdx.x*72) + 16)]))
-        conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[47]*kernel.shared_1[((threadIdx.x*72) + 17)]))
-        conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[42]*kernel.shared_1[((threadIdx.x*72) + 9)]))
-        conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[43]*kernel.shared_1[((threadIdx.x*72) + 10)]))
-        conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[44]*kernel.shared_1[((threadIdx.x*72) + 11)]))
-        conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[45]*kernel.shared_1[((threadIdx.x*72) + 12)]))
-        conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[46]*kernel.shared_1[((threadIdx.x*72) + 13)]))
-        conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[47]*kernel.shared_1[((threadIdx.x*72) + 14)]))
-        conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[48]*kernel.shared_1[((threadIdx.x*72) + 15)]))
-        conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[49]*kernel.shared_1[((threadIdx.x*72) + 16)]))
-        conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[50]*kernel.shared_1[((threadIdx.x*72) + 17)]))
-        conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[45]*kernel.shared_1[((threadIdx.x*72) + 9)]))
-        conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[46]*kernel.shared_1[((threadIdx.x*72) + 10)]))
-        conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[47]*kernel.shared_1[((threadIdx.x*72) + 11)]))
-        conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[48]*kernel.shared_1[((threadIdx.x*72) + 12)]))
-        conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[49]*kernel.shared_1[((threadIdx.x*72) + 13)]))
-        conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[50]*kernel.shared_1[((threadIdx.x*72) + 14)]))
-        conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[51]*kernel.shared_1[((threadIdx.x*72) + 15)]))
-        conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[52]*kernel.shared_1[((threadIdx.x*72) + 16)]))
-        conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[53]*kernel.shared_1[((threadIdx.x*72) + 17)]))
-        conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[27]*kernel.shared_1[((threadIdx.x*72) + 45)]))
-        conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[28]*kernel.shared_1[((threadIdx.x*72) + 46)]))
-        conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[29]*kernel.shared_1[((threadIdx.x*72) + 47)]))
-        conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[30]*kernel.shared_1[((threadIdx.x*72) + 48)]))
-        conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[31]*kernel.shared_1[((threadIdx.x*72) + 49)]))
-        conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[32]*kernel.shared_1[((threadIdx.x*72) + 50)]))
-        conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[33]*kernel.shared_1[((threadIdx.x*72) + 51)]))
-        conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[34]*kernel.shared_1[((threadIdx.x*72) + 52)]))
-        conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[35]*kernel.shared_1[((threadIdx.x*72) + 53)]))
-        conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[30]*kernel.shared_1[((threadIdx.x*72) + 45)]))
-        conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[31]*kernel.shared_1[((threadIdx.x*72) + 46)]))
-        conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[32]*kernel.shared_1[((threadIdx.x*72) + 47)]))
-        conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[33]*kernel.shared_1[((threadIdx.x*72) + 48)]))
-        conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[34]*kernel.shared_1[((threadIdx.x*72) + 49)]))
-        conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[35]*kernel.shared_1[((threadIdx.x*72) + 50)]))
-        conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[36]*kernel.shared_1[((threadIdx.x*72) + 51)]))
-        conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[37]*kernel.shared_1[((threadIdx.x*72) + 52)]))
-        conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[38]*kernel.shared_1[((threadIdx.x*72) + 53)]))
-        conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[33]*kernel.shared_1[((threadIdx.x*72) + 45)]))
-        conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[34]*kernel.shared_1[((threadIdx.x*72) + 46)]))
-        conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[35]*kernel.shared_1[((threadIdx.x*72) + 47)]))
-        conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[36]*kernel.shared_1[((threadIdx.x*72) + 48)]))
-        conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[37]*kernel.shared_1[((threadIdx.x*72) + 49)]))
-        conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[38]*kernel.shared_1[((threadIdx.x*72) + 50)]))
-        conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[39]*kernel.shared_1[((threadIdx.x*72) + 51)]))
-        conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[40]*kernel.shared_1[((threadIdx.x*72) + 52)]))
-        conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[41]*kernel.shared_1[((threadIdx.x*72) + 53)]))
-        conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[36]*kernel.shared_1[((threadIdx.x*72) + 45)]))
-        conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[37]*kernel.shared_1[((threadIdx.x*72) + 46)]))
-        conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[38]*kernel.shared_1[((threadIdx.x*72) + 47)]))
-        conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[39]*kernel.shared_1[((threadIdx.x*72) + 48)]))
-        conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[40]*kernel.shared_1[((threadIdx.x*72) + 49)]))
-        conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[41]*kernel.shared_1[((threadIdx.x*72) + 50)]))
-        conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[42]*kernel.shared_1[((threadIdx.x*72) + 51)]))
-        conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[43]*kernel.shared_1[((threadIdx.x*72) + 52)]))
-        conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[44]*kernel.shared_1[((threadIdx.x*72) + 53)]))
-        conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[39]*kernel.shared_1[((threadIdx.x*72) + 45)]))
-        conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[40]*kernel.shared_1[((threadIdx.x*72) + 46)]))
-        conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[41]*kernel.shared_1[((threadIdx.x*72) + 47)]))
-        conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[42]*kernel.shared_1[((threadIdx.x*72) + 48)]))
-        conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[43]*kernel.shared_1[((threadIdx.x*72) + 49)]))
-        conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[44]*kernel.shared_1[((threadIdx.x*72) + 50)]))
-        conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[45]*kernel.shared_1[((threadIdx.x*72) + 51)]))
-        conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[46]*kernel.shared_1[((threadIdx.x*72) + 52)]))
-        conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[47]*kernel.shared_1[((threadIdx.x*72) + 53)]))
-        conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[42]*kernel.shared_1[((threadIdx.x*72) + 45)]))
-        conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[43]*kernel.shared_1[((threadIdx.x*72) + 46)]))
-        conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[44]*kernel.shared_1[((threadIdx.x*72) + 47)]))
-        conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[45]*kernel.shared_1[((threadIdx.x*72) + 48)]))
-        conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[46]*kernel.shared_1[((threadIdx.x*72) + 49)]))
-        conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[47]*kernel.shared_1[((threadIdx.x*72) + 50)]))
-        conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[48]*kernel.shared_1[((threadIdx.x*72) + 51)]))
-        conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[49]*kernel.shared_1[((threadIdx.x*72) + 52)]))
-        conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[50]*kernel.shared_1[((threadIdx.x*72) + 53)]))
-        conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[45]*kernel.shared_1[((threadIdx.x*72) + 45)]))
-        conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[46]*kernel.shared_1[((threadIdx.x*72) + 46)]))
-        conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[47]*kernel.shared_1[((threadIdx.x*72) + 47)]))
-        conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[48]*kernel.shared_1[((threadIdx.x*72) + 48)]))
-        conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[49]*kernel.shared_1[((threadIdx.x*72) + 49)]))
-        conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[50]*kernel.shared_1[((threadIdx.x*72) + 50)]))
-        conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[51]*kernel.shared_1[((threadIdx.x*72) + 51)]))
-        conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[52]*kernel.shared_1[((threadIdx.x*72) + 52)]))
-        conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[53]*kernel.shared_1[((threadIdx.x*72) + 53)]))
-        conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[54]*kernel.shared_1[((threadIdx.x*72) + 18)]))
-        conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[55]*kernel.shared_1[((threadIdx.x*72) + 19)]))
-        conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[56]*kernel.shared_1[((threadIdx.x*72) + 20)]))
-        conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[57]*kernel.shared_1[((threadIdx.x*72) + 21)]))
-        conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[58]*kernel.shared_1[((threadIdx.x*72) + 22)]))
-        conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[59]*kernel.shared_1[((threadIdx.x*72) + 23)]))
-        conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[60]*kernel.shared_1[((threadIdx.x*72) + 24)]))
-        conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[61]*kernel.shared_1[((threadIdx.x*72) + 25)]))
-        conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[62]*kernel.shared_1[((threadIdx.x*72) + 26)]))
-        conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[57]*kernel.shared_1[((threadIdx.x*72) + 18)]))
-        conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[58]*kernel.shared_1[((threadIdx.x*72) + 19)]))
-        conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[59]*kernel.shared_1[((threadIdx.x*72) + 20)]))
-        conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[60]*kernel.shared_1[((threadIdx.x*72) + 21)]))
-        conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[61]*kernel.shared_1[((threadIdx.x*72) + 22)]))
-        conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[62]*kernel.shared_1[((threadIdx.x*72) + 23)]))
-        conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[63]*kernel.shared_1[((threadIdx.x*72) + 24)]))
-        conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[64]*kernel.shared_1[((threadIdx.x*72) + 25)]))
-        conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[65]*kernel.shared_1[((threadIdx.x*72) + 26)]))
-        conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[60]*kernel.shared_1[((threadIdx.x*72) + 18)]))
-        conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[61]*kernel.shared_1[((threadIdx.x*72) + 19)]))
-        conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[62]*kernel.shared_1[((threadIdx.x*72) + 20)]))
-        conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[63]*kernel.shared_1[((threadIdx.x*72) + 21)]))
-        conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[64]*kernel.shared_1[((threadIdx.x*72) + 22)]))
-        conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[65]*kernel.shared_1[((threadIdx.x*72) + 23)]))
-        conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[66]*kernel.shared_1[((threadIdx.x*72) + 24)]))
-        conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[67]*kernel.shared_1[((threadIdx.x*72) + 25)]))
-        conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[68]*kernel.shared_1[((threadIdx.x*72) + 26)]))
-        conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[63]*kernel.shared_1[((threadIdx.x*72) + 18)]))
-        conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[64]*kernel.shared_1[((threadIdx.x*72) + 19)]))
-        conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[65]*kernel.shared_1[((threadIdx.x*72) + 20)]))
-        conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[66]*kernel.shared_1[((threadIdx.x*72) + 21)]))
-        conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[67]*kernel.shared_1[((threadIdx.x*72) + 22)]))
-        conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[68]*kernel.shared_1[((threadIdx.x*72) + 23)]))
-        conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[69]*kernel.shared_1[((threadIdx.x*72) + 24)]))
-        conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[70]*kernel.shared_1[((threadIdx.x*72) + 25)]))
-        conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[71]*kernel.shared_1[((threadIdx.x*72) + 26)]))
-        conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[66]*kernel.shared_1[((threadIdx.x*72) + 18)]))
-        conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[67]*kernel.shared_1[((threadIdx.x*72) + 19)]))
-        conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[68]*kernel.shared_1[((threadIdx.x*72) + 20)]))
-        conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[69]*kernel.shared_1[((threadIdx.x*72) + 21)]))
-        conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[70]*kernel.shared_1[((threadIdx.x*72) + 22)]))
-        conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[71]*kernel.shared_1[((threadIdx.x*72) + 23)]))
-        conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[72]*kernel.shared_1[((threadIdx.x*72) + 24)]))
-        conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[73]*kernel.shared_1[((threadIdx.x*72) + 25)]))
-        conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[74]*kernel.shared_1[((threadIdx.x*72) + 26)]))
-        conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[69]*kernel.shared_1[((threadIdx.x*72) + 18)]))
-        conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[70]*kernel.shared_1[((threadIdx.x*72) + 19)]))
-        conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[71]*kernel.shared_1[((threadIdx.x*72) + 20)]))
-        conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[72]*kernel.shared_1[((threadIdx.x*72) + 21)]))
-        conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[73]*kernel.shared_1[((threadIdx.x*72) + 22)]))
-        conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[74]*kernel.shared_1[((threadIdx.x*72) + 23)]))
-        conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[75]*kernel.shared_1[((threadIdx.x*72) + 24)]))
-        conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[76]*kernel.shared_1[((threadIdx.x*72) + 25)]))
-        conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[77]*kernel.shared_1[((threadIdx.x*72) + 26)]))
-        conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[72]*kernel.shared_1[((threadIdx.x*72) + 18)]))
-        conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[73]*kernel.shared_1[((threadIdx.x*72) + 19)]))
-        conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[74]*kernel.shared_1[((threadIdx.x*72) + 20)]))
-        conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[75]*kernel.shared_1[((threadIdx.x*72) + 21)]))
-        conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[76]*kernel.shared_1[((threadIdx.x*72) + 22)]))
-        conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[77]*kernel.shared_1[((threadIdx.x*72) + 23)]))
-        conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[78]*kernel.shared_1[((threadIdx.x*72) + 24)]))
-        conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[79]*kernel.shared_1[((threadIdx.x*72) + 25)]))
-        conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[80]*kernel.shared_1[((threadIdx.x*72) + 26)]))
-        conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[54]*kernel.shared_1[((threadIdx.x*72) + 54)]))
-        conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[55]*kernel.shared_1[((threadIdx.x*72) + 55)]))
-        conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[56]*kernel.shared_1[((threadIdx.x*72) + 56)]))
-        conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[57]*kernel.shared_1[((threadIdx.x*72) + 57)]))
-        conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[58]*kernel.shared_1[((threadIdx.x*72) + 58)]))
-        conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[59]*kernel.shared_1[((threadIdx.x*72) + 59)]))
-        conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[60]*kernel.shared_1[((threadIdx.x*72) + 60)]))
-        conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[61]*kernel.shared_1[((threadIdx.x*72) + 61)]))
-        conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[62]*kernel.shared_1[((threadIdx.x*72) + 62)]))
-        conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[57]*kernel.shared_1[((threadIdx.x*72) + 54)]))
-        conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[58]*kernel.shared_1[((threadIdx.x*72) + 55)]))
-        conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[59]*kernel.shared_1[((threadIdx.x*72) + 56)]))
-        conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[60]*kernel.shared_1[((threadIdx.x*72) + 57)]))
-        conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[61]*kernel.shared_1[((threadIdx.x*72) + 58)]))
-        conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[62]*kernel.shared_1[((threadIdx.x*72) + 59)]))
-        conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[63]*kernel.shared_1[((threadIdx.x*72) + 60)]))
-        conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[64]*kernel.shared_1[((threadIdx.x*72) + 61)]))
-        conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[65]*kernel.shared_1[((threadIdx.x*72) + 62)]))
-        conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[60]*kernel.shared_1[((threadIdx.x*72) + 54)]))
-        conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[61]*kernel.shared_1[((threadIdx.x*72) + 55)]))
-        conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[62]*kernel.shared_1[((threadIdx.x*72) + 56)]))
-        conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[63]*kernel.shared_1[((threadIdx.x*72) + 57)]))
-        conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[64]*kernel.shared_1[((threadIdx.x*72) + 58)]))
-        conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[65]*kernel.shared_1[((threadIdx.x*72) + 59)]))
-        conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[66]*kernel.shared_1[((threadIdx.x*72) + 60)]))
-        conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[67]*kernel.shared_1[((threadIdx.x*72) + 61)]))
-        conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[68]*kernel.shared_1[((threadIdx.x*72) + 62)]))
-        conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[63]*kernel.shared_1[((threadIdx.x*72) + 54)]))
-        conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[64]*kernel.shared_1[((threadIdx.x*72) + 55)]))
-        conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[65]*kernel.shared_1[((threadIdx.x*72) + 56)]))
-        conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[66]*kernel.shared_1[((threadIdx.x*72) + 57)]))
-        conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[67]*kernel.shared_1[((threadIdx.x*72) + 58)]))
-        conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[68]*kernel.shared_1[((threadIdx.x*72) + 59)]))
-        conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[69]*kernel.shared_1[((threadIdx.x*72) + 60)]))
-        conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[70]*kernel.shared_1[((threadIdx.x*72) + 61)]))
-        conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[71]*kernel.shared_1[((threadIdx.x*72) + 62)]))
-        conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[66]*kernel.shared_1[((threadIdx.x*72) + 54)]))
-        conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[67]*kernel.shared_1[((threadIdx.x*72) + 55)]))
-        conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[68]*kernel.shared_1[((threadIdx.x*72) + 56)]))
-        conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[69]*kernel.shared_1[((threadIdx.x*72) + 57)]))
-        conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[70]*kernel.shared_1[((threadIdx.x*72) + 58)]))
-        conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[71]*kernel.shared_1[((threadIdx.x*72) + 59)]))
-        conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[72]*kernel.shared_1[((threadIdx.x*72) + 60)]))
-        conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[73]*kernel.shared_1[((threadIdx.x*72) + 61)]))
-        conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[74]*kernel.shared_1[((threadIdx.x*72) + 62)]))
-        conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[69]*kernel.shared_1[((threadIdx.x*72) + 54)]))
-        conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[70]*kernel.shared_1[((threadIdx.x*72) + 55)]))
-        conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[71]*kernel.shared_1[((threadIdx.x*72) + 56)]))
-        conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[72]*kernel.shared_1[((threadIdx.x*72) + 57)]))
-        conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[73]*kernel.shared_1[((threadIdx.x*72) + 58)]))
-        conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[74]*kernel.shared_1[((threadIdx.x*72) + 59)]))
-        conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[75]*kernel.shared_1[((threadIdx.x*72) + 60)]))
-        conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[76]*kernel.shared_1[((threadIdx.x*72) + 61)]))
-        conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[77]*kernel.shared_1[((threadIdx.x*72) + 62)]))
-        conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[72]*kernel.shared_1[((threadIdx.x*72) + 54)]))
-        conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[73]*kernel.shared_1[((threadIdx.x*72) + 55)]))
-        conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[74]*kernel.shared_1[((threadIdx.x*72) + 56)]))
-        conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[75]*kernel.shared_1[((threadIdx.x*72) + 57)]))
-        conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[76]*kernel.shared_1[((threadIdx.x*72) + 58)]))
-        conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[77]*kernel.shared_1[((threadIdx.x*72) + 59)]))
-        conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[78]*kernel.shared_1[((threadIdx.x*72) + 60)]))
-        conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[79]*kernel.shared_1[((threadIdx.x*72) + 61)]))
-        conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[80]*kernel.shared_1[((threadIdx.x*72) + 62)]))
-        conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[81]*kernel.shared_1[((threadIdx.x*72) + 27)]))
-        conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[82]*kernel.shared_1[((threadIdx.x*72) + 28)]))
-        conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[83]*kernel.shared_1[((threadIdx.x*72) + 29)]))
-        conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[84]*kernel.shared_1[((threadIdx.x*72) + 30)]))
-        conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[85]*kernel.shared_1[((threadIdx.x*72) + 31)]))
-        conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[86]*kernel.shared_1[((threadIdx.x*72) + 32)]))
-        conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[87]*kernel.shared_1[((threadIdx.x*72) + 33)]))
-        conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[88]*kernel.shared_1[((threadIdx.x*72) + 34)]))
-        conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[89]*kernel.shared_1[((threadIdx.x*72) + 35)]))
-        conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[84]*kernel.shared_1[((threadIdx.x*72) + 27)]))
-        conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[85]*kernel.shared_1[((threadIdx.x*72) + 28)]))
-        conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[86]*kernel.shared_1[((threadIdx.x*72) + 29)]))
-        conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[87]*kernel.shared_1[((threadIdx.x*72) + 30)]))
-        conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[88]*kernel.shared_1[((threadIdx.x*72) + 31)]))
-        conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[89]*kernel.shared_1[((threadIdx.x*72) + 32)]))
-        conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[90]*kernel.shared_1[((threadIdx.x*72) + 33)]))
-        conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[91]*kernel.shared_1[((threadIdx.x*72) + 34)]))
-        conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[92]*kernel.shared_1[((threadIdx.x*72) + 35)]))
-        conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[87]*kernel.shared_1[((threadIdx.x*72) + 27)]))
-        conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[88]*kernel.shared_1[((threadIdx.x*72) + 28)]))
-        conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[89]*kernel.shared_1[((threadIdx.x*72) + 29)]))
-        conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[90]*kernel.shared_1[((threadIdx.x*72) + 30)]))
-        conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[91]*kernel.shared_1[((threadIdx.x*72) + 31)]))
-        conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[92]*kernel.shared_1[((threadIdx.x*72) + 32)]))
-        conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[93]*kernel.shared_1[((threadIdx.x*72) + 33)]))
-        conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[94]*kernel.shared_1[((threadIdx.x*72) + 34)]))
-        conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[95]*kernel.shared_1[((threadIdx.x*72) + 35)]))
-        conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[90]*kernel.shared_1[((threadIdx.x*72) + 27)]))
-        conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[91]*kernel.shared_1[((threadIdx.x*72) + 28)]))
-        conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[92]*kernel.shared_1[((threadIdx.x*72) + 29)]))
-        conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[93]*kernel.shared_1[((threadIdx.x*72) + 30)]))
-        conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[94]*kernel.shared_1[((threadIdx.x*72) + 31)]))
-        conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[95]*kernel.shared_1[((threadIdx.x*72) + 32)]))
-        conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[96]*kernel.shared_1[((threadIdx.x*72) + 33)]))
-        conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[97]*kernel.shared_1[((threadIdx.x*72) + 34)]))
-        conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[98]*kernel.shared_1[((threadIdx.x*72) + 35)]))
-        conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[93]*kernel.shared_1[((threadIdx.x*72) + 27)]))
-        conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[94]*kernel.shared_1[((threadIdx.x*72) + 28)]))
-        conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[95]*kernel.shared_1[((threadIdx.x*72) + 29)]))
-        conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[96]*kernel.shared_1[((threadIdx.x*72) + 30)]))
-        conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[97]*kernel.shared_1[((threadIdx.x*72) + 31)]))
-        conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[98]*kernel.shared_1[((threadIdx.x*72) + 32)]))
-        conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[99]*kernel.shared_1[((threadIdx.x*72) + 33)]))
-        conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[100]*kernel.shared_1[((threadIdx.x*72) + 34)]))
-        conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[101]*kernel.shared_1[((threadIdx.x*72) + 35)]))
-        conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[96]*kernel.shared_1[((threadIdx.x*72) + 27)]))
-        conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[97]*kernel.shared_1[((threadIdx.x*72) + 28)]))
-        conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[98]*kernel.shared_1[((threadIdx.x*72) + 29)]))
-        conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[99]*kernel.shared_1[((threadIdx.x*72) + 30)]))
-        conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[100]*kernel.shared_1[((threadIdx.x*72) + 31)]))
-        conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[101]*kernel.shared_1[((threadIdx.x*72) + 32)]))
-        conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[102]*kernel.shared_1[((threadIdx.x*72) + 33)]))
-        conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[103]*kernel.shared_1[((threadIdx.x*72) + 34)]))
-        conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[104]*kernel.shared_1[((threadIdx.x*72) + 35)]))
-        conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[99]*kernel.shared_1[((threadIdx.x*72) + 27)]))
-        conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[100]*kernel.shared_1[((threadIdx.x*72) + 28)]))
-        conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[101]*kernel.shared_1[((threadIdx.x*72) + 29)]))
-        conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[102]*kernel.shared_1[((threadIdx.x*72) + 30)]))
-        conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[103]*kernel.shared_1[((threadIdx.x*72) + 31)]))
-        conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[104]*kernel.shared_1[((threadIdx.x*72) + 32)]))
-        conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[105]*kernel.shared_1[((threadIdx.x*72) + 33)]))
-        conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[106]*kernel.shared_1[((threadIdx.x*72) + 34)]))
-        conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[107]*kernel.shared_1[((threadIdx.x*72) + 35)]))
-        conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[81]*kernel.shared_1[((threadIdx.x*72) + 63)]))
-        conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[82]*kernel.shared_1[((threadIdx.x*72) + 64)]))
-        conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[83]*kernel.shared_1[((threadIdx.x*72) + 65)]))
-        conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[84]*kernel.shared_1[((threadIdx.x*72) + 66)]))
-        conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[85]*kernel.shared_1[((threadIdx.x*72) + 67)]))
-        conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[86]*kernel.shared_1[((threadIdx.x*72) + 68)]))
-        conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[87]*kernel.shared_1[((threadIdx.x*72) + 69)]))
-        conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[88]*kernel.shared_1[((threadIdx.x*72) + 70)]))
-        conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[89]*kernel.shared_1[((threadIdx.x*72) + 71)]))
-        conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[84]*kernel.shared_1[((threadIdx.x*72) + 63)]))
-        conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[85]*kernel.shared_1[((threadIdx.x*72) + 64)]))
-        conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[86]*kernel.shared_1[((threadIdx.x*72) + 65)]))
-        conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[87]*kernel.shared_1[((threadIdx.x*72) + 66)]))
-        conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[88]*kernel.shared_1[((threadIdx.x*72) + 67)]))
-        conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[89]*kernel.shared_1[((threadIdx.x*72) + 68)]))
-        conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[90]*kernel.shared_1[((threadIdx.x*72) + 69)]))
-        conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[91]*kernel.shared_1[((threadIdx.x*72) + 70)]))
-        conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[92]*kernel.shared_1[((threadIdx.x*72) + 71)]))
-        conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[87]*kernel.shared_1[((threadIdx.x*72) + 63)]))
-        conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[88]*kernel.shared_1[((threadIdx.x*72) + 64)]))
-        conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[89]*kernel.shared_1[((threadIdx.x*72) + 65)]))
-        conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[90]*kernel.shared_1[((threadIdx.x*72) + 66)]))
-        conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[91]*kernel.shared_1[((threadIdx.x*72) + 67)]))
-        conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[92]*kernel.shared_1[((threadIdx.x*72) + 68)]))
-        conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[93]*kernel.shared_1[((threadIdx.x*72) + 69)]))
-        conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[94]*kernel.shared_1[((threadIdx.x*72) + 70)]))
-        conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[95]*kernel.shared_1[((threadIdx.x*72) + 71)]))
-        conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[90]*kernel.shared_1[((threadIdx.x*72) + 63)]))
-        conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[91]*kernel.shared_1[((threadIdx.x*72) + 64)]))
-        conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[92]*kernel.shared_1[((threadIdx.x*72) + 65)]))
-        conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[93]*kernel.shared_1[((threadIdx.x*72) + 66)]))
-        conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[94]*kernel.shared_1[((threadIdx.x*72) + 67)]))
-        conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[95]*kernel.shared_1[((threadIdx.x*72) + 68)]))
-        conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[96]*kernel.shared_1[((threadIdx.x*72) + 69)]))
-        conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[97]*kernel.shared_1[((threadIdx.x*72) + 70)]))
-        conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[98]*kernel.shared_1[((threadIdx.x*72) + 71)]))
-        conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[93]*kernel.shared_1[((threadIdx.x*72) + 63)]))
-        conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[94]*kernel.shared_1[((threadIdx.x*72) + 64)]))
-        conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[95]*kernel.shared_1[((threadIdx.x*72) + 65)]))
-        conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[96]*kernel.shared_1[((threadIdx.x*72) + 66)]))
-        conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[97]*kernel.shared_1[((threadIdx.x*72) + 67)]))
-        conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[98]*kernel.shared_1[((threadIdx.x*72) + 68)]))
-        conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[99]*kernel.shared_1[((threadIdx.x*72) + 69)]))
-        conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[100]*kernel.shared_1[((threadIdx.x*72) + 70)]))
-        conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[101]*kernel.shared_1[((threadIdx.x*72) + 71)]))
-        conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[96]*kernel.shared_1[((threadIdx.x*72) + 63)]))
-        conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[97]*kernel.shared_1[((threadIdx.x*72) + 64)]))
-        conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[98]*kernel.shared_1[((threadIdx.x*72) + 65)]))
-        conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[99]*kernel.shared_1[((threadIdx.x*72) + 66)]))
-        conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[100]*kernel.shared_1[((threadIdx.x*72) + 67)]))
-        conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[101]*kernel.shared_1[((threadIdx.x*72) + 68)]))
-        conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[102]*kernel.shared_1[((threadIdx.x*72) + 69)]))
-        conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[103]*kernel.shared_1[((threadIdx.x*72) + 70)]))
-        conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[104]*kernel.shared_1[((threadIdx.x*72) + 71)]))
-        conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[99]*kernel.shared_1[((threadIdx.x*72) + 63)]))
-        conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[100]*kernel.shared_1[((threadIdx.x*72) + 64)]))
-        conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[101]*kernel.shared_1[((threadIdx.x*72) + 65)]))
-        conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[102]*kernel.shared_1[((threadIdx.x*72) + 66)]))
-        conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[103]*kernel.shared_1[((threadIdx.x*72) + 67)]))
-        conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[104]*kernel.shared_1[((threadIdx.x*72) + 68)]))
-        conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[105]*kernel.shared_1[((threadIdx.x*72) + 69)]))
-        conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[106]*kernel.shared_1[((threadIdx.x*72) + 70)]))
-        conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[107]*kernel.shared_1[((threadIdx.x*72) + 71)]))
       }
     }
     for (i1.inner: int32, 0, 2) {
-      for (i2.inner: int32, 0, 7) {
-        compute[(((((floordiv(blockIdx.x, 7)*1568) + (threadIdx.x*98)) + (i1.inner*49)) + (i2.inner*7)) + floormod(blockIdx.x, 7))] = max((conv2d_nchw_1[((i1.inner*7) + i2.inner)] + bias[(((floordiv(blockIdx.x, 7)*32) + (threadIdx.x*2)) + i1.inner)]), 0f32)
-      }
+      compute[(((blockIdx.x*196) + (i1.inner*49)) + threadIdx.x)] = max((conv2d_nchw_1[i1.inner] + bias[((blockIdx.x*4) + i1.inner)]), 0f32)
+      compute[((((blockIdx.x*196) + (i1.inner*49)) + threadIdx.x) + 7)] = max((conv2d_nchw_1[(i1.inner + 2)] + bias[((blockIdx.x*4) + i1.inner)]), 0f32)
+      compute[((((blockIdx.x*196) + (i1.inner*49)) + threadIdx.x) + 14)] = max((conv2d_nchw_1[(i1.inner + 4)] + bias[((blockIdx.x*4) + i1.inner)]), 0f32)
+      compute[((((blockIdx.x*196) + (i1.inner*49)) + threadIdx.x) + 21)] = max((conv2d_nchw_1[(i1.inner + 6)] + bias[((blockIdx.x*4) + i1.inner)]), 0f32)
+      compute[((((blockIdx.x*196) + (i1.inner*49)) + threadIdx.x) + 28)] = max((conv2d_nchw_1[(i1.inner + 8)] + bias[((blockIdx.x*4) + i1.inner)]), 0f32)
+      compute[((((blockIdx.x*196) + (i1.inner*49)) + threadIdx.x) + 35)] = max((conv2d_nchw_1[(i1.inner + 10)] + bias[((blockIdx.x*4) + i1.inner)]), 0f32)
+      compute[((((blockIdx.x*196) + (i1.inner*49)) + threadIdx.x) + 42)] = max((conv2d_nchw_1[(i1.inner + 12)] + bias[((blockIdx.x*4) + i1.inner)]), 0f32)
+      compute[((((blockIdx.x*196) + (i1.inner*49)) + threadIdx.x) + 98)] = max((conv2d_nchw_1[(i1.inner + 14)] + bias[(((blockIdx.x*4) + i1.inner) + 2)]), 0f32)
+      compute[((((blockIdx.x*196) + (i1.inner*49)) + threadIdx.x) + 105)] = max((conv2d_nchw_1[(i1.inner + 16)] + bias[(((blockIdx.x*4) + i1.inner) + 2)]), 0f32)
+      compute[((((blockIdx.x*196) + (i1.inner*49)) + threadIdx.x) + 112)] = max((conv2d_nchw_1[(i1.inner + 18)] + bias[(((blockIdx.x*4) + i1.inner) + 2)]), 0f32)
+      compute[((((blockIdx.x*196) + (i1.inner*49)) + threadIdx.x) + 119)] = max((conv2d_nchw_1[(i1.inner + 20)] + bias[(((blockIdx.x*4) + i1.inner) + 2)]), 0f32)
+      compute[((((blockIdx.x*196) + (i1.inner*49)) + threadIdx.x) + 126)] = max((conv2d_nchw_1[(i1.inner + 22)] + bias[(((blockIdx.x*4) + i1.inner) + 2)]), 0f32)
+      compute[((((blockIdx.x*196) + (i1.inner*49)) + threadIdx.x) + 133)] = max((conv2d_nchw_1[(i1.inner + 24)] + bias[(((blockIdx.x*4) + i1.inner) + 2)]), 0f32)
+      compute[((((blockIdx.x*196) + (i1.inner*49)) + threadIdx.x) + 140)] = max((conv2d_nchw_1[(i1.inner + 26)] + bias[(((blockIdx.x*4) + i1.inner) + 2)]), 0f32)
     }
   }
 }
@@ -1215,7 +1001,7 @@ cooperative fetching, unrolling and operator fusion.</p>
 <span class="p">)</span>
 </pre></div>
 </div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Execution time of this operator: 0.361 ms
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Execution time of this operator: 0.369 ms
 </pre></div>
 </div>
 </div>
@@ -1246,34 +1032,34 @@ conv2d_nchw_nn_o_o_o_i, conv2d_nchw_nn_o_o_i = s[conv2d_nchw].split(conv2d_nchw_
 conv2d_nchw_nn_o_o_o_o, conv2d_nchw_nn_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_nn_o_o_o_i, factor=1)
 conv2d_nchw_ff_o_i, conv2d_nchw_ff_i = s[conv2d_nchw].split(conv2d_nchw_ff, factor=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=1)
+conv2d_nchw_ff_o_o_o_i, conv2d_nchw_ff_o_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_o_i, factor=1)
+conv2d_nchw_ff_o_o_o_o, conv2d_nchw_ff_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_o_o_i, factor=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=7)
+conv2d_nchw_yy_o_o_i, conv2d_nchw_yy_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_i, factor=1)
 conv2d_nchw_yy_o_o_o_i, conv2d_nchw_yy_o_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_o_i, factor=1)
-conv2d_nchw_yy_o_o_o_o, conv2d_nchw_yy_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_o_o_i, factor=1)
+conv2d_nchw_yy_o_o_o_o, conv2d_nchw_yy_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_o_o_i, factor=7)
 conv2d_nchw_xx_o_i, conv2d_nchw_xx_i = s[conv2d_nchw].split(conv2d_nchw_xx, factor=1)
 conv2d_nchw_xx_o_o_i, conv2d_nchw_xx_o_i = s[conv2d_nchw].split(conv2d_nchw_xx_o_i, factor=1)
-conv2d_nchw_xx_o_o_o_i, conv2d_nchw_xx_o_o_i = s[conv2d_nchw].split(conv2d_nchw_xx_o_o_i, factor=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=4)
-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=3)
+conv2d_nchw_ry_o_i, conv2d_nchw_ry_i = s[conv2d_nchw].split(conv2d_nchw_ry, factor=1)
+conv2d_nchw_ry_o_o, conv2d_nchw_ry_o_i = s[conv2d_nchw].split(conv2d_nchw_ry_o_i, factor=3)
+conv2d_nchw_rx_o_i, conv2d_nchw_rx_i = s[conv2d_nchw].split(conv2d_nchw_rx, factor=1)
 conv2d_nchw_rx_o_o, conv2d_nchw_rx_o_i = s[conv2d_nchw].split(conv2d_nchw_rx_o_i, factor=1)
 s[conv2d_nchw].reorder(conv2d_nchw_nn_o_o_o_o, conv2d_nchw_ff_o_o_o_o, conv2d_nchw_yy_o_o_o_o, conv2d_nchw_xx_o_o_o_o, conv2d_nchw_nn_o_o_o_i, conv2d_nchw_ff_o_o_o_i, conv2d_nchw_yy_o_o_o_i, conv2d_nchw_xx_o_o_o_i, conv2d_nchw_nn_o_o_i, conv2d_nchw_ff_o_o_i, conv2d_nchw_yy_o_o_i, conv2d_nchw_xx_o_o_i, conv2d_nchw_rc_o_o, conv2d_nchw_ry_o_o, conv2d_nchw_rx_o_o, conv2d_nchw_rc_o_i, conv2d_nchw_ry_o_i, conv2d_nchw_rx_o_i, conv2d_nchw_nn_o_i, conv2d_nchw_ff_o_i, conv2d_nchw_yy_o_i, conv2d_nc [...]
 compute_i0_o_i, compute_i0_i = s[compute].split(compute_i0, factor=1)
 compute_i0_o_o_i, compute_i0_o_i = s[compute].split(compute_i0_o_i, factor=1)
 compute_i0_o_o_o, compute_i0_o_o_i = s[compute].split(compute_i0_o_o_i, factor=1)
 compute_i1_o_i, compute_i1_i = s[compute].split(compute_i1, factor=2)
-compute_i1_o_o_i, compute_i1_o_i = s[compute].split(compute_i1_o_i, factor=16)
-compute_i1_o_o_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=7)
+compute_i1_o_o_i, compute_i1_o_i = s[compute].split(compute_i1_o_i, factor=1)
+compute_i1_o_o_o, compute_i1_o_o_i = s[compute].split(compute_i1_o_o_i, factor=2)
+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=1)
-compute_i2_o_o_o, compute_i2_o_o_i = s[compute].split(compute_i2_o_o_i, factor=1)
+compute_i2_o_o_o, compute_i2_o_o_i = s[compute].split(compute_i2_o_o_i, factor=7)
 compute_i3_o_i, compute_i3_i = s[compute].split(compute_i3, factor=1)
-compute_i3_o_o_i, compute_i3_o_i = s[compute].split(compute_i3_o_i, factor=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)
@@ -1293,14 +1079,14 @@ 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=16)
+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=7)
 s[kernel_shared].bind(kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i, te.thread_axis(&quot;threadIdx.x&quot;))
 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=16)
+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=7)
 s[pad_temp_shared].bind(pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i, te.thread_axis(&quot;threadIdx.x&quot;))
-s[conv2d_nchw].pragma(conv2d_nchw_nn_o_o_o_o, &quot;auto_unroll_max_step&quot;, 512)
+s[conv2d_nchw].pragma(conv2d_nchw_nn_o_o_o_o, &quot;auto_unroll_max_step&quot;, 1024)
 s[conv2d_nchw].pragma(conv2d_nchw_nn_o_o_o_o, &quot;unroll_explicit&quot;, True)
 
 CUDA source code:
@@ -1318,617 +1104,440 @@ CUDA source code:
   #define int64_t long long
   #define uint64_t unsigned long long
 #endif
-extern &quot;C&quot; __global__ void __launch_bounds__(16) default_function_kernel0(float* __restrict__ data, float* __restrict__ kernel, float* __restrict__ compute, float* __restrict__ bias) {
-  float conv2d_nchw[14];
-  __shared__ float pad_temp_shared[108];
-  __shared__ float kernel_shared[1152];
+extern &quot;C&quot; __global__ void __launch_bounds__(7) default_function_kernel0(float* __restrict__ data, float* __restrict__ kernel, float* __restrict__ compute, float* __restrict__ bias) {
+  float conv2d_nchw[28];
+  __shared__ float pad_temp_shared[252];
+  __shared__ float kernel_shared[48];
   conv2d_nchw[0] = 0.000000e+00f;
-  conv2d_nchw[1] = 0.000000e+00f;
   conv2d_nchw[2] = 0.000000e+00f;
-  conv2d_nchw[3] = 0.000000e+00f;
   conv2d_nchw[4] = 0.000000e+00f;
-  conv2d_nchw[5] = 0.000000e+00f;
   conv2d_nchw[6] = 0.000000e+00f;
-  conv2d_nchw[7] = 0.000000e+00f;
   conv2d_nchw[8] = 0.000000e+00f;
-  conv2d_nchw[9] = 0.000000e+00f;
   conv2d_nchw[10] = 0.000000e+00f;
-  conv2d_nchw[11] = 0.000000e+00f;
   conv2d_nchw[12] = 0.000000e+00f;
+  conv2d_nchw[14] = 0.000000e+00f;
+  conv2d_nchw[16] = 0.000000e+00f;
+  conv2d_nchw[18] = 0.000000e+00f;
+  conv2d_nchw[20] = 0.000000e+00f;
+  conv2d_nchw[22] = 0.000000e+00f;
+  conv2d_nchw[24] = 0.000000e+00f;
+  conv2d_nchw[26] = 0.000000e+00f;
+  conv2d_nchw[1] = 0.000000e+00f;
+  conv2d_nchw[3] = 0.000000e+00f;
+  conv2d_nchw[5] = 0.000000e+00f;
+  conv2d_nchw[7] = 0.000000e+00f;
+  conv2d_nchw[9] = 0.000000e+00f;
+  conv2d_nchw[11] = 0.000000e+00f;
   conv2d_nchw[13] = 0.000000e+00f;
+  conv2d_nchw[15] = 0.000000e+00f;
+  conv2d_nchw[17] = 0.000000e+00f;
+  conv2d_nchw[19] = 0.000000e+00f;
+  conv2d_nchw[21] = 0.000000e+00f;
+  conv2d_nchw[23] = 0.000000e+00f;
+  conv2d_nchw[25] = 0.000000e+00f;
+  conv2d_nchw[27] = 0.000000e+00f;
   for (int rc_outer_outer = 0; rc_outer_outer &lt; 128; ++rc_outer_outer) {
-    __syncthreads();
-    pad_temp_shared[((int)threadIdx.x)] = ((((3 &lt;= ((int)threadIdx.x)) &amp;&amp; (1 &lt;= ((((int)blockIdx.x) % 7) + (((int)threadIdx.x) % 3)))) &amp;&amp; (((((int)blockIdx.x) % 7) + (((int)threadIdx.x) % 3)) &lt; 8)) ? data[(((((rc_outer_outer * 196) + ((((int)threadIdx.x) / 3) * 7)) + (((int)blockIdx.x) % 7)) + (((int)threadIdx.x) % 3)) - 8)] : 0.000000e+00f);
-    pad_temp_shared[(((int)threadIdx.x) + 16)] = (((((3 &lt;= ((((int)threadIdx.x) + 16) % 27)) &amp;&amp; (((((int)threadIdx.x) + 16) % 27) &lt; 24)) &amp;&amp; (1 &lt;= ((((int)blockIdx.x) % 7) + ((((int)threadIdx.x) + 1) % 3)))) &amp;&amp; (((((int)blockIdx.x) % 7) + ((((int)threadIdx.x) + 1) % 3)) &lt; 8)) ? data[((((((rc_outer_outer * 196) + (((((int)threadIdx.x) + 16) / 27) * 49)) + ((((((int)threadIdx.x) + 16) % 27) / 3) * 7)) + (((int)blockIdx.x) % 7)) + ((((int)threadIdx.x) + 1) [...]
-    pad_temp_shared[(((int)threadIdx.x) + 32)] = (((1 &lt;= ((((int)blockIdx.x) % 7) + ((((int)threadIdx.x) + 2) % 3))) &amp;&amp; (((((int)blockIdx.x) % 7) + ((((int)threadIdx.x) + 2) % 3)) &lt; 8)) ? data[((((((rc_outer_outer * 196) + (((((int)threadIdx.x) + 32) / 27) * 49)) + (((((int)threadIdx.x) + 5) / 3) * 7)) + (((int)blockIdx.x) % 7)) + ((((int)threadIdx.x) + 2) % 3)) - 8)] : 0.000000e+00f);
-    pad_temp_shared[(((int)threadIdx.x) + 48)] = (((((1 &lt;= (((((int)threadIdx.x) / 3) + 7) % 9)) &amp;&amp; (((((int)threadIdx.x) + 21) % 27) &lt; 24)) &amp;&amp; (1 &lt;= ((((int)blockIdx.x) % 7) + (((int)threadIdx.x) % 3)))) &amp;&amp; (((((int)blockIdx.x) % 7) + (((int)threadIdx.x) % 3)) &lt; 8)) ? data[((((((rc_outer_outer * 196) + (((((int)threadIdx.x) + 48) / 27) * 49)) + ((((((int)threadIdx.x) / 3) + 7) % 9) * 7)) + (((int)blockIdx.x) % 7)) + (((int)threadIdx.x) % 3)) - 8)] : 0 [...]
-    pad_temp_shared[(((int)threadIdx.x) + 64)] = ((((((int)threadIdx.x) &lt; 14) &amp;&amp; (1 &lt;= ((((int)blockIdx.x) % 7) + ((((int)threadIdx.x) + 1) % 3)))) &amp;&amp; (((((int)blockIdx.x) % 7) + ((((int)threadIdx.x) + 1) % 3)) &lt; 8)) ? data[((((((rc_outer_outer * 196) + (((((int)threadIdx.x) + 64) / 27) * 49)) + (((((int)threadIdx.x) + 10) / 3) * 7)) + (((int)blockIdx.x) % 7)) + ((((int)threadIdx.x) + 1) % 3)) - 8)] : 0.000000e+00f);
-    pad_temp_shared[(((int)threadIdx.x) + 80)] = (((((3 &lt;= ((((int)threadIdx.x) + 26) % 27)) &amp;&amp; (((((int)threadIdx.x) + 26) % 27) &lt; 24)) &amp;&amp; (1 &lt;= ((((int)blockIdx.x) % 7) + ((((int)threadIdx.x) + 2) % 3)))) &amp;&amp; (((((int)blockIdx.x) % 7) + ((((int)threadIdx.x) + 2) % 3)) &lt; 8)) ? data[((((((rc_outer_outer * 196) + (((((int)threadIdx.x) + 80) / 27) * 49)) + ((((((int)threadIdx.x) + 26) % 27) / 3) * 7)) + (((int)blockIdx.x) % 7)) + ((((int)threadIdx.x) + 2) [...]
-    if (((int)threadIdx.x) &lt; 12) {
-      pad_temp_shared[(((int)threadIdx.x) + 96)] = ((((((int)threadIdx.x) &lt; 9) &amp;&amp; (1 &lt;= ((((int)blockIdx.x) % 7) + (((int)threadIdx.x) % 3)))) &amp;&amp; (((((int)blockIdx.x) % 7) + (((int)threadIdx.x) % 3)) &lt; 8)) ? data[((((((rc_outer_outer * 196) + (((((int)threadIdx.x) + 96) / 27) * 49)) + ((((int)threadIdx.x) / 3) * 7)) + (((int)blockIdx.x) % 7)) + (((int)threadIdx.x) % 3)) + 27)] : 0.000000e+00f);
+    for (int rx_outer_outer = 0; rx_outer_outer &lt; 3; ++rx_outer_outer) {
+      __syncthreads();
+      pad_temp_shared[((int)threadIdx.x)] = 0.000000e+00f;
+      pad_temp_shared[(((int)threadIdx.x) + 7)] = (((1 &lt;= (((int)threadIdx.x) + rx_outer_outer)) &amp;&amp; ((((int)threadIdx.x) + rx_outer_outer) &lt; 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) - 1)] : 0.000000e+00f);
+      pad_temp_shared[(((int)threadIdx.x) + 14)] = (((1 &lt;= (((int)threadIdx.x) + rx_outer_outer)) &amp;&amp; ((((int)threadIdx.x) + rx_outer_outer) &lt; 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 6)] : 0.000000e+00f);
+      pad_temp_shared[(((int)threadIdx.x) + 21)] = (((1 &lt;= (((int)threadIdx.x) + rx_outer_outer)) &amp;&amp; ((((int)threadIdx.x) + rx_outer_outer) &lt; 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 13)] : 0.000000e+00f);
+      pad_temp_shared[(((int)threadIdx.x) + 28)] = (((1 &lt;= (((int)threadIdx.x) + rx_outer_outer)) &amp;&amp; ((((int)threadIdx.x) + rx_outer_outer) &lt; 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 20)] : 0.000000e+00f);
+      pad_temp_shared[(((int)threadIdx.x) + 35)] = (((1 &lt;= (((int)threadIdx.x) + rx_outer_outer)) &amp;&amp; ((((int)threadIdx.x) + rx_outer_outer) &lt; 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 27)] : 0.000000e+00f);
+      pad_temp_shared[(((int)threadIdx.x) + 42)] = (((1 &lt;= (((int)threadIdx.x) + rx_outer_outer)) &amp;&amp; ((((int)threadIdx.x) + rx_outer_outer) &lt; 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 34)] : 0.000000e+00f);
+      pad_temp_shared[(((int)threadIdx.x) + 49)] = (((1 &lt;= (((int)threadIdx.x) + rx_outer_outer)) &amp;&amp; ((((int)threadIdx.x) + rx_outer_outer) &lt; 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 41)] : 0.000000e+00f);
+      pad_temp_shared[(((int)threadIdx.x) + 56)] = 0.000000e+00f;
+      pad_temp_shared[(((int)threadIdx.x) + 63)] = 0.000000e+00f;
+      pad_temp_shared[(((int)threadIdx.x) + 70)] = (((1 &lt;= (((int)threadIdx.x) + rx_outer_outer)) &amp;&amp; ((((int)threadIdx.x) + rx_outer_outer) &lt; 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 48)] : 0.000000e+00f);
+      pad_temp_shared[(((int)threadIdx.x) + 77)] = (((1 &lt;= (((int)threadIdx.x) + rx_outer_outer)) &amp;&amp; ((((int)threadIdx.x) + rx_outer_outer) &lt; 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 55)] : 0.000000e+00f);
+      pad_temp_shared[(((int)threadIdx.x) + 84)] = (((1 &lt;= (((int)threadIdx.x) + rx_outer_outer)) &amp;&amp; ((((int)threadIdx.x) + rx_outer_outer) &lt; 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 62)] : 0.000000e+00f);
+      pad_temp_shared[(((int)threadIdx.x) + 91)] = (((1 &lt;= (((int)threadIdx.x) + rx_outer_outer)) &amp;&amp; ((((int)threadIdx.x) + rx_outer_outer) &lt; 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 69)] : 0.000000e+00f);
+      pad_temp_shared[(((int)threadIdx.x) + 98)] = (((1 &lt;= (((int)threadIdx.x) + rx_outer_outer)) &amp;&amp; ((((int)threadIdx.x) + rx_outer_outer) &lt; 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 76)] : 0.000000e+00f);
+      pad_temp_shared[(((int)threadIdx.x) + 105)] = (((1 &lt;= (((int)threadIdx.x) + rx_outer_outer)) &amp;&amp; ((((int)threadIdx.x) + rx_outer_outer) &lt; 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 83)] : 0.000000e+00f);
+      pad_temp_shared[(((int)threadIdx.x) + 112)] = (((1 &lt;= (((int)threadIdx.x) + rx_outer_outer)) &amp;&amp; ((((int)threadIdx.x) + rx_outer_outer) &lt; 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 90)] : 0.000000e+00f);
+      pad_temp_shared[(((int)threadIdx.x) + 119)] = 0.000000e+00f;
+      pad_temp_shared[(((int)threadIdx.x) + 126)] = 0.000000e+00f;
+      pad_temp_shared[(((int)threadIdx.x) + 133)] = (((1 &lt;= (((int)threadIdx.x) + rx_outer_outer)) &amp;&amp; ((((int)threadIdx.x) + rx_outer_outer) &lt; 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 97)] : 0.000000e+00f);
+      pad_temp_shared[(((int)threadIdx.x) + 140)] = (((1 &lt;= (((int)threadIdx.x) + rx_outer_outer)) &amp;&amp; ((((int)threadIdx.x) + rx_outer_outer) &lt; 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 104)] : 0.000000e+00f);
+      pad_temp_shared[(((int)threadIdx.x) + 147)] = (((1 &lt;= (((int)threadIdx.x) + rx_outer_outer)) &amp;&amp; ((((int)threadIdx.x) + rx_outer_outer) &lt; 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 111)] : 0.000000e+00f);
+      pad_temp_shared[(((int)threadIdx.x) + 154)] = (((1 &lt;= (((int)threadIdx.x) + rx_outer_outer)) &amp;&amp; ((((int)threadIdx.x) + rx_outer_outer) &lt; 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 118)] : 0.000000e+00f);
+      pad_temp_shared[(((int)threadIdx.x) + 161)] = (((1 &lt;= (((int)threadIdx.x) + rx_outer_outer)) &amp;&amp; ((((int)threadIdx.x) + rx_outer_outer) &lt; 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 125)] : 0.000000e+00f);
+      pad_temp_shared[(((int)threadIdx.x) + 168)] = (((1 &lt;= (((int)threadIdx.x) + rx_outer_outer)) &amp;&amp; ((((int)threadIdx.x) + rx_outer_outer) &lt; 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 132)] : 0.000000e+00f);
+      pad_temp_shared[(((int)threadIdx.x) + 175)] = (((1 &lt;= (((int)threadIdx.x) + rx_outer_outer)) &amp;&amp; ((((int)threadIdx.x) + rx_outer_outer) &lt; 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 139)] : 0.000000e+00f);
+      pad_temp_shared[(((int)threadIdx.x) + 182)] = 0.000000e+00f;
+      pad_temp_shared[(((int)threadIdx.x) + 189)] = 0.000000e+00f;
+      pad_temp_shared[(((int)threadIdx.x) + 196)] = (((1 &lt;= (((int)threadIdx.x) + rx_outer_outer)) &amp;&amp; ((((int)threadIdx.x) + rx_outer_outer) &lt; 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 146)] : 0.000000e+00f);
+      pad_temp_shared[(((int)threadIdx.x) + 203)] = (((1 &lt;= (((int)threadIdx.x) + rx_outer_outer)) &amp;&amp; ((((int)threadIdx.x) + rx_outer_outer) &lt; 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 153)] : 0.000000e+00f);
+      pad_temp_shared[(((int)threadIdx.x) + 210)] = (((1 &lt;= (((int)threadIdx.x) + rx_outer_outer)) &amp;&amp; ((((int)threadIdx.x) + rx_outer_outer) &lt; 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 160)] : 0.000000e+00f);
+      pad_temp_shared[(((int)threadIdx.x) + 217)] = (((1 &lt;= (((int)threadIdx.x) + rx_outer_outer)) &amp;&amp; ((((int)threadIdx.x) + rx_outer_outer) &lt; 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 167)] : 0.000000e+00f);
+      pad_temp_shared[(((int)threadIdx.x) + 224)] = (((1 &lt;= (((int)threadIdx.x) + rx_outer_outer)) &amp;&amp; ((((int)threadIdx.x) + rx_outer_outer) &lt; 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 174)] : 0.000000e+00f);
+      pad_temp_shared[(((int)threadIdx.x) + 231)] = (((1 &lt;= (((int)threadIdx.x) + rx_outer_outer)) &amp;&amp; ((((int)threadIdx.x) + rx_outer_outer) &lt; 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 181)] : 0.000000e+00f);
+      pad_temp_shared[(((int)threadIdx.x) + 238)] = (((1 &lt;= (((int)threadIdx.x) + rx_outer_outer)) &amp;&amp; ((((int)threadIdx.x) + rx_outer_outer) &lt; 8)) ? data[((((rc_outer_outer * 196) + rx_outer_outer) + ((int)threadIdx.x)) + 188)] : 0.000000e+00f);
+      pad_temp_shared[(((int)threadIdx.x) + 245)] = 0.000000e+00f;
+      kernel_shared[((int)threadIdx.x)] = kernel[((((((int)blockIdx.x) * 18432) + (rc_outer_outer * 36)) + (((int)threadIdx.x) * 3)) + rx_outer_outer)];
+      kernel_shared[(((int)threadIdx.x) + 7)] = kernel[((((((((int)blockIdx.x) * 18432) + (((((int)threadIdx.x) + 7) / 12) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 7) % 12) / 3) * 9)) + (((((int)threadIdx.x) + 1) % 3) * 3)) + rx_outer_outer)];
+      kernel_shared[(((int)threadIdx.x) + 14)] = kernel[(((((((int)blockIdx.x) * 18432) + (((((int)threadIdx.x) + 14) / 12) * 4608)) + (rc_outer_outer * 36)) + ((((int)threadIdx.x) + 2) * 3)) + rx_outer_outer)];
+      kernel_shared[(((int)threadIdx.x) + 21)] = kernel[((((((((int)blockIdx.x) * 18432) + (((((int)threadIdx.x) + 21) / 12) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) / 3) + 3) &amp; 3) * 9)) + ((((int)threadIdx.x) % 3) * 3)) + rx_outer_outer)];
+      kernel_shared[(((int)threadIdx.x) + 28)] = kernel[((((((((int)blockIdx.x) * 18432) + (((((int)threadIdx.x) + 28) / 12) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 4) / 3) * 9)) + (((((int)threadIdx.x) + 1) % 3) * 3)) + rx_outer_outer)];
+      kernel_shared[(((int)threadIdx.x) + 35)] = kernel[((((((((int)blockIdx.x) * 18432) + (((((int)threadIdx.x) + 35) / 12) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 11) % 12) / 3) * 9)) + (((((int)threadIdx.x) + 2) % 3) * 3)) + rx_outer_outer)];
+      if (((int)threadIdx.x) &lt; 6) {
+        kernel_shared[(((int)threadIdx.x) + 42)] = kernel[((((((((int)blockIdx.x) * 18432) + (((((int)threadIdx.x) + 42) / 12) * 4608)) + (rc_outer_outer * 36)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 18)];
+      }
+      __syncthreads();
+      conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((int)threadIdx.x)] * kernel_shared[0]));
+      conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((int)threadIdx.x) + 7)] * kernel_shared[0]));
+      conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((int)threadIdx.x) + 14)] * kernel_shared[0]));
+      conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((int)threadIdx.x) + 21)] * kernel_shared[0]));
+      conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((int)threadIdx.x) + 28)] * kernel_shared[0]));
+      conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((int)threadIdx.x) + 35)] * kernel_shared[0]));
+      conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((int)threadIdx.x) + 42)] * kernel_shared[0]));
+      conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[((int)threadIdx.x)] * kernel_shared[24]));
+      conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((int)threadIdx.x) + 7)] * kernel_shared[24]));
+      conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((int)threadIdx.x) + 14)] * kernel_shared[24]));
+      conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((int)threadIdx.x) + 21)] * kernel_shared[24]));
+      conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((int)threadIdx.x) + 28)] * kernel_shared[24]));
+      conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((int)threadIdx.x) + 35)] * kernel_shared[24]));
+      conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((int)threadIdx.x) + 42)] * kernel_shared[24]));
+      conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((int)threadIdx.x)] * kernel_shared[12]));
+      conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((int)threadIdx.x) + 7)] * kernel_shared[12]));
+      conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((int)threadIdx.x) + 14)] * kernel_shared[12]));
+      conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((int)threadIdx.x) + 21)] * kernel_shared[12]));
+      conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((int)threadIdx.x) + 28)] * kernel_shared[12]));
+      conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((int)threadIdx.x) + 35)] * kernel_shared[12]));
+      conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((int)threadIdx.x) + 42)] * kernel_shared[12]));
+      conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[((int)threadIdx.x)] * kernel_shared[36]));
+      conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((int)threadIdx.x) + 7)] * kernel_shared[36]));
+      conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((int)threadIdx.x) + 14)] * kernel_shared[36]));
+      conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((int)threadIdx.x) + 21)] * kernel_shared[36]));
+      conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((int)threadIdx.x) + 28)] * kernel_shared[36]));
+      conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((int)threadIdx.x) + 35)] * kernel_shared[36]));
+      conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((int)threadIdx.x) + 42)] * kernel_shared[36]));
+      conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((int)threadIdx.x) + 7)] * kernel_shared[1]));
+      conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((int)threadIdx.x) + 14)] * kernel_shared[1]));
+      conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((int)threadIdx.x) + 21)] * kernel_shared[1]));
+      conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((int)threadIdx.x) + 28)] * kernel_shared[1]));
+      conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((int)threadIdx.x) + 35)] * kernel_shared[1]));
+      conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((int)threadIdx.x) + 42)] * kernel_shared[1]));
+      conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((int)threadIdx.x) + 49)] * kernel_shared[1]));
+      conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((int)threadIdx.x) + 7)] * kernel_shared[25]));
+      conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((int)threadIdx.x) + 14)] * kernel_shared[25]));
+      conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((int)threadIdx.x) + 21)] * kernel_shared[25]));
+      conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((int)threadIdx.x) + 28)] * kernel_shared[25]));
+      conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((int)threadIdx.x) + 35)] * kernel_shared[25]));
+      conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((int)threadIdx.x) + 42)] * kernel_shared[25]));
+      conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((int)threadIdx.x) + 49)] * kernel_shared[25]));
+      conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((int)threadIdx.x) + 7)] * kernel_shared[13]));
+      conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((int)threadIdx.x) + 14)] * kernel_shared[13]));
+      conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((int)threadIdx.x) + 21)] * kernel_shared[13]));
+      conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((int)threadIdx.x) + 28)] * kernel_shared[13]));
+      conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((int)threadIdx.x) + 35)] * kernel_shared[13]));
+      conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((int)threadIdx.x) + 42)] * kernel_shared[13]));
+      conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((int)threadIdx.x) + 49)] * kernel_shared[13]));
+      conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((int)threadIdx.x) + 7)] * kernel_shared[37]));
+      conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((int)threadIdx.x) + 14)] * kernel_shared[37]));
+      conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((int)threadIdx.x) + 21)] * kernel_shared[37]));
+      conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((int)threadIdx.x) + 28)] * kernel_shared[37]));
+      conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((int)threadIdx.x) + 35)] * kernel_shared[37]));
+      conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((int)threadIdx.x) + 42)] * kernel_shared[37]));
+      conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((int)threadIdx.x) + 49)] * kernel_shared[37]));
+      conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((int)threadIdx.x) + 14)] * kernel_shared[2]));
+      conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((int)threadIdx.x) + 21)] * kernel_shared[2]));
+      conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((int)threadIdx.x) + 28)] * kernel_shared[2]));
+      conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((int)threadIdx.x) + 35)] * kernel_shared[2]));
+      conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((int)threadIdx.x) + 42)] * kernel_shared[2]));
+      conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((int)threadIdx.x) + 49)] * kernel_shared[2]));
+      conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((int)threadIdx.x) + 56)] * kernel_shared[2]));
+      conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((int)threadIdx.x) + 14)] * kernel_shared[26]));
+      conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((int)threadIdx.x) + 21)] * kernel_shared[26]));
+      conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((int)threadIdx.x) + 28)] * kernel_shared[26]));
+      conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((int)threadIdx.x) + 35)] * kernel_shared[26]));
+      conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((int)threadIdx.x) + 42)] * kernel_shared[26]));
+      conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((int)threadIdx.x) + 49)] * kernel_shared[26]));
+      conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((int)threadIdx.x) + 56)] * kernel_shared[26]));
+      conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((int)threadIdx.x) + 14)] * kernel_shared[14]));
+      conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((int)threadIdx.x) + 21)] * kernel_shared[14]));
+      conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((int)threadIdx.x) + 28)] * kernel_shared[14]));
+      conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((int)threadIdx.x) + 35)] * kernel_shared[14]));
+      conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((int)threadIdx.x) + 42)] * kernel_shared[14]));
+      conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((int)threadIdx.x) + 49)] * kernel_shared[14]));
+      conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((int)threadIdx.x) + 56)] * kernel_shared[14]));
+      conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((int)threadIdx.x) + 14)] * kernel_shared[38]));
+      conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((int)threadIdx.x) + 21)] * kernel_shared[38]));
+      conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((int)threadIdx.x) + 28)] * kernel_shared[38]));
+      conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((int)threadIdx.x) + 35)] * kernel_shared[38]));
+      conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((int)threadIdx.x) + 42)] * kernel_shared[38]));
+      conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((int)threadIdx.x) + 49)] * kernel_shared[38]));
+      conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((int)threadIdx.x) + 56)] * kernel_shared[38]));
+      conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((int)threadIdx.x) + 63)] * kernel_shared[3]));
+      conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((int)threadIdx.x) + 70)] * kernel_shared[3]));
+      conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((int)threadIdx.x) + 77)] * kernel_shared[3]));
+      conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((int)threadIdx.x) + 84)] * kernel_shared[3]));
+      conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((int)threadIdx.x) + 91)] * kernel_shared[3]));
+      conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((int)threadIdx.x) + 98)] * kernel_shared[3]));
+      conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((int)threadIdx.x) + 105)] * kernel_shared[3]));
+      conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((int)threadIdx.x) + 63)] * kernel_shared[27]));
+      conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((int)threadIdx.x) + 70)] * kernel_shared[27]));
+      conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((int)threadIdx.x) + 77)] * kernel_shared[27]));
+      conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((int)threadIdx.x) + 84)] * kernel_shared[27]));
+      conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((int)threadIdx.x) + 91)] * kernel_shared[27]));
+      conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((int)threadIdx.x) + 98)] * kernel_shared[27]));
+      conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((int)threadIdx.x) + 105)] * kernel_shared[27]));
+      conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((int)threadIdx.x) + 63)] * kernel_shared[15]));
+      conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((int)threadIdx.x) + 70)] * kernel_shared[15]));
+      conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((int)threadIdx.x) + 77)] * kernel_shared[15]));
+      conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((int)threadIdx.x) + 84)] * kernel_shared[15]));
+      conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((int)threadIdx.x) + 91)] * kernel_shared[15]));
+      conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((int)threadIdx.x) + 98)] * kernel_shared[15]));
+      conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((int)threadIdx.x) + 105)] * kernel_shared[15]));
+      conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((int)threadIdx.x) + 63)] * kernel_shared[39]));
+      conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((int)threadIdx.x) + 70)] * kernel_shared[39]));
+      conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((int)threadIdx.x) + 77)] * kernel_shared[39]));
+      conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((int)threadIdx.x) + 84)] * kernel_shared[39]));
+      conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((int)threadIdx.x) + 91)] * kernel_shared[39]));
+      conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((int)threadIdx.x) + 98)] * kernel_shared[39]));
+      conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((int)threadIdx.x) + 105)] * kernel_shared[39]));
+      conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((int)threadIdx.x) + 70)] * kernel_shared[4]));
+      conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((int)threadIdx.x) + 77)] * kernel_shared[4]));
+      conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((int)threadIdx.x) + 84)] * kernel_shared[4]));
+      conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((int)threadIdx.x) + 91)] * kernel_shared[4]));
+      conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((int)threadIdx.x) + 98)] * kernel_shared[4]));
+      conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((int)threadIdx.x) + 105)] * kernel_shared[4]));
+      conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((int)threadIdx.x) + 112)] * kernel_shared[4]));
+      conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((int)threadIdx.x) + 70)] * kernel_shared[28]));
+      conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((int)threadIdx.x) + 77)] * kernel_shared[28]));
+      conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((int)threadIdx.x) + 84)] * kernel_shared[28]));
+      conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((int)threadIdx.x) + 91)] * kernel_shared[28]));
+      conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((int)threadIdx.x) + 98)] * kernel_shared[28]));
+      conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((int)threadIdx.x) + 105)] * kernel_shared[28]));
+      conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((int)threadIdx.x) + 112)] * kernel_shared[28]));
+      conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((int)threadIdx.x) + 70)] * kernel_shared[16]));
+      conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((int)threadIdx.x) + 77)] * kernel_shared[16]));
+      conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((int)threadIdx.x) + 84)] * kernel_shared[16]));
+      conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((int)threadIdx.x) + 91)] * kernel_shared[16]));
+      conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((int)threadIdx.x) + 98)] * kernel_shared[16]));
+      conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((int)threadIdx.x) + 105)] * kernel_shared[16]));
+      conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((int)threadIdx.x) + 112)] * kernel_shared[16]));
+      conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((int)threadIdx.x) + 70)] * kernel_shared[40]));
+      conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((int)threadIdx.x) + 77)] * kernel_shared[40]));
+      conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((int)threadIdx.x) + 84)] * kernel_shared[40]));
+      conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((int)threadIdx.x) + 91)] * kernel_shared[40]));
+      conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((int)threadIdx.x) + 98)] * kernel_shared[40]));
+      conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((int)threadIdx.x) + 105)] * kernel_shared[40]));
+      conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((int)threadIdx.x) + 112)] * kernel_shared[40]));
+      conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((int)threadIdx.x) + 77)] * kernel_shared[5]));
+      conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((int)threadIdx.x) + 84)] * kernel_shared[5]));
+      conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((int)threadIdx.x) + 91)] * kernel_shared[5]));
+      conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((int)threadIdx.x) + 98)] * kernel_shared[5]));
+      conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((int)threadIdx.x) + 105)] * kernel_shared[5]));
+      conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((int)threadIdx.x) + 112)] * kernel_shared[5]));
+      conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((int)threadIdx.x) + 119)] * kernel_shared[5]));
+      conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((int)threadIdx.x) + 77)] * kernel_shared[29]));
+      conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((int)threadIdx.x) + 84)] * kernel_shared[29]));
+      conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((int)threadIdx.x) + 91)] * kernel_shared[29]));
+      conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((int)threadIdx.x) + 98)] * kernel_shared[29]));
+      conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((int)threadIdx.x) + 105)] * kernel_shared[29]));
+      conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((int)threadIdx.x) + 112)] * kernel_shared[29]));
+      conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((int)threadIdx.x) + 119)] * kernel_shared[29]));
+      conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((int)threadIdx.x) + 77)] * kernel_shared[17]));
+      conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((int)threadIdx.x) + 84)] * kernel_shared[17]));
+      conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((int)threadIdx.x) + 91)] * kernel_shared[17]));
+      conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((int)threadIdx.x) + 98)] * kernel_shared[17]));
+      conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((int)threadIdx.x) + 105)] * kernel_shared[17]));
+      conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((int)threadIdx.x) + 112)] * kernel_shared[17]));
+      conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((int)threadIdx.x) + 119)] * kernel_shared[17]));
+      conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((int)threadIdx.x) + 77)] * kernel_shared[41]));
+      conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((int)threadIdx.x) + 84)] * kernel_shared[41]));
+      conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((int)threadIdx.x) + 91)] * kernel_shared[41]));
+      conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((int)threadIdx.x) + 98)] * kernel_shared[41]));
+      conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((int)threadIdx.x) + 105)] * kernel_shared[41]));
+      conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((int)threadIdx.x) + 112)] * kernel_shared[41]));
+      conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((int)threadIdx.x) + 119)] * kernel_shared[41]));
+      conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((int)threadIdx.x) + 126)] * kernel_shared[6]));
+      conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((int)threadIdx.x) + 133)] * kernel_shared[6]));
+      conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((int)threadIdx.x) + 140)] * kernel_shared[6]));
+      conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((int)threadIdx.x) + 147)] * kernel_shared[6]));
+      conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((int)threadIdx.x) + 154)] * kernel_shared[6]));
+      conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((int)threadIdx.x) + 161)] * kernel_shared[6]));
+      conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((int)threadIdx.x) + 168)] * kernel_shared[6]));
+      conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((int)threadIdx.x) + 126)] * kernel_shared[30]));
+      conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((int)threadIdx.x) + 133)] * kernel_shared[30]));
+      conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((int)threadIdx.x) + 140)] * kernel_shared[30]));
+      conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((int)threadIdx.x) + 147)] * kernel_shared[30]));
+      conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((int)threadIdx.x) + 154)] * kernel_shared[30]));
+      conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((int)threadIdx.x) + 161)] * kernel_shared[30]));
+      conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((int)threadIdx.x) + 168)] * kernel_shared[30]));
+      conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((int)threadIdx.x) + 126)] * kernel_shared[18]));
+      conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((int)threadIdx.x) + 133)] * kernel_shared[18]));
+      conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((int)threadIdx.x) + 140)] * kernel_shared[18]));
+      conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((int)threadIdx.x) + 147)] * kernel_shared[18]));
+      conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((int)threadIdx.x) + 154)] * kernel_shared[18]));
+      conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((int)threadIdx.x) + 161)] * kernel_shared[18]));
+      conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((int)threadIdx.x) + 168)] * kernel_shared[18]));
+      conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((int)threadIdx.x) + 126)] * kernel_shared[42]));
+      conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((int)threadIdx.x) + 133)] * kernel_shared[42]));
+      conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((int)threadIdx.x) + 140)] * kernel_shared[42]));
+      conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((int)threadIdx.x) + 147)] * kernel_shared[42]));
+      conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((int)threadIdx.x) + 154)] * kernel_shared[42]));
+      conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((int)threadIdx.x) + 161)] * kernel_shared[42]));
+      conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((int)threadIdx.x) + 168)] * kernel_shared[42]));
+      conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((int)threadIdx.x) + 133)] * kernel_shared[7]));
+      conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((int)threadIdx.x) + 140)] * kernel_shared[7]));
+      conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((int)threadIdx.x) + 147)] * kernel_shared[7]));
+      conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((int)threadIdx.x) + 154)] * kernel_shared[7]));
+      conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((int)threadIdx.x) + 161)] * kernel_shared[7]));
+      conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((int)threadIdx.x) + 168)] * kernel_shared[7]));
+      conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((int)threadIdx.x) + 175)] * kernel_shared[7]));
+      conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((int)threadIdx.x) + 133)] * kernel_shared[31]));
+      conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((int)threadIdx.x) + 140)] * kernel_shared[31]));
+      conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((int)threadIdx.x) + 147)] * kernel_shared[31]));
+      conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((int)threadIdx.x) + 154)] * kernel_shared[31]));
+      conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((int)threadIdx.x) + 161)] * kernel_shared[31]));
+      conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((int)threadIdx.x) + 168)] * kernel_shared[31]));
+      conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((int)threadIdx.x) + 175)] * kernel_shared[31]));
+      conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((int)threadIdx.x) + 133)] * kernel_shared[19]));
+      conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((int)threadIdx.x) + 140)] * kernel_shared[19]));
+      conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((int)threadIdx.x) + 147)] * kernel_shared[19]));
+      conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((int)threadIdx.x) + 154)] * kernel_shared[19]));
+      conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((int)threadIdx.x) + 161)] * kernel_shared[19]));
+      conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((int)threadIdx.x) + 168)] * kernel_shared[19]));
+      conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((int)threadIdx.x) + 175)] * kernel_shared[19]));
+      conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((int)threadIdx.x) + 133)] * kernel_shared[43]));
+      conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((int)threadIdx.x) + 140)] * kernel_shared[43]));
+      conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((int)threadIdx.x) + 147)] * kernel_shared[43]));
+      conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((int)threadIdx.x) + 154)] * kernel_shared[43]));
+      conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((int)threadIdx.x) + 161)] * kernel_shared[43]));
+      conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((int)threadIdx.x) + 168)] * kernel_shared[43]));
+      conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((int)threadIdx.x) + 175)] * kernel_shared[43]));
+      conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((int)threadIdx.x) + 140)] * kernel_shared[8]));
+      conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((int)threadIdx.x) + 147)] * kernel_shared[8]));
+      conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((int)threadIdx.x) + 154)] * kernel_shared[8]));
+      conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((int)threadIdx.x) + 161)] * kernel_shared[8]));
+      conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((int)threadIdx.x) + 168)] * kernel_shared[8]));
+      conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((int)threadIdx.x) + 175)] * kernel_shared[8]));
+      conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((int)threadIdx.x) + 182)] * kernel_shared[8]));
+      conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((int)threadIdx.x) + 140)] * kernel_shared[32]));
+      conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((int)threadIdx.x) + 147)] * kernel_shared[32]));
+      conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((int)threadIdx.x) + 154)] * kernel_shared[32]));
+      conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((int)threadIdx.x) + 161)] * kernel_shared[32]));
+      conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((int)threadIdx.x) + 168)] * kernel_shared[32]));
+      conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((int)threadIdx.x) + 175)] * kernel_shared[32]));
+      conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((int)threadIdx.x) + 182)] * kernel_shared[32]));
+      conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((int)threadIdx.x) + 140)] * kernel_shared[20]));
+      conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((int)threadIdx.x) + 147)] * kernel_shared[20]));
+      conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((int)threadIdx.x) + 154)] * kernel_shared[20]));
+      conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((int)threadIdx.x) + 161)] * kernel_shared[20]));
+      conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((int)threadIdx.x) + 168)] * kernel_shared[20]));
+      conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((int)threadIdx.x) + 175)] * kernel_shared[20]));
+      conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((int)threadIdx.x) + 182)] * kernel_shared[20]));
+      conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((int)threadIdx.x) + 140)] * kernel_shared[44]));
+      conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((int)threadIdx.x) + 147)] * kernel_shared[44]));
+      conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((int)threadIdx.x) + 154)] * kernel_shared[44]));
+      conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((int)threadIdx.x) + 161)] * kernel_shared[44]));
+      conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((int)threadIdx.x) + 168)] * kernel_shared[44]));
+      conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((int)threadIdx.x) + 175)] * kernel_shared[44]));
+      conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((int)threadIdx.x) + 182)] * kernel_shared[44]));
+      conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((int)threadIdx.x) + 189)] * kernel_shared[9]));
+      conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((int)threadIdx.x) + 196)] * kernel_shared[9]));
+      conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((int)threadIdx.x) + 203)] * kernel_shared[9]));
+      conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((int)threadIdx.x) + 210)] * kernel_shared[9]));
+      conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((int)threadIdx.x) + 217)] * kernel_shared[9]));
+      conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((int)threadIdx.x) + 224)] * kernel_shared[9]));
+      conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((int)threadIdx.x) + 231)] * kernel_shared[9]));
+      conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((int)threadIdx.x) + 189)] * kernel_shared[33]));
+      conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((int)threadIdx.x) + 196)] * kernel_shared[33]));
+      conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((int)threadIdx.x) + 203)] * kernel_shared[33]));
+      conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((int)threadIdx.x) + 210)] * kernel_shared[33]));
+      conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((int)threadIdx.x) + 217)] * kernel_shared[33]));
+      conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((int)threadIdx.x) + 224)] * kernel_shared[33]));
+      conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((int)threadIdx.x) + 231)] * kernel_shared[33]));
+      conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((int)threadIdx.x) + 189)] * kernel_shared[21]));
+      conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((int)threadIdx.x) + 196)] * kernel_shared[21]));
+      conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((int)threadIdx.x) + 203)] * kernel_shared[21]));
+      conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((int)threadIdx.x) + 210)] * kernel_shared[21]));
+      conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((int)threadIdx.x) + 217)] * kernel_shared[21]));
+      conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((int)threadIdx.x) + 224)] * kernel_shared[21]));
+      conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((int)threadIdx.x) + 231)] * kernel_shared[21]));
+      conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((int)threadIdx.x) + 189)] * kernel_shared[45]));
+      conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((int)threadIdx.x) + 196)] * kernel_shared[45]));
+      conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((int)threadIdx.x) + 203)] * kernel_shared[45]));
+      conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((int)threadIdx.x) + 210)] * kernel_shared[45]));
+      conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((int)threadIdx.x) + 217)] * kernel_shared[45]));
+      conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((int)threadIdx.x) + 224)] * kernel_shared[45]));
+      conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((int)threadIdx.x) + 231)] * kernel_shared[45]));
+      conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((int)threadIdx.x) + 196)] * kernel_shared[10]));
+      conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((int)threadIdx.x) + 203)] * kernel_shared[10]));
+      conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((int)threadIdx.x) + 210)] * kernel_shared[10]));
+      conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((int)threadIdx.x) + 217)] * kernel_shared[10]));
+      conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((int)threadIdx.x) + 224)] * kernel_shared[10]));
+      conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((int)threadIdx.x) + 231)] * kernel_shared[10]));
+      conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((int)threadIdx.x) + 238)] * kernel_shared[10]));
+      conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((int)threadIdx.x) + 196)] * kernel_shared[34]));
+      conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((int)threadIdx.x) + 203)] * kernel_shared[34]));
+      conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((int)threadIdx.x) + 210)] * kernel_shared[34]));
+      conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((int)threadIdx.x) + 217)] * kernel_shared[34]));
+      conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((int)threadIdx.x) + 224)] * kernel_shared[34]));
+      conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((int)threadIdx.x) + 231)] * kernel_shared[34]));
+      conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((int)threadIdx.x) + 238)] * kernel_shared[34]));
+      conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((int)threadIdx.x) + 196)] * kernel_shared[22]));
+      conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((int)threadIdx.x) + 203)] * kernel_shared[22]));
+      conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((int)threadIdx.x) + 210)] * kernel_shared[22]));
+      conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((int)threadIdx.x) + 217)] * kernel_shared[22]));
+      conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((int)threadIdx.x) + 224)] * kernel_shared[22]));
+      conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((int)threadIdx.x) + 231)] * kernel_shared[22]));
+      conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((int)threadIdx.x) + 238)] * kernel_shared[22]));
+      conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((int)threadIdx.x) + 196)] * kernel_shared[46]));
+      conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((int)threadIdx.x) + 203)] * kernel_shared[46]));
+      conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((int)threadIdx.x) + 210)] * kernel_shared[46]));
+      conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((int)threadIdx.x) + 217)] * kernel_shared[46]));
+      conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((int)threadIdx.x) + 224)] * kernel_shared[46]));
+      conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((int)threadIdx.x) + 231)] * kernel_shared[46]));
+      conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((int)threadIdx.x) + 238)] * kernel_shared[46]));
+      conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((int)threadIdx.x) + 203)] * kernel_shared[11]));
+      conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((int)threadIdx.x) + 210)] * kernel_shared[11]));
+      conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((int)threadIdx.x) + 217)] * kernel_shared[11]));
+      conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((int)threadIdx.x) + 224)] * kernel_shared[11]));
+      conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((int)threadIdx.x) + 231)] * kernel_shared[11]));
+      conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((int)threadIdx.x) + 238)] * kernel_shared[11]));
+      conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((int)threadIdx.x) + 245)] * kernel_shared[11]));
+      conv2d_nchw[14] = (conv2d_nchw[14] + (pad_temp_shared[(((int)threadIdx.x) + 203)] * kernel_shared[35]));
+      conv2d_nchw[16] = (conv2d_nchw[16] + (pad_temp_shared[(((int)threadIdx.x) + 210)] * kernel_shared[35]));
+      conv2d_nchw[18] = (conv2d_nchw[18] + (pad_temp_shared[(((int)threadIdx.x) + 217)] * kernel_shared[35]));
+      conv2d_nchw[20] = (conv2d_nchw[20] + (pad_temp_shared[(((int)threadIdx.x) + 224)] * kernel_shared[35]));
+      conv2d_nchw[22] = (conv2d_nchw[22] + (pad_temp_shared[(((int)threadIdx.x) + 231)] * kernel_shared[35]));
+      conv2d_nchw[24] = (conv2d_nchw[24] + (pad_temp_shared[(((int)threadIdx.x) + 238)] * kernel_shared[35]));
+      conv2d_nchw[26] = (conv2d_nchw[26] + (pad_temp_shared[(((int)threadIdx.x) + 245)] * kernel_shared[35]));
+      conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((int)threadIdx.x) + 203)] * kernel_shared[23]));
+      conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((int)threadIdx.x) + 210)] * kernel_shared[23]));
+      conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((int)threadIdx.x) + 217)] * kernel_shared[23]));
+      conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((int)threadIdx.x) + 224)] * kernel_shared[23]));
+      conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((int)threadIdx.x) + 231)] * kernel_shared[23]));
+      conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((int)threadIdx.x) + 238)] * kernel_shared[23]));
+      conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((int)threadIdx.x) + 245)] * kernel_shared[23]));
+      conv2d_nchw[15] = (conv2d_nchw[15] + (pad_temp_shared[(((int)threadIdx.x) + 203)] * kernel_shared[47]));
+      conv2d_nchw[17] = (conv2d_nchw[17] + (pad_temp_shared[(((int)threadIdx.x) + 210)] * kernel_shared[47]));
+      conv2d_nchw[19] = (conv2d_nchw[19] + (pad_temp_shared[(((int)threadIdx.x) + 217)] * kernel_shared[47]));
+      conv2d_nchw[21] = (conv2d_nchw[21] + (pad_temp_shared[(((int)threadIdx.x) + 224)] * kernel_shared[47]));
+      conv2d_nchw[23] = (conv2d_nchw[23] + (pad_temp_shared[(((int)threadIdx.x) + 231)] * kernel_shared[47]));
+      conv2d_nchw[25] = (conv2d_nchw[25] + (pad_temp_shared[(((int)threadIdx.x) + 238)] * kernel_shared[47]));
+      conv2d_nchw[27] = (conv2d_nchw[27] + (pad_temp_shared[(((int)threadIdx.x) + 245)] * kernel_shared[47]));
     }
-    kernel_shared[((int)threadIdx.x)] = kernel[((((((int)blockIdx.x) / 7) * 147456) + (rc_outer_outer * 36)) + ((int)threadIdx.x))];
-    kernel_shared[(((int)threadIdx.x) + 16)] = kernel[(((((((int)blockIdx.x) / 7) * 147456) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 16) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 32)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 32) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 32) % 36) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 48)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 48) / 36) * 4608)) + (rc_outer_outer * 36)) + ((int)threadIdx.x)) + 12)];
-    kernel_shared[(((int)threadIdx.x) + 64)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 64) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 28) % 36) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 80)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 80) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 8) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 96)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 96) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) / 3) + 8) % 12) * 3)) + (((int)threadIdx.x) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 112)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 112) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 4) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 128)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 128) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 20) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 144)] = kernel[(((((((int)blockIdx.x) / 7) * 147456) + (rc_outer_outer * 36)) + ((int)threadIdx.x)) + 18432)];
-    kernel_shared[(((int)threadIdx.x) + 160)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 160) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 16) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 176)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 176) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 32) % 36) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 192)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 192) / 36) * 4608)) + (rc_outer_outer * 36)) + ((int)threadIdx.x)) + 12)];
-    kernel_shared[(((int)threadIdx.x) + 208)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 208) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 28) % 36) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 224)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 224) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 8) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 240)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 240) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) / 3) + 8) % 12) * 3)) + (((int)threadIdx.x) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 256)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 256) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 4) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 272)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 272) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 20) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 288)] = kernel[(((((((int)blockIdx.x) / 7) * 147456) + (rc_outer_outer * 36)) + ((int)threadIdx.x)) + 36864)];
-    kernel_shared[(((int)threadIdx.x) + 304)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 304) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 16) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 320)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 320) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 32) % 36) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 336)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 336) / 36) * 4608)) + (rc_outer_outer * 36)) + ((int)threadIdx.x)) + 12)];
-    kernel_shared[(((int)threadIdx.x) + 352)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 352) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 28) % 36) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 368)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 368) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 8) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 384)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 384) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) / 3) + 8) % 12) * 3)) + (((int)threadIdx.x) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 400)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 400) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 4) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 416)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 416) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 20) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 432)] = kernel[(((((((int)blockIdx.x) / 7) * 147456) + (rc_outer_outer * 36)) + ((int)threadIdx.x)) + 55296)];
-    kernel_shared[(((int)threadIdx.x) + 448)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 448) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 16) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 464)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 464) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 32) % 36) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 480)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 480) / 36) * 4608)) + (rc_outer_outer * 36)) + ((int)threadIdx.x)) + 12)];
-    kernel_shared[(((int)threadIdx.x) + 496)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 496) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 28) % 36) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 512)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 512) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 8) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 528)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 528) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) / 3) + 8) % 12) * 3)) + (((int)threadIdx.x) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 544)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 544) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 4) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 560)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 560) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 20) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 576)] = kernel[(((((((int)blockIdx.x) / 7) * 147456) + (rc_outer_outer * 36)) + ((int)threadIdx.x)) + 73728)];
-    kernel_shared[(((int)threadIdx.x) + 592)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 592) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 16) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 608)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 608) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 32) % 36) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 624)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 624) / 36) * 4608)) + (rc_outer_outer * 36)) + ((int)threadIdx.x)) + 12)];
-    kernel_shared[(((int)threadIdx.x) + 640)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 640) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 28) % 36) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 656)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 656) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 8) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 672)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 672) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) / 3) + 8) % 12) * 3)) + (((int)threadIdx.x) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 688)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 688) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 4) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 704)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 704) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 20) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 720)] = kernel[(((((((int)blockIdx.x) / 7) * 147456) + (rc_outer_outer * 36)) + ((int)threadIdx.x)) + 92160)];
-    kernel_shared[(((int)threadIdx.x) + 736)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 736) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 16) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 752)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 752) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 32) % 36) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 768)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 768) / 36) * 4608)) + (rc_outer_outer * 36)) + ((int)threadIdx.x)) + 12)];
-    kernel_shared[(((int)threadIdx.x) + 784)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 784) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 28) % 36) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 800)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 800) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 8) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 816)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 816) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) / 3) + 8) % 12) * 3)) + (((int)threadIdx.x) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 832)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 832) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 4) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 848)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 848) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 20) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 864)] = kernel[(((((((int)blockIdx.x) / 7) * 147456) + (rc_outer_outer * 36)) + ((int)threadIdx.x)) + 110592)];
-    kernel_shared[(((int)threadIdx.x) + 880)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 880) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 16) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 896)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 896) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 32) % 36) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 912)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 912) / 36) * 4608)) + (rc_outer_outer * 36)) + ((int)threadIdx.x)) + 12)];
-    kernel_shared[(((int)threadIdx.x) + 928)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 928) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 28) % 36) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 944)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 944) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 8) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 960)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 960) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) / 3) + 8) % 12) * 3)) + (((int)threadIdx.x) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 976)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 976) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 4) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 992)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 992) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 20) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 1008)] = kernel[(((((((int)blockIdx.x) / 7) * 147456) + (rc_outer_outer * 36)) + ((int)threadIdx.x)) + 129024)];
-    kernel_shared[(((int)threadIdx.x) + 1024)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 1024) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 16) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 1040)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 1040) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 32) % 36) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 1056)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 1056) / 36) * 4608)) + (rc_outer_outer * 36)) + ((int)threadIdx.x)) + 12)];
-    kernel_shared[(((int)threadIdx.x) + 1072)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 1072) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) + 28) % 36) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 1088)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 1088) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 8) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 1104)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 1104) / 36) * 4608)) + (rc_outer_outer * 36)) + ((((((int)threadIdx.x) / 3) + 8) % 12) * 3)) + (((int)threadIdx.x) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 1120)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 1120) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 4) / 3) * 3)) + ((((int)threadIdx.x) + 1) % 3))];
-    kernel_shared[(((int)threadIdx.x) + 1136)] = kernel[((((((((int)blockIdx.x) / 7) * 147456) + (((((int)threadIdx.x) + 1136) / 36) * 4608)) + (rc_outer_outer * 36)) + (((((int)threadIdx.x) + 20) / 3) * 3)) + ((((int)threadIdx.x) + 2) % 3))];
-    __syncthreads();
-    conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[0] * kernel_shared[(((int)threadIdx.x) * 72)]));
-    conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[1] * kernel_shared[((((int)threadIdx.x) * 72) + 1)]));
-    conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[2] * kernel_shared[((((int)threadIdx.x) * 72) + 2)]));
-    conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[3] * kernel_shared[((((int)threadIdx.x) * 72) + 3)]));
-    conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[4] * kernel_shared[((((int)threadIdx.x) * 72) + 4)]));
-    conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[5] * kernel_shared[((((int)threadIdx.x) * 72) + 5)]));
-    conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[6] * kernel_shared[((((int)threadIdx.x) * 72) + 6)]));
-    conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[7] * kernel_shared[((((int)threadIdx.x) * 72) + 7)]));
-    conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[8] * kernel_shared[((((int)threadIdx.x) * 72) + 8)]));
-    conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[3] * kernel_shared[(((int)threadIdx.x) * 72)]));
-    conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[4] * kernel_shared[((((int)threadIdx.x) * 72) + 1)]));
-    conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[5] * kernel_shared[((((int)threadIdx.x) * 72) + 2)]));
-    conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[6] * kernel_shared[((((int)threadIdx.x) * 72) + 3)]));
-    conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[7] * kernel_shared[((((int)threadIdx.x) * 72) + 4)]));
-    conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[8] * kernel_shared[((((int)threadIdx.x) * 72) + 5)]));
-    conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[9] * kernel_shared[((((int)threadIdx.x) * 72) + 6)]));
-    conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[10] * kernel_shared[((((int)threadIdx.x) * 72) + 7)]));
-    conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[11] * kernel_shared[((((int)threadIdx.x) * 72) + 8)]));
-    conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[6] * kernel_shared[(((int)threadIdx.x) * 72)]));
-    conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[7] * kernel_shared[((((int)threadIdx.x) * 72) + 1)]));
-    conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[8] * kernel_shared[((((int)threadIdx.x) * 72) + 2)]));
-    conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[9] * kernel_shared[((((int)threadIdx.x) * 72) + 3)]));
-    conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[10] * kernel_shared[((((int)threadIdx.x) * 72) + 4)]));
-    conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[11] * kernel_shared[((((int)threadIdx.x) * 72) + 5)]));
-    conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[12] * kernel_shared[((((int)threadIdx.x) * 72) + 6)]));
-    conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[13] * kernel_shared[((((int)threadIdx.x) * 72) + 7)]));
-    conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[14] * kernel_shared[((((int)threadIdx.x) * 72) + 8)]));
-    conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[9] * kernel_shared[(((int)threadIdx.x) * 72)]));
-    conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[10] * kernel_shared[((((int)threadIdx.x) * 72) + 1)]));
-    conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[11] * kernel_shared[((((int)threadIdx.x) * 72) + 2)]));
-    conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[12] * kernel_shared[((((int)threadIdx.x) * 72) + 3)]));
-    conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[13] * kernel_shared[((((int)threadIdx.x) * 72) + 4)]));
-    conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[14] * kernel_shared[((((int)threadIdx.x) * 72) + 5)]));
-    conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[15] * kernel_shared[((((int)threadIdx.x) * 72) + 6)]));
-    conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[16] * kernel_shared[((((int)threadIdx.x) * 72) + 7)]));
-    conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[17] * kernel_shared[((((int)threadIdx.x) * 72) + 8)]));
-    conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[12] * kernel_shared[(((int)threadIdx.x) * 72)]));
-    conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[13] * kernel_shared[((((int)threadIdx.x) * 72) + 1)]));
-    conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[14] * kernel_shared[((((int)threadIdx.x) * 72) + 2)]));
-    conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[15] * kernel_shared[((((int)threadIdx.x) * 72) + 3)]));
-    conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[16] * kernel_shared[((((int)threadIdx.x) * 72) + 4)]));
-    conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[17] * kernel_shared[((((int)threadIdx.x) * 72) + 5)]));
-    conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[18] * kernel_shared[((((int)threadIdx.x) * 72) + 6)]));
-    conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[19] * kernel_shared[((((int)threadIdx.x) * 72) + 7)]));
-    conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[20] * kernel_shared[((((int)threadIdx.x) * 72) + 8)]));
-    conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[15] * kernel_shared[(((int)threadIdx.x) * 72)]));
-    conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[16] * kernel_shared[((((int)threadIdx.x) * 72) + 1)]));
-    conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[17] * kernel_shared[((((int)threadIdx.x) * 72) + 2)]));
-    conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[18] * kernel_shared[((((int)threadIdx.x) * 72) + 3)]));
-    conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[19] * kernel_shared[((((int)threadIdx.x) * 72) + 4)]));
-    conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[20] * kernel_shared[((((int)threadIdx.x) * 72) + 5)]));
-    conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[21] * kernel_shared[((((int)threadIdx.x) * 72) + 6)]));
-    conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[22] * kernel_shared[((((int)threadIdx.x) * 72) + 7)]));
-    conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[23] * kernel_shared[((((int)threadIdx.x) * 72) + 8)]));
-    conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[18] * kernel_shared[(((int)threadIdx.x) * 72)]));
-    conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[19] * kernel_shared[((((int)threadIdx.x) * 72) + 1)]));
-    conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[20] * kernel_shared[((((int)threadIdx.x) * 72) + 2)]));
-    conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[21] * kernel_shared[((((int)threadIdx.x) * 72) + 3)]));
-    conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[22] * kernel_shared[((((int)threadIdx.x) * 72) + 4)]));
-    conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[23] * kernel_shared[((((int)threadIdx.x) * 72) + 5)]));
-    conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[24] * kernel_shared[((((int)threadIdx.x) * 72) + 6)]));
-    conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[25] * kernel_shared[((((int)threadIdx.x) * 72) + 7)]));
-    conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[26] * kernel_shared[((((int)threadIdx.x) * 72) + 8)]));
-    conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[0] * kernel_shared[((((int)threadIdx.x) * 72) + 36)]));
-    conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[1] * kernel_shared[((((int)threadIdx.x) * 72) + 37)]));
-    conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[2] * kernel_shared[((((int)threadIdx.x) * 72) + 38)]));
-    conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[3] * kernel_shared[((((int)threadIdx.x) * 72) + 39)]));
-    conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[4] * kernel_shared[((((int)threadIdx.x) * 72) + 40)]));
-    conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[5] * kernel_shared[((((int)threadIdx.x) * 72) + 41)]));
-    conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[6] * kernel_shared[((((int)threadIdx.x) * 72) + 42)]));
-    conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[7] * kernel_shared[((((int)threadIdx.x) * 72) + 43)]));
-    conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[8] * kernel_shared[((((int)threadIdx.x) * 72) + 44)]));
-    conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[3] * kernel_shared[((((int)threadIdx.x) * 72) + 36)]));
-    conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[4] * kernel_shared[((((int)threadIdx.x) * 72) + 37)]));
-    conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[5] * kernel_shared[((((int)threadIdx.x) * 72) + 38)]));
-    conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[6] * kernel_shared[((((int)threadIdx.x) * 72) + 39)]));
-    conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[7] * kernel_shared[((((int)threadIdx.x) * 72) + 40)]));
-    conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[8] * kernel_shared[((((int)threadIdx.x) * 72) + 41)]));
-    conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[9] * kernel_shared[((((int)threadIdx.x) * 72) + 42)]));
-    conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[10] * kernel_shared[((((int)threadIdx.x) * 72) + 43)]));
-    conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[11] * kernel_shared[((((int)threadIdx.x) * 72) + 44)]));
-    conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[6] * kernel_shared[((((int)threadIdx.x) * 72) + 36)]));
-    conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[7] * kernel_shared[((((int)threadIdx.x) * 72) + 37)]));
-    conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[8] * kernel_shared[((((int)threadIdx.x) * 72) + 38)]));
-    conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[9] * kernel_shared[((((int)threadIdx.x) * 72) + 39)]));
-    conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[10] * kernel_shared[((((int)threadIdx.x) * 72) + 40)]));
-    conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[11] * kernel_shared[((((int)threadIdx.x) * 72) + 41)]));
-    conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[12] * kernel_shared[((((int)threadIdx.x) * 72) + 42)]));
-    conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[13] * kernel_shared[((((int)threadIdx.x) * 72) + 43)]));
-    conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[14] * kernel_shared[((((int)threadIdx.x) * 72) + 44)]));
-    conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[9] * kernel_shared[((((int)threadIdx.x) * 72) + 36)]));
-    conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[10] * kernel_shared[((((int)threadIdx.x) * 72) + 37)]));
-    conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[11] * kernel_shared[((((int)threadIdx.x) * 72) + 38)]));
-    conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[12] * kernel_shared[((((int)threadIdx.x) * 72) + 39)]));
-    conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[13] * kernel_shared[((((int)threadIdx.x) * 72) + 40)]));
-    conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[14] * kernel_shared[((((int)threadIdx.x) * 72) + 41)]));
-    conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[15] * kernel_shared[((((int)threadIdx.x) * 72) + 42)]));
-    conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[16] * kernel_shared[((((int)threadIdx.x) * 72) + 43)]));
-    conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[17] * kernel_shared[((((int)threadIdx.x) * 72) + 44)]));
-    conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[12] * kernel_shared[((((int)threadIdx.x) * 72) + 36)]));
-    conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[13] * kernel_shared[((((int)threadIdx.x) * 72) + 37)]));
-    conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[14] * kernel_shared[((((int)threadIdx.x) * 72) + 38)]));
-    conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[15] * kernel_shared[((((int)threadIdx.x) * 72) + 39)]));
-    conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[16] * kernel_shared[((((int)threadIdx.x) * 72) + 40)]));
-    conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[17] * kernel_shared[((((int)threadIdx.x) * 72) + 41)]));
-    conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[18] * kernel_shared[((((int)threadIdx.x) * 72) + 42)]));
-    conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[19] * kernel_shared[((((int)threadIdx.x) * 72) + 43)]));
-    conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[20] * kernel_shared[((((int)threadIdx.x) * 72) + 44)]));
-    conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[15] * kernel_shared[((((int)threadIdx.x) * 72) + 36)]));
-    conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[16] * kernel_shared[((((int)threadIdx.x) * 72) + 37)]));
-    conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[17] * kernel_shared[((((int)threadIdx.x) * 72) + 38)]));
-    conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[18] * kernel_shared[((((int)threadIdx.x) * 72) + 39)]));
-    conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[19] * kernel_shared[((((int)threadIdx.x) * 72) + 40)]));
-    conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[20] * kernel_shared[((((int)threadIdx.x) * 72) + 41)]));
-    conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[21] * kernel_shared[((((int)threadIdx.x) * 72) + 42)]));
-    conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[22] * kernel_shared[((((int)threadIdx.x) * 72) + 43)]));
-    conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[23] * kernel_shared[((((int)threadIdx.x) * 72) + 44)]));
-    conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[18] * kernel_shared[((((int)threadIdx.x) * 72) + 36)]));
-    conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[19] * kernel_shared[((((int)threadIdx.x) * 72) + 37)]));
-    conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[20] * kernel_shared[((((int)threadIdx.x) * 72) + 38)]));
-    conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[21] * kernel_shared[((((int)threadIdx.x) * 72) + 39)]));
-    conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[22] * kernel_shared[((((int)threadIdx.x) * 72) + 40)]));
-    conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[23] * kernel_shared[((((int)threadIdx.x) * 72) + 41)]));
-    conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[24] * kernel_shared[((((int)threadIdx.x) * 72) + 42)]));
-    conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[25] * kernel_shared[((((int)threadIdx.x) * 72) + 43)]));
-    conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[26] * kernel_shared[((((int)threadIdx.x) * 72) + 44)]));
-    conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[27] * kernel_shared[((((int)threadIdx.x) * 72) + 9)]));
-    conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[28] * kernel_shared[((((int)threadIdx.x) * 72) + 10)]));
-    conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[29] * kernel_shared[((((int)threadIdx.x) * 72) + 11)]));
-    conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[30] * kernel_shared[((((int)threadIdx.x) * 72) + 12)]));
-    conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[31] * kernel_shared[((((int)threadIdx.x) * 72) + 13)]));
-    conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[32] * kernel_shared[((((int)threadIdx.x) * 72) + 14)]));
-    conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[33] * kernel_shared[((((int)threadIdx.x) * 72) + 15)]));
-    conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[34] * kernel_shared[((((int)threadIdx.x) * 72) + 16)]));
-    conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[35] * kernel_shared[((((int)threadIdx.x) * 72) + 17)]));
-    conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[30] * kernel_shared[((((int)threadIdx.x) * 72) + 9)]));
-    conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[31] * kernel_shared[((((int)threadIdx.x) * 72) + 10)]));
-    conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[32] * kernel_shared[((((int)threadIdx.x) * 72) + 11)]));
-    conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[33] * kernel_shared[((((int)threadIdx.x) * 72) + 12)]));
-    conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[34] * kernel_shared[((((int)threadIdx.x) * 72) + 13)]));
-    conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[35] * kernel_shared[((((int)threadIdx.x) * 72) + 14)]));
-    conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[36] * kernel_shared[((((int)threadIdx.x) * 72) + 15)]));
-    conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[37] * kernel_shared[((((int)threadIdx.x) * 72) + 16)]));
-    conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[38] * kernel_shared[((((int)threadIdx.x) * 72) + 17)]));
-    conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[33] * kernel_shared[((((int)threadIdx.x) * 72) + 9)]));
-    conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[34] * kernel_shared[((((int)threadIdx.x) * 72) + 10)]));
-    conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[35] * kernel_shared[((((int)threadIdx.x) * 72) + 11)]));
-    conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[36] * kernel_shared[((((int)threadIdx.x) * 72) + 12)]));
-    conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[37] * kernel_shared[((((int)threadIdx.x) * 72) + 13)]));
-    conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[38] * kernel_shared[((((int)threadIdx.x) * 72) + 14)]));
-    conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[39] * kernel_shared[((((int)threadIdx.x) * 72) + 15)]));
-    conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[40] * kernel_shared[((((int)threadIdx.x) * 72) + 16)]));
-    conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[41] * kernel_shared[((((int)threadIdx.x) * 72) + 17)]));
-    conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[36] * kernel_shared[((((int)threadIdx.x) * 72) + 9)]));
-    conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[37] * kernel_shared[((((int)threadIdx.x) * 72) + 10)]));
-    conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[38] * kernel_shared[((((int)threadIdx.x) * 72) + 11)]));
-    conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[39] * kernel_shared[((((int)threadIdx.x) * 72) + 12)]));
-    conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[40] * kernel_shared[((((int)threadIdx.x) * 72) + 13)]));
-    conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[41] * kernel_shared[((((int)threadIdx.x) * 72) + 14)]));
-    conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[42] * kernel_shared[((((int)threadIdx.x) * 72) + 15)]));
-    conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[43] * kernel_shared[((((int)threadIdx.x) * 72) + 16)]));
-    conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[44] * kernel_shared[((((int)threadIdx.x) * 72) + 17)]));
-    conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[39] * kernel_shared[((((int)threadIdx.x) * 72) + 9)]));
-    conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[40] * kernel_shared[((((int)threadIdx.x) * 72) + 10)]));
-    conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[41] * kernel_shared[((((int)threadIdx.x) * 72) + 11)]));
-    conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[42] * kernel_shared[((((int)threadIdx.x) * 72) + 12)]));
-    conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[43] * kernel_shared[((((int)threadIdx.x) * 72) + 13)]));
-    conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[44] * kernel_shared[((((int)threadIdx.x) * 72) + 14)]));
-    conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[45] * kernel_shared[((((int)threadIdx.x) * 72) + 15)]));
-    conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[46] * kernel_shared[((((int)threadIdx.x) * 72) + 16)]));
-    conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[47] * kernel_shared[((((int)threadIdx.x) * 72) + 17)]));
-    conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[42] * kernel_shared[((((int)threadIdx.x) * 72) + 9)]));
-    conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[43] * kernel_shared[((((int)threadIdx.x) * 72) + 10)]));
-    conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[44] * kernel_shared[((((int)threadIdx.x) * 72) + 11)]));
-    conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[45] * kernel_shared[((((int)threadIdx.x) * 72) + 12)]));
-    conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[46] * kernel_shared[((((int)threadIdx.x) * 72) + 13)]));
-    conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[47] * kernel_shared[((((int)threadIdx.x) * 72) + 14)]));
-    conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[48] * kernel_shared[((((int)threadIdx.x) * 72) + 15)]));
-    conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[49] * kernel_shared[((((int)threadIdx.x) * 72) + 16)]));
-    conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[50] * kernel_shared[((((int)threadIdx.x) * 72) + 17)]));
-    conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[45] * kernel_shared[((((int)threadIdx.x) * 72) + 9)]));
-    conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[46] * kernel_shared[((((int)threadIdx.x) * 72) + 10)]));
-    conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[47] * kernel_shared[((((int)threadIdx.x) * 72) + 11)]));
-    conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[48] * kernel_shared[((((int)threadIdx.x) * 72) + 12)]));
-    conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[49] * kernel_shared[((((int)threadIdx.x) * 72) + 13)]));
-    conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[50] * kernel_shared[((((int)threadIdx.x) * 72) + 14)]));
-    conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[51] * kernel_shared[((((int)threadIdx.x) * 72) + 15)]));
-    conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[52] * kernel_shared[((((int)threadIdx.x) * 72) + 16)]));
-    conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[53] * kernel_shared[((((int)threadIdx.x) * 72) + 17)]));
-    conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[27] * kernel_shared[((((int)threadIdx.x) * 72) + 45)]));
-    conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[28] * kernel_shared[((((int)threadIdx.x) * 72) + 46)]));
-    conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[29] * kernel_shared[((((int)threadIdx.x) * 72) + 47)]));
-    conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[30] * kernel_shared[((((int)threadIdx.x) * 72) + 48)]));
-    conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[31] * kernel_shared[((((int)threadIdx.x) * 72) + 49)]));
-    conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[32] * kernel_shared[((((int)threadIdx.x) * 72) + 50)]));
-    conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[33] * kernel_shared[((((int)threadIdx.x) * 72) + 51)]));
-    conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[34] * kernel_shared[((((int)threadIdx.x) * 72) + 52)]));
-    conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[35] * kernel_shared[((((int)threadIdx.x) * 72) + 53)]));
-    conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[30] * kernel_shared[((((int)threadIdx.x) * 72) + 45)]));
-    conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[31] * kernel_shared[((((int)threadIdx.x) * 72) + 46)]));
-    conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[32] * kernel_shared[((((int)threadIdx.x) * 72) + 47)]));
-    conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[33] * kernel_shared[((((int)threadIdx.x) * 72) + 48)]));
-    conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[34] * kernel_shared[((((int)threadIdx.x) * 72) + 49)]));
-    conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[35] * kernel_shared[((((int)threadIdx.x) * 72) + 50)]));
-    conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[36] * kernel_shared[((((int)threadIdx.x) * 72) + 51)]));
-    conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[37] * kernel_shared[((((int)threadIdx.x) * 72) + 52)]));
-    conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[38] * kernel_shared[((((int)threadIdx.x) * 72) + 53)]));
-    conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[33] * kernel_shared[((((int)threadIdx.x) * 72) + 45)]));
-    conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[34] * kernel_shared[((((int)threadIdx.x) * 72) + 46)]));
-    conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[35] * kernel_shared[((((int)threadIdx.x) * 72) + 47)]));
-    conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[36] * kernel_shared[((((int)threadIdx.x) * 72) + 48)]));
-    conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[37] * kernel_shared[((((int)threadIdx.x) * 72) + 49)]));
-    conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[38] * kernel_shared[((((int)threadIdx.x) * 72) + 50)]));
-    conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[39] * kernel_shared[((((int)threadIdx.x) * 72) + 51)]));
-    conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[40] * kernel_shared[((((int)threadIdx.x) * 72) + 52)]));
-    conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[41] * kernel_shared[((((int)threadIdx.x) * 72) + 53)]));
-    conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[36] * kernel_shared[((((int)threadIdx.x) * 72) + 45)]));
-    conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[37] * kernel_shared[((((int)threadIdx.x) * 72) + 46)]));
-    conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[38] * kernel_shared[((((int)threadIdx.x) * 72) + 47)]));
-    conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[39] * kernel_shared[((((int)threadIdx.x) * 72) + 48)]));
-    conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[40] * kernel_shared[((((int)threadIdx.x) * 72) + 49)]));
-    conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[41] * kernel_shared[((((int)threadIdx.x) * 72) + 50)]));
-    conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[42] * kernel_shared[((((int)threadIdx.x) * 72) + 51)]));
-    conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[43] * kernel_shared[((((int)threadIdx.x) * 72) + 52)]));
-    conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[44] * kernel_shared[((((int)threadIdx.x) * 72) + 53)]));
-    conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[39] * kernel_shared[((((int)threadIdx.x) * 72) + 45)]));
-    conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[40] * kernel_shared[((((int)threadIdx.x) * 72) + 46)]));
-    conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[41] * kernel_shared[((((int)threadIdx.x) * 72) + 47)]));
-    conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[42] * kernel_shared[((((int)threadIdx.x) * 72) + 48)]));
-    conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[43] * kernel_shared[((((int)threadIdx.x) * 72) + 49)]));
-    conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[44] * kernel_shared[((((int)threadIdx.x) * 72) + 50)]));
-    conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[45] * kernel_shared[((((int)threadIdx.x) * 72) + 51)]));
-    conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[46] * kernel_shared[((((int)threadIdx.x) * 72) + 52)]));
-    conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[47] * kernel_shared[((((int)threadIdx.x) * 72) + 53)]));
-    conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[42] * kernel_shared[((((int)threadIdx.x) * 72) + 45)]));
-    conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[43] * kernel_shared[((((int)threadIdx.x) * 72) + 46)]));
-    conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[44] * kernel_shared[((((int)threadIdx.x) * 72) + 47)]));
-    conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[45] * kernel_shared[((((int)threadIdx.x) * 72) + 48)]));
-    conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[46] * kernel_shared[((((int)threadIdx.x) * 72) + 49)]));
-    conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[47] * kernel_shared[((((int)threadIdx.x) * 72) + 50)]));
-    conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[48] * kernel_shared[((((int)threadIdx.x) * 72) + 51)]));
-    conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[49] * kernel_shared[((((int)threadIdx.x) * 72) + 52)]));
-    conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[50] * kernel_shared[((((int)threadIdx.x) * 72) + 53)]));
-    conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[45] * kernel_shared[((((int)threadIdx.x) * 72) + 45)]));
-    conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[46] * kernel_shared[((((int)threadIdx.x) * 72) + 46)]));
-    conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[47] * kernel_shared[((((int)threadIdx.x) * 72) + 47)]));
-    conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[48] * kernel_shared[((((int)threadIdx.x) * 72) + 48)]));
-    conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[49] * kernel_shared[((((int)threadIdx.x) * 72) + 49)]));
-    conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[50] * kernel_shared[((((int)threadIdx.x) * 72) + 50)]));
-    conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[51] * kernel_shared[((((int)threadIdx.x) * 72) + 51)]));
-    conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[52] * kernel_shared[((((int)threadIdx.x) * 72) + 52)]));
-    conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[53] * kernel_shared[((((int)threadIdx.x) * 72) + 53)]));
-    conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[54] * kernel_shared[((((int)threadIdx.x) * 72) + 18)]));
-    conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[55] * kernel_shared[((((int)threadIdx.x) * 72) + 19)]));
-    conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[56] * kernel_shared[((((int)threadIdx.x) * 72) + 20)]));
-    conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[57] * kernel_shared[((((int)threadIdx.x) * 72) + 21)]));
-    conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[58] * kernel_shared[((((int)threadIdx.x) * 72) + 22)]));
-    conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[59] * kernel_shared[((((int)threadIdx.x) * 72) + 23)]));
-    conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[60] * kernel_shared[((((int)threadIdx.x) * 72) + 24)]));
-    conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[61] * kernel_shared[((((int)threadIdx.x) * 72) + 25)]));
-    conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[62] * kernel_shared[((((int)threadIdx.x) * 72) + 26)]));
-    conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[57] * kernel_shared[((((int)threadIdx.x) * 72) + 18)]));
-    conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[58] * kernel_shared[((((int)threadIdx.x) * 72) + 19)]));
-    conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[59] * kernel_shared[((((int)threadIdx.x) * 72) + 20)]));
-    conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[60] * kernel_shared[((((int)threadIdx.x) * 72) + 21)]));
-    conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[61] * kernel_shared[((((int)threadIdx.x) * 72) + 22)]));
-    conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[62] * kernel_shared[((((int)threadIdx.x) * 72) + 23)]));
-    conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[63] * kernel_shared[((((int)threadIdx.x) * 72) + 24)]));
-    conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[64] * kernel_shared[((((int)threadIdx.x) * 72) + 25)]));
-    conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[65] * kernel_shared[((((int)threadIdx.x) * 72) + 26)]));
-    conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[60] * kernel_shared[((((int)threadIdx.x) * 72) + 18)]));
-    conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[61] * kernel_shared[((((int)threadIdx.x) * 72) + 19)]));
-    conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[62] * kernel_shared[((((int)threadIdx.x) * 72) + 20)]));
-    conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[63] * kernel_shared[((((int)threadIdx.x) * 72) + 21)]));
-    conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[64] * kernel_shared[((((int)threadIdx.x) * 72) + 22)]));
-    conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[65] * kernel_shared[((((int)threadIdx.x) * 72) + 23)]));
-    conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[66] * kernel_shared[((((int)threadIdx.x) * 72) + 24)]));
-    conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[67] * kernel_shared[((((int)threadIdx.x) * 72) + 25)]));
-    conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[68] * kernel_shared[((((int)threadIdx.x) * 72) + 26)]));
-    conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[63] * kernel_shared[((((int)threadIdx.x) * 72) + 18)]));
-    conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[64] * kernel_shared[((((int)threadIdx.x) * 72) + 19)]));
-    conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[65] * kernel_shared[((((int)threadIdx.x) * 72) + 20)]));
-    conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[66] * kernel_shared[((((int)threadIdx.x) * 72) + 21)]));
-    conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[67] * kernel_shared[((((int)threadIdx.x) * 72) + 22)]));
-    conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[68] * kernel_shared[((((int)threadIdx.x) * 72) + 23)]));
-    conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[69] * kernel_shared[((((int)threadIdx.x) * 72) + 24)]));
-    conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[70] * kernel_shared[((((int)threadIdx.x) * 72) + 25)]));
-    conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[71] * kernel_shared[((((int)threadIdx.x) * 72) + 26)]));
-    conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[66] * kernel_shared[((((int)threadIdx.x) * 72) + 18)]));
-    conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[67] * kernel_shared[((((int)threadIdx.x) * 72) + 19)]));
-    conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[68] * kernel_shared[((((int)threadIdx.x) * 72) + 20)]));
-    conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[69] * kernel_shared[((((int)threadIdx.x) * 72) + 21)]));
-    conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[70] * kernel_shared[((((int)threadIdx.x) * 72) + 22)]));
-    conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[71] * kernel_shared[((((int)threadIdx.x) * 72) + 23)]));
-    conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[72] * kernel_shared[((((int)threadIdx.x) * 72) + 24)]));
-    conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[73] * kernel_shared[((((int)threadIdx.x) * 72) + 25)]));
-    conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[74] * kernel_shared[((((int)threadIdx.x) * 72) + 26)]));
-    conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[69] * kernel_shared[((((int)threadIdx.x) * 72) + 18)]));
-    conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[70] * kernel_shared[((((int)threadIdx.x) * 72) + 19)]));
-    conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[71] * kernel_shared[((((int)threadIdx.x) * 72) + 20)]));
-    conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[72] * kernel_shared[((((int)threadIdx.x) * 72) + 21)]));
-    conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[73] * kernel_shared[((((int)threadIdx.x) * 72) + 22)]));
-    conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[74] * kernel_shared[((((int)threadIdx.x) * 72) + 23)]));
-    conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[75] * kernel_shared[((((int)threadIdx.x) * 72) + 24)]));
-    conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[76] * kernel_shared[((((int)threadIdx.x) * 72) + 25)]));
-    conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[77] * kernel_shared[((((int)threadIdx.x) * 72) + 26)]));
-    conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[72] * kernel_shared[((((int)threadIdx.x) * 72) + 18)]));
-    conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[73] * kernel_shared[((((int)threadIdx.x) * 72) + 19)]));
-    conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[74] * kernel_shared[((((int)threadIdx.x) * 72) + 20)]));
-    conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[75] * kernel_shared[((((int)threadIdx.x) * 72) + 21)]));
-    conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[76] * kernel_shared[((((int)threadIdx.x) * 72) + 22)]));
-    conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[77] * kernel_shared[((((int)threadIdx.x) * 72) + 23)]));
-    conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[78] * kernel_shared[((((int)threadIdx.x) * 72) + 24)]));
-    conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[79] * kernel_shared[((((int)threadIdx.x) * 72) + 25)]));
-    conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[80] * kernel_shared[((((int)threadIdx.x) * 72) + 26)]));
-    conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[54] * kernel_shared[((((int)threadIdx.x) * 72) + 54)]));
-    conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[55] * kernel_shared[((((int)threadIdx.x) * 72) + 55)]));
-    conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[56] * kernel_shared[((((int)threadIdx.x) * 72) + 56)]));
-    conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[57] * kernel_shared[((((int)threadIdx.x) * 72) + 57)]));
-    conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[58] * kernel_shared[((((int)threadIdx.x) * 72) + 58)]));
-    conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[59] * kernel_shared[((((int)threadIdx.x) * 72) + 59)]));
-    conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[60] * kernel_shared[((((int)threadIdx.x) * 72) + 60)]));
-    conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[61] * kernel_shared[((((int)threadIdx.x) * 72) + 61)]));
-    conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[62] * kernel_shared[((((int)threadIdx.x) * 72) + 62)]));
-    conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[57] * kernel_shared[((((int)threadIdx.x) * 72) + 54)]));
-    conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[58] * kernel_shared[((((int)threadIdx.x) * 72) + 55)]));
-    conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[59] * kernel_shared[((((int)threadIdx.x) * 72) + 56)]));
-    conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[60] * kernel_shared[((((int)threadIdx.x) * 72) + 57)]));
-    conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[61] * kernel_shared[((((int)threadIdx.x) * 72) + 58)]));
-    conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[62] * kernel_shared[((((int)threadIdx.x) * 72) + 59)]));
-    conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[63] * kernel_shared[((((int)threadIdx.x) * 72) + 60)]));
-    conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[64] * kernel_shared[((((int)threadIdx.x) * 72) + 61)]));
-    conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[65] * kernel_shared[((((int)threadIdx.x) * 72) + 62)]));
-    conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[60] * kernel_shared[((((int)threadIdx.x) * 72) + 54)]));
-    conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[61] * kernel_shared[((((int)threadIdx.x) * 72) + 55)]));
-    conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[62] * kernel_shared[((((int)threadIdx.x) * 72) + 56)]));
-    conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[63] * kernel_shared[((((int)threadIdx.x) * 72) + 57)]));
-    conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[64] * kernel_shared[((((int)threadIdx.x) * 72) + 58)]));
-    conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[65] * kernel_shared[((((int)threadIdx.x) * 72) + 59)]));
-    conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[66] * kernel_shared[((((int)threadIdx.x) * 72) + 60)]));
-    conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[67] * kernel_shared[((((int)threadIdx.x) * 72) + 61)]));
-    conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[68] * kernel_shared[((((int)threadIdx.x) * 72) + 62)]));
-    conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[63] * kernel_shared[((((int)threadIdx.x) * 72) + 54)]));
-    conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[64] * kernel_shared[((((int)threadIdx.x) * 72) + 55)]));
-    conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[65] * kernel_shared[((((int)threadIdx.x) * 72) + 56)]));
-    conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[66] * kernel_shared[((((int)threadIdx.x) * 72) + 57)]));
-    conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[67] * kernel_shared[((((int)threadIdx.x) * 72) + 58)]));
-    conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[68] * kernel_shared[((((int)threadIdx.x) * 72) + 59)]));
-    conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[69] * kernel_shared[((((int)threadIdx.x) * 72) + 60)]));
-    conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[70] * kernel_shared[((((int)threadIdx.x) * 72) + 61)]));
-    conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[71] * kernel_shared[((((int)threadIdx.x) * 72) + 62)]));
-    conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[66] * kernel_shared[((((int)threadIdx.x) * 72) + 54)]));
-    conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[67] * kernel_shared[((((int)threadIdx.x) * 72) + 55)]));
-    conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[68] * kernel_shared[((((int)threadIdx.x) * 72) + 56)]));
-    conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[69] * kernel_shared[((((int)threadIdx.x) * 72) + 57)]));
-    conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[70] * kernel_shared[((((int)threadIdx.x) * 72) + 58)]));
-    conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[71] * kernel_shared[((((int)threadIdx.x) * 72) + 59)]));
-    conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[72] * kernel_shared[((((int)threadIdx.x) * 72) + 60)]));
-    conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[73] * kernel_shared[((((int)threadIdx.x) * 72) + 61)]));
-    conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[74] * kernel_shared[((((int)threadIdx.x) * 72) + 62)]));
-    conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[69] * kernel_shared[((((int)threadIdx.x) * 72) + 54)]));
-    conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[70] * kernel_shared[((((int)threadIdx.x) * 72) + 55)]));
-    conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[71] * kernel_shared[((((int)threadIdx.x) * 72) + 56)]));
-    conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[72] * kernel_shared[((((int)threadIdx.x) * 72) + 57)]));
-    conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[73] * kernel_shared[((((int)threadIdx.x) * 72) + 58)]));
-    conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[74] * kernel_shared[((((int)threadIdx.x) * 72) + 59)]));
-    conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[75] * kernel_shared[((((int)threadIdx.x) * 72) + 60)]));
-    conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[76] * kernel_shared[((((int)threadIdx.x) * 72) + 61)]));
-    conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[77] * kernel_shared[((((int)threadIdx.x) * 72) + 62)]));
-    conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[72] * kernel_shared[((((int)threadIdx.x) * 72) + 54)]));
-    conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[73] * kernel_shared[((((int)threadIdx.x) * 72) + 55)]));
-    conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[74] * kernel_shared[((((int)threadIdx.x) * 72) + 56)]));
-    conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[75] * kernel_shared[((((int)threadIdx.x) * 72) + 57)]));
-    conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[76] * kernel_shared[((((int)threadIdx.x) * 72) + 58)]));
-    conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[77] * kernel_shared[((((int)threadIdx.x) * 72) + 59)]));
-    conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[78] * kernel_shared[((((int)threadIdx.x) * 72) + 60)]));
-    conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[79] * kernel_shared[((((int)threadIdx.x) * 72) + 61)]));
-    conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[80] * kernel_shared[((((int)threadIdx.x) * 72) + 62)]));
-    conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[81] * kernel_shared[((((int)threadIdx.x) * 72) + 27)]));
-    conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[82] * kernel_shared[((((int)threadIdx.x) * 72) + 28)]));
-    conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[83] * kernel_shared[((((int)threadIdx.x) * 72) + 29)]));
-    conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[84] * kernel_shared[((((int)threadIdx.x) * 72) + 30)]));
-    conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[85] * kernel_shared[((((int)threadIdx.x) * 72) + 31)]));
-    conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[86] * kernel_shared[((((int)threadIdx.x) * 72) + 32)]));
-    conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[87] * kernel_shared[((((int)threadIdx.x) * 72) + 33)]));
-    conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[88] * kernel_shared[((((int)threadIdx.x) * 72) + 34)]));
-    conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[89] * kernel_shared[((((int)threadIdx.x) * 72) + 35)]));
-    conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[84] * kernel_shared[((((int)threadIdx.x) * 72) + 27)]));
-    conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[85] * kernel_shared[((((int)threadIdx.x) * 72) + 28)]));
-    conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[86] * kernel_shared[((((int)threadIdx.x) * 72) + 29)]));
-    conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[87] * kernel_shared[((((int)threadIdx.x) * 72) + 30)]));
-    conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[88] * kernel_shared[((((int)threadIdx.x) * 72) + 31)]));
-    conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[89] * kernel_shared[((((int)threadIdx.x) * 72) + 32)]));
-    conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[90] * kernel_shared[((((int)threadIdx.x) * 72) + 33)]));
-    conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[91] * kernel_shared[((((int)threadIdx.x) * 72) + 34)]));
-    conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[92] * kernel_shared[((((int)threadIdx.x) * 72) + 35)]));
-    conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[87] * kernel_shared[((((int)threadIdx.x) * 72) + 27)]));
-    conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[88] * kernel_shared[((((int)threadIdx.x) * 72) + 28)]));
-    conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[89] * kernel_shared[((((int)threadIdx.x) * 72) + 29)]));
-    conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[90] * kernel_shared[((((int)threadIdx.x) * 72) + 30)]));
-    conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[91] * kernel_shared[((((int)threadIdx.x) * 72) + 31)]));
-    conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[92] * kernel_shared[((((int)threadIdx.x) * 72) + 32)]));
-    conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[93] * kernel_shared[((((int)threadIdx.x) * 72) + 33)]));
-    conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[94] * kernel_shared[((((int)threadIdx.x) * 72) + 34)]));
-    conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[95] * kernel_shared[((((int)threadIdx.x) * 72) + 35)]));
-    conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[90] * kernel_shared[((((int)threadIdx.x) * 72) + 27)]));
-    conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[91] * kernel_shared[((((int)threadIdx.x) * 72) + 28)]));
-    conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[92] * kernel_shared[((((int)threadIdx.x) * 72) + 29)]));
-    conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[93] * kernel_shared[((((int)threadIdx.x) * 72) + 30)]));
-    conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[94] * kernel_shared[((((int)threadIdx.x) * 72) + 31)]));
-    conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[95] * kernel_shared[((((int)threadIdx.x) * 72) + 32)]));
-    conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[96] * kernel_shared[((((int)threadIdx.x) * 72) + 33)]));
-    conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[97] * kernel_shared[((((int)threadIdx.x) * 72) + 34)]));
-    conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[98] * kernel_shared[((((int)threadIdx.x) * 72) + 35)]));
-    conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[93] * kernel_shared[((((int)threadIdx.x) * 72) + 27)]));
-    conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[94] * kernel_shared[((((int)threadIdx.x) * 72) + 28)]));
-    conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[95] * kernel_shared[((((int)threadIdx.x) * 72) + 29)]));
-    conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[96] * kernel_shared[((((int)threadIdx.x) * 72) + 30)]));
-    conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[97] * kernel_shared[((((int)threadIdx.x) * 72) + 31)]));
-    conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[98] * kernel_shared[((((int)threadIdx.x) * 72) + 32)]));
-    conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[99] * kernel_shared[((((int)threadIdx.x) * 72) + 33)]));
-    conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[100] * kernel_shared[((((int)threadIdx.x) * 72) + 34)]));
-    conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[101] * kernel_shared[((((int)threadIdx.x) * 72) + 35)]));
-    conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[96] * kernel_shared[((((int)threadIdx.x) * 72) + 27)]));
-    conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[97] * kernel_shared[((((int)threadIdx.x) * 72) + 28)]));
-    conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[98] * kernel_shared[((((int)threadIdx.x) * 72) + 29)]));
-    conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[99] * kernel_shared[((((int)threadIdx.x) * 72) + 30)]));
-    conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[100] * kernel_shared[((((int)threadIdx.x) * 72) + 31)]));
-    conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[101] * kernel_shared[((((int)threadIdx.x) * 72) + 32)]));
-    conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[102] * kernel_shared[((((int)threadIdx.x) * 72) + 33)]));
-    conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[103] * kernel_shared[((((int)threadIdx.x) * 72) + 34)]));
-    conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[104] * kernel_shared[((((int)threadIdx.x) * 72) + 35)]));
-    conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[99] * kernel_shared[((((int)threadIdx.x) * 72) + 27)]));
-    conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[100] * kernel_shared[((((int)threadIdx.x) * 72) + 28)]));
-    conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[101] * kernel_shared[((((int)threadIdx.x) * 72) + 29)]));
-    conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[102] * kernel_shared[((((int)threadIdx.x) * 72) + 30)]));
-    conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[103] * kernel_shared[((((int)threadIdx.x) * 72) + 31)]));
-    conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[104] * kernel_shared[((((int)threadIdx.x) * 72) + 32)]));
-    conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[105] * kernel_shared[((((int)threadIdx.x) * 72) + 33)]));
-    conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[106] * kernel_shared[((((int)threadIdx.x) * 72) + 34)]));
-    conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[107] * kernel_shared[((((int)threadIdx.x) * 72) + 35)]));
-    conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[81] * kernel_shared[((((int)threadIdx.x) * 72) + 63)]));
-    conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[82] * kernel_shared[((((int)threadIdx.x) * 72) + 64)]));
-    conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[83] * kernel_shared[((((int)threadIdx.x) * 72) + 65)]));
-    conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[84] * kernel_shared[((((int)threadIdx.x) * 72) + 66)]));
-    conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[85] * kernel_shared[((((int)threadIdx.x) * 72) + 67)]));
-    conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[86] * kernel_shared[((((int)threadIdx.x) * 72) + 68)]));
-    conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[87] * kernel_shared[((((int)threadIdx.x) * 72) + 69)]));
-    conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[88] * kernel_shared[((((int)threadIdx.x) * 72) + 70)]));
-    conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[89] * kernel_shared[((((int)threadIdx.x) * 72) + 71)]));
-    conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[84] * kernel_shared[((((int)threadIdx.x) * 72) + 63)]));
-    conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[85] * kernel_shared[((((int)threadIdx.x) * 72) + 64)]));
-    conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[86] * kernel_shared[((((int)threadIdx.x) * 72) + 65)]));
-    conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[87] * kernel_shared[((((int)threadIdx.x) * 72) + 66)]));
-    conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[88] * kernel_shared[((((int)threadIdx.x) * 72) + 67)]));
-    conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[89] * kernel_shared[((((int)threadIdx.x) * 72) + 68)]));
-    conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[90] * kernel_shared[((((int)threadIdx.x) * 72) + 69)]));
-    conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[91] * kernel_shared[((((int)threadIdx.x) * 72) + 70)]));
-    conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[92] * kernel_shared[((((int)threadIdx.x) * 72) + 71)]));
-    conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[87] * kernel_shared[((((int)threadIdx.x) * 72) + 63)]));
-    conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[88] * kernel_shared[((((int)threadIdx.x) * 72) + 64)]));
-    conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[89] * kernel_shared[((((int)threadIdx.x) * 72) + 65)]));
-    conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[90] * kernel_shared[((((int)threadIdx.x) * 72) + 66)]));
-    conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[91] * kernel_shared[((((int)threadIdx.x) * 72) + 67)]));
-    conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[92] * kernel_shared[((((int)threadIdx.x) * 72) + 68)]));
-    conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[93] * kernel_shared[((((int)threadIdx.x) * 72) + 69)]));
-    conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[94] * kernel_shared[((((int)threadIdx.x) * 72) + 70)]));
-    conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[95] * kernel_shared[((((int)threadIdx.x) * 72) + 71)]));
-    conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[90] * kernel_shared[((((int)threadIdx.x) * 72) + 63)]));
-    conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[91] * kernel_shared[((((int)threadIdx.x) * 72) + 64)]));
-    conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[92] * kernel_shared[((((int)threadIdx.x) * 72) + 65)]));
-    conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[93] * kernel_shared[((((int)threadIdx.x) * 72) + 66)]));
-    conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[94] * kernel_shared[((((int)threadIdx.x) * 72) + 67)]));
-    conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[95] * kernel_shared[((((int)threadIdx.x) * 72) + 68)]));
-    conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[96] * kernel_shared[((((int)threadIdx.x) * 72) + 69)]));
-    conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[97] * kernel_shared[((((int)threadIdx.x) * 72) + 70)]));
-    conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[98] * kernel_shared[((((int)threadIdx.x) * 72) + 71)]));
-    conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[93] * kernel_shared[((((int)threadIdx.x) * 72) + 63)]));
-    conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[94] * kernel_shared[((((int)threadIdx.x) * 72) + 64)]));
-    conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[95] * kernel_shared[((((int)threadIdx.x) * 72) + 65)]));
-    conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[96] * kernel_shared[((((int)threadIdx.x) * 72) + 66)]));
-    conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[97] * kernel_shared[((((int)threadIdx.x) * 72) + 67)]));
-    conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[98] * kernel_shared[((((int)threadIdx.x) * 72) + 68)]));
-    conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[99] * kernel_shared[((((int)threadIdx.x) * 72) + 69)]));
-    conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[100] * kernel_shared[((((int)threadIdx.x) * 72) + 70)]));
-    conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[101] * kernel_shared[((((int)threadIdx.x) * 72) + 71)]));
-    conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[96] * kernel_shared[((((int)threadIdx.x) * 72) + 63)]));
-    conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[97] * kernel_shared[((((int)threadIdx.x) * 72) + 64)]));
-    conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[98] * kernel_shared[((((int)threadIdx.x) * 72) + 65)]));
-    conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[99] * kernel_shared[((((int)threadIdx.x) * 72) + 66)]));
-    conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[100] * kernel_shared[((((int)threadIdx.x) * 72) + 67)]));
-    conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[101] * kernel_shared[((((int)threadIdx.x) * 72) + 68)]));
-    conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[102] * kernel_shared[((((int)threadIdx.x) * 72) + 69)]));
-    conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[103] * kernel_shared[((((int)threadIdx.x) * 72) + 70)]));
-    conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[104] * kernel_shared[((((int)threadIdx.x) * 72) + 71)]));
-    conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[99] * kernel_shared[((((int)threadIdx.x) * 72) + 63)]));
-    conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[100] * kernel_shared[((((int)threadIdx.x) * 72) + 64)]));
-    conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[101] * kernel_shared[((((int)threadIdx.x) * 72) + 65)]));
-    conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[102] * kernel_shared[((((int)threadIdx.x) * 72) + 66)]));
-    conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[103] * kernel_shared[((((int)threadIdx.x) * 72) + 67)]));
-    conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[104] * kernel_shared[((((int)threadIdx.x) * 72) + 68)]));
-    conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[105] * kernel_shared[((((int)threadIdx.x) * 72) + 69)]));
-    conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[106] * kernel_shared[((((int)threadIdx.x) * 72) + 70)]));
-    conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[107] * kernel_shared[((((int)threadIdx.x) * 72) + 71)]));
   }
   for (int i1_inner = 0; i1_inner &lt; 2; ++i1_inner) {
-    for (int i2_inner = 0; i2_inner &lt; 7; ++i2_inner) {
-      compute[((((((((int)blockIdx.x) / 7) * 1568) + (((int)threadIdx.x) * 98)) + (i1_inner * 49)) + (i2_inner * 7)) + (((int)blockIdx.x) % 7))] = max((conv2d_nchw[((i1_inner * 7) + i2_inner)] + bias[((((((int)blockIdx.x) / 7) * 32) + (((int)threadIdx.x) * 2)) + i1_inner)]), 0.000000e+00f);
-    }
+    compute[(((((int)blockIdx.x) * 196) + (i1_inner * 49)) + ((int)threadIdx.x))] = max((conv2d_nchw[i1_inner] + bias[((((int)blockIdx.x) * 4) + i1_inner)]), 0.000000e+00f);
+    compute[((((((int)blockIdx.x) * 196) + (i1_inner * 49)) + ((int)threadIdx.x)) + 7)] = max((conv2d_nchw[(i1_inner + 2)] + bias[((((int)blockIdx.x) * 4) + i1_inner)]), 0.000000e+00f);
+    compute[((((((int)blockIdx.x) * 196) + (i1_inner * 49)) + ((int)threadIdx.x)) + 14)] = max((conv2d_nchw[(i1_inner + 4)] + bias[((((int)blockIdx.x) * 4) + i1_inner)]), 0.000000e+00f);
+    compute[((((((int)blockIdx.x) * 196) + (i1_inner * 49)) + ((int)threadIdx.x)) + 21)] = max((conv2d_nchw[(i1_inner + 6)] + bias[((((int)blockIdx.x) * 4) + i1_inner)]), 0.000000e+00f);
+    compute[((((((int)blockIdx.x) * 196) + (i1_inner * 49)) + ((int)threadIdx.x)) + 28)] = max((conv2d_nchw[(i1_inner + 8)] + bias[((((int)blockIdx.x) * 4) + i1_inner)]), 0.000000e+00f);
+    compute[((((((int)blockIdx.x) * 196) + (i1_inner * 49)) + ((int)threadIdx.x)) + 35)] = max((conv2d_nchw[(i1_inner + 10)] + bias[((((int)blockIdx.x) * 4) + i1_inner)]), 0.000000e+00f);
+    compute[((((((int)blockIdx.x) * 196) + (i1_inner * 49)) + ((int)threadIdx.x)) + 42)] = max((conv2d_nchw[(i1_inner + 12)] + bias[((((int)blockIdx.x) * 4) + i1_inner)]), 0.000000e+00f);
+    compute[((((((int)blockIdx.x) * 196) + (i1_inner * 49)) + ((int)threadIdx.x)) + 98)] = max((conv2d_nchw[(i1_inner + 14)] + bias[(((((int)blockIdx.x) * 4) + i1_inner) + 2)]), 0.000000e+00f);
+    compute[((((((int)blockIdx.x) * 196) + (i1_inner * 49)) + ((int)threadIdx.x)) + 105)] = max((conv2d_nchw[(i1_inner + 16)] + bias[(((((int)blockIdx.x) * 4) + i1_inner) + 2)]), 0.000000e+00f);
+    compute[((((((int)blockIdx.x) * 196) + (i1_inner * 49)) + ((int)threadIdx.x)) + 112)] = max((conv2d_nchw[(i1_inner + 18)] + bias[(((((int)blockIdx.x) * 4) + i1_inner) + 2)]), 0.000000e+00f);
+    compute[((((((int)blockIdx.x) * 196) + (i1_inner * 49)) + ((int)threadIdx.x)) + 119)] = max((conv2d_nchw[(i1_inner + 20)] + bias[(((((int)blockIdx.x) * 4) + i1_inner) + 2)]), 0.000000e+00f);
+    compute[((((((int)blockIdx.x) * 196) + (i1_inner * 49)) + ((int)threadIdx.x)) + 126)] = max((conv2d_nchw[(i1_inner + 22)] + bias[(((((int)blockIdx.x) * 4) + i1_inner) + 2)]), 0.000000e+00f);
+    compute[((((((int)blockIdx.x) * 196) + (i1_inner * 49)) + ((int)threadIdx.x)) + 133)] = max((conv2d_nchw[(i1_inner + 24)] + bias[(((((int)blockIdx.x) * 4) + i1_inner) + 2)]), 0.000000e+00f);
+    compute[((((((int)blockIdx.x) * 196) + (i1_inner * 49)) + ((int)threadIdx.x)) + 140)] = max((conv2d_nchw[(i1_inner + 26)] + bias[(((((int)blockIdx.x) * 4) + i1_inner) + 2)]), 0.000000e+00f);
   }
 }
 </pre></div>
@@ -1965,7 +1574,7 @@ 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> ( 3 minutes  11.157 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 3 minutes  10.386 seconds)</p>
 <div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-tune-with-autoscheduler-tune-conv2d-layer-cuda-py">
 <div class="sphx-glr-download sphx-glr-download-python docutils container">
 <p><a class="reference download internal" download="" href="../../_downloads/e3e540f3b477c0c52d8eb73e674e8ffd/tune_conv2d_layer_cuda.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">tune_conv2d_layer_cuda.py</span></code></a></p>
diff --git a/docs/how_to/tune_with_autoscheduler/tune_network_cuda.html b/docs/how_to/tune_with_autoscheduler/tune_network_cuda.html
index 5c879ff7f..d37db89c0 100644
--- a/docs/how_to/tune_with_autoscheduler/tune_network_cuda.html
+++ b/docs/how_to/tune_with_autoscheduler/tune_network_cuda.html
@@ -901,7 +901,7 @@ so we can read the log file and load the best schedules.</p>
 Evaluate inference time cost...
 Execution time summary:
  mean (ms)   median (ms)    max (ms)     min (ms)     std (ms)
-   9.7592       9.7663       9.7725       9.7386       0.0147
+   9.6540       9.6583       9.6798       9.6239       0.0230
 </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 74d522300..761429a8e 100644
--- a/docs/how_to/tune_with_autoscheduler/tune_network_x86.html
+++ b/docs/how_to/tune_with_autoscheduler/tune_network_x86.html
@@ -920,7 +920,7 @@ so we can read the log file and load the best schedules.</p>
 Evaluate inference time cost...
 Execution time summary:
  mean (ms)   median (ms)    max (ms)     min (ms)     std (ms)
-  762.3544     762.0270     763.7512     761.2850      1.0331
+  762.9937     760.9339     767.2873     760.7600      3.0368
 </pre></div>
 </div>
 </div>
@@ -942,7 +942,7 @@ to learn how to use the RPC Tracker and RPC Server.
 To use the RPC Tracker in auto-scheduler, replace the runner in <code class="code docutils literal notranslate"><span class="pre">TuningOptions</span></code>
 with <a class="reference internal" href="../../reference/api/python/auto_scheduler.html#tvm.auto_scheduler.RPCRunner" title="tvm.auto_scheduler.RPCRunner"><code class="xref any py py-class docutils literal notranslate"><span class="pre">auto_scheduler.RPCRunner</span></code></a>.</p></li>
 </ol>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes  21.967 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes  22.515 seconds)</p>
 <div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-tune-with-autoscheduler-tune-network-x86-py">
 <div class="sphx-glr-download sphx-glr-download-python docutils container">
 <p><a class="reference download internal" download="" href="../../_downloads/e416b94ca1090b0897c0f6e0df95b911/tune_network_x86.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">tune_network_x86.py</span></code></a></p>
diff --git a/docs/how_to/tune_with_autoscheduler/tune_sparse_x86.html b/docs/how_to/tune_with_autoscheduler/tune_sparse_x86.html
index dbb411752..7909271a6 100644
--- a/docs/how_to/tune_with_autoscheduler/tune_sparse_x86.html
+++ b/docs/how_to/tune_with_autoscheduler/tune_sparse_x86.html
@@ -620,28 +620,659 @@ layout transformation, parallelization, vectorization, unrolling, and operator f
              placeholder_4: Buffer(placeholder_14: Pointer(float32), float32, [65536], []),
              compute: Buffer(compute_2: Pointer(float32), float32, [65536], [])}
   buffer_map = {placeholder_5: placeholder, placeholder_6: placeholder_1, placeholder_7: placeholder_2, placeholder_8: placeholder_3, placeholder_9: placeholder_4, compute_1: compute}
-  preflattened_buffer_map = {compute_1: compute_3: Buffer(compute_2, float32, [128, 512], []), placeholder_8: placeholder_15: Buffer(placeholder_13, int32, [33], []), placeholder_5: placeholder_16: Buffer(placeholder_10, float32, [128, 256], []), placeholder_6: placeholder_17: Buffer(placeholder_11, float32, [4916, 16, 1], []), placeholder_7: placeholder_18: Buffer(placeholder_12, int32, [4916], []), placeholder_9: placeholder_19: Buffer(placeholder_14, float32, [128, 512], [])} {
-  for (i0.outer.i1.outer.fused: int32, 0, 16) &quot;parallel&quot; {
-    allocate(compute_4: Pointer(global float32), float32, [4096]), storage_scope = global {
-      for (i.outer.inner: int32, 0, 128) {
-        for (nb_j.inner: int32, 0, 2) {
-          for (j.init: int32, 0, 16) {
-            compute_5: Buffer(compute_4, float32, [4096], [])[(((i.outer.inner*32) + (nb_j.inner*16)) + j.init)] = 0f32
-          }
-          for (elem_idx: int32, 0, let cse_var_1: int32 = ((i0.outer.i1.outer.fused*2) + nb_j.inner) in (placeholder_3[(cse_var_1 + 1)] - placeholder_3[cse_var_1])) {
-            for (j: int32, 0, 16) {
-              let cse_var_3: int32 = ((i0.outer.i1.outer.fused*2) + nb_j.inner)
-              let cse_var_2: int32 = (((i.outer.inner*32) + (nb_j.inner*16)) + j)
-              compute_5[cse_var_2] = (compute_5[cse_var_2] + (placeholder_1[(((placeholder_3[cse_var_3]*16) + (elem_idx*16)) + j)]*max(placeholder[((i.outer.inner*256) + placeholder_2[(placeholder_3[cse_var_3] + elem_idx)])], 0f32)))
+  preflattened_buffer_map = {placeholder_5: placeholder_15: Buffer(placeholder_10, float32, [128, 256], []), placeholder_6: placeholder_16: Buffer(placeholder_11, float32, [4916, 16, 1], []), placeholder_8: placeholder_17: Buffer(placeholder_13, int32, [33], []), placeholder_9: placeholder_18: Buffer(placeholder_14, float32, [128, 512], []), placeholder_7: placeholder_19: Buffer(placeholder_12, int32, [4916], []), compute_1: compute_3: Buffer(compute_2, float32, [128, 512], [])} {
+  for (i0.outer.i1.outer.fused: int32, 0, 128) &quot;parallel&quot; {
+    allocate(compute_4: Pointer(global float32), float32, [512]), storage_scope = global {
+      for (i.outer.inner: int32, 0, 4) {
+        let cse_var_2: int32 = floormod(i0.outer.i1.outer.fused, 32)
+        let cse_var_1: int32 = (i.outer.inner*128)
+         {
+          compute_5: Buffer(compute_4, float32, [512], [])[cse_var_1] = 0f32
+          compute_5[(cse_var_1 + 1)] = 0f32
+          compute_5[(cse_var_1 + 2)] = 0f32
+          compute_5[(cse_var_1 + 3)] = 0f32
+          compute_5[(cse_var_1 + 4)] = 0f32
+          compute_5[(cse_var_1 + 5)] = 0f32
+          compute_5[(cse_var_1 + 6)] = 0f32
+          compute_5[(cse_var_1 + 7)] = 0f32
+          compute_5[(cse_var_1 + 8)] = 0f32
+          compute_5[(cse_var_1 + 9)] = 0f32
+          compute_5[(cse_var_1 + 10)] = 0f32
+          compute_5[(cse_var_1 + 11)] = 0f32
+          compute_5[(cse_var_1 + 12)] = 0f32
+          compute_5[(cse_var_1 + 13)] = 0f32
+          compute_5[(cse_var_1 + 14)] = 0f32
+          compute_5[(cse_var_1 + 15)] = 0f32
+          compute_5[(cse_var_1 + 16)] = 0f32
+          compute_5[(cse_var_1 + 17)] = 0f32
+          compute_5[(cse_var_1 + 18)] = 0f32
+          compute_5[(cse_var_1 + 19)] = 0f32
+          compute_5[(cse_var_1 + 20)] = 0f32
+          compute_5[(cse_var_1 + 21)] = 0f32
+          compute_5[(cse_var_1 + 22)] = 0f32
+          compute_5[(cse_var_1 + 23)] = 0f32
+          compute_5[(cse_var_1 + 24)] = 0f32
+          compute_5[(cse_var_1 + 25)] = 0f32
+          compute_5[(cse_var_1 + 26)] = 0f32
+          compute_5[(cse_var_1 + 27)] = 0f32
+          compute_5[(cse_var_1 + 28)] = 0f32
+          compute_5[(cse_var_1 + 29)] = 0f32
+          compute_5[(cse_var_1 + 30)] = 0f32
+          compute_5[(cse_var_1 + 31)] = 0f32
+          compute_5[(cse_var_1 + 32)] = 0f32
+          compute_5[(cse_var_1 + 33)] = 0f32
+          compute_5[(cse_var_1 + 34)] = 0f32
+          compute_5[(cse_var_1 + 35)] = 0f32
+          compute_5[(cse_var_1 + 36)] = 0f32
+          compute_5[(cse_var_1 + 37)] = 0f32
+          compute_5[(cse_var_1 + 38)] = 0f32
+          compute_5[(cse_var_1 + 39)] = 0f32
+          compute_5[(cse_var_1 + 40)] = 0f32
+          compute_5[(cse_var_1 + 41)] = 0f32
+          compute_5[(cse_var_1 + 42)] = 0f32
+          compute_5[(cse_var_1 + 43)] = 0f32
+          compute_5[(cse_var_1 + 44)] = 0f32
+          compute_5[(cse_var_1 + 45)] = 0f32
+          compute_5[(cse_var_1 + 46)] = 0f32
+          compute_5[(cse_var_1 + 47)] = 0f32
+          compute_5[(cse_var_1 + 48)] = 0f32
+          compute_5[(cse_var_1 + 49)] = 0f32
+          compute_5[(cse_var_1 + 50)] = 0f32
+          compute_5[(cse_var_1 + 51)] = 0f32
+          compute_5[(cse_var_1 + 52)] = 0f32
+          compute_5[(cse_var_1 + 53)] = 0f32
+          compute_5[(cse_var_1 + 54)] = 0f32
+          compute_5[(cse_var_1 + 55)] = 0f32
+          compute_5[(cse_var_1 + 56)] = 0f32
+          compute_5[(cse_var_1 + 57)] = 0f32
+          compute_5[(cse_var_1 + 58)] = 0f32
+          compute_5[(cse_var_1 + 59)] = 0f32
+          compute_5[(cse_var_1 + 60)] = 0f32
+          compute_5[(cse_var_1 + 61)] = 0f32
+          compute_5[(cse_var_1 + 62)] = 0f32
+          compute_5[(cse_var_1 + 63)] = 0f32
+          compute_5[(cse_var_1 + 64)] = 0f32
+          compute_5[(cse_var_1 + 65)] = 0f32
+          compute_5[(cse_var_1 + 66)] = 0f32
+          compute_5[(cse_var_1 + 67)] = 0f32
+          compute_5[(cse_var_1 + 68)] = 0f32
+          compute_5[(cse_var_1 + 69)] = 0f32
+          compute_5[(cse_var_1 + 70)] = 0f32
+          compute_5[(cse_var_1 + 71)] = 0f32
+          compute_5[(cse_var_1 + 72)] = 0f32
+          compute_5[(cse_var_1 + 73)] = 0f32
+          compute_5[(cse_var_1 + 74)] = 0f32
+          compute_5[(cse_var_1 + 75)] = 0f32
+          compute_5[(cse_var_1 + 76)] = 0f32
+          compute_5[(cse_var_1 + 77)] = 0f32
+          compute_5[(cse_var_1 + 78)] = 0f32
+          compute_5[(cse_var_1 + 79)] = 0f32
+          compute_5[(cse_var_1 + 80)] = 0f32
+          compute_5[(cse_var_1 + 81)] = 0f32
+          compute_5[(cse_var_1 + 82)] = 0f32
+          compute_5[(cse_var_1 + 83)] = 0f32
+          compute_5[(cse_var_1 + 84)] = 0f32
+          compute_5[(cse_var_1 + 85)] = 0f32
+          compute_5[(cse_var_1 + 86)] = 0f32
+          compute_5[(cse_var_1 + 87)] = 0f32
+          compute_5[(cse_var_1 + 88)] = 0f32
+          compute_5[(cse_var_1 + 89)] = 0f32
+          compute_5[(cse_var_1 + 90)] = 0f32
+          compute_5[(cse_var_1 + 91)] = 0f32
+          compute_5[(cse_var_1 + 92)] = 0f32
+          compute_5[(cse_var_1 + 93)] = 0f32
+          compute_5[(cse_var_1 + 94)] = 0f32
+          compute_5[(cse_var_1 + 95)] = 0f32
+          compute_5[(cse_var_1 + 96)] = 0f32
+          compute_5[(cse_var_1 + 97)] = 0f32
+          compute_5[(cse_var_1 + 98)] = 0f32
+          compute_5[(cse_var_1 + 99)] = 0f32
+          compute_5[(cse_var_1 + 100)] = 0f32
+          compute_5[(cse_var_1 + 101)] = 0f32
+          compute_5[(cse_var_1 + 102)] = 0f32
+          compute_5[(cse_var_1 + 103)] = 0f32
+          compute_5[(cse_var_1 + 104)] = 0f32
+          compute_5[(cse_var_1 + 105)] = 0f32
+          compute_5[(cse_var_1 + 106)] = 0f32
+          compute_5[(cse_var_1 + 107)] = 0f32
+          compute_5[(cse_var_1 + 108)] = 0f32
+          compute_5[(cse_var_1 + 109)] = 0f32
+          compute_5[(cse_var_1 + 110)] = 0f32
+          compute_5[(cse_var_1 + 111)] = 0f32
+          compute_5[(cse_var_1 + 112)] = 0f32
+          compute_5[(cse_var_1 + 113)] = 0f32
+          compute_5[(cse_var_1 + 114)] = 0f32
+          compute_5[(cse_var_1 + 115)] = 0f32
+          compute_5[(cse_var_1 + 116)] = 0f32
+          compute_5[(cse_var_1 + 117)] = 0f32
+          compute_5[(cse_var_1 + 118)] = 0f32
+          compute_5[(cse_var_1 + 119)] = 0f32
+          compute_5[(cse_var_1 + 120)] = 0f32
+          compute_5[(cse_var_1 + 121)] = 0f32
+          compute_5[(cse_var_1 + 122)] = 0f32
+          compute_5[(cse_var_1 + 123)] = 0f32
+          compute_5[(cse_var_1 + 124)] = 0f32
+          compute_5[(cse_var_1 + 125)] = 0f32
+          compute_5[(cse_var_1 + 126)] = 0f32
+          compute_5[(cse_var_1 + 127)] = 0f32
+          for (elem_idx: int32, 0, (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])) {
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              compute_5[cse_var_1] = (compute_5[cse_var_1] + (placeholder_1[((placeholder_3[cse_var_2]*16) + (elem_idx*16))]*max(placeholder[(((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)])], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_3: int32 = (cse_var_1 + 1)
+              compute_5[cse_var_3] = (compute_5[cse_var_3] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 1)]*max(placeholder[(((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)])], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_4: int32 = (cse_var_1 + 2)
+              compute_5[cse_var_4] = (compute_5[cse_var_4] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 2)]*max(placeholder[(((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)])], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_5: int32 = (cse_var_1 + 3)
+              compute_5[cse_var_5] = (compute_5[cse_var_5] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 3)]*max(placeholder[(((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)])], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_6: int32 = (cse_var_1 + 4)
+              compute_5[cse_var_6] = (compute_5[cse_var_6] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 4)]*max(placeholder[(((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)])], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_7: int32 = (cse_var_1 + 5)
+              compute_5[cse_var_7] = (compute_5[cse_var_7] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 5)]*max(placeholder[(((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)])], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_8: int32 = (cse_var_1 + 6)
+              compute_5[cse_var_8] = (compute_5[cse_var_8] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 6)]*max(placeholder[(((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)])], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_9: int32 = (cse_var_1 + 7)
+              compute_5[cse_var_9] = (compute_5[cse_var_9] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 7)]*max(placeholder[(((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)])], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_10: int32 = (cse_var_1 + 8)
+              compute_5[cse_var_10] = (compute_5[cse_var_10] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 8)]*max(placeholder[(((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)])], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_11: int32 = (cse_var_1 + 9)
+              compute_5[cse_var_11] = (compute_5[cse_var_11] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 9)]*max(placeholder[(((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)])], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_12: int32 = (cse_var_1 + 10)
+              compute_5[cse_var_12] = (compute_5[cse_var_12] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 10)]*max(placeholder[(((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)])], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_13: int32 = (cse_var_1 + 11)
+              compute_5[cse_var_13] = (compute_5[cse_var_13] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 11)]*max(placeholder[(((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)])], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_14: int32 = (cse_var_1 + 12)
+              compute_5[cse_var_14] = (compute_5[cse_var_14] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 12)]*max(placeholder[(((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)])], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_15: int32 = (cse_var_1 + 13)
+              compute_5[cse_var_15] = (compute_5[cse_var_15] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 13)]*max(placeholder[(((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)])], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_16: int32 = (cse_var_1 + 14)
+              compute_5[cse_var_16] = (compute_5[cse_var_16] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 14)]*max(placeholder[(((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)])], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_17: int32 = (cse_var_1 + 15)
+              compute_5[cse_var_17] = (compute_5[cse_var_17] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 15)]*max(placeholder[(((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)])], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_18: int32 = (cse_var_1 + 16)
+              compute_5[cse_var_18] = (compute_5[cse_var_18] + (placeholder_1[((placeholder_3[cse_var_2]*16) + (elem_idx*16))]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 256)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_19: int32 = (cse_var_1 + 17)
+              compute_5[cse_var_19] = (compute_5[cse_var_19] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 1)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 256)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_20: int32 = (cse_var_1 + 18)
+              compute_5[cse_var_20] = (compute_5[cse_var_20] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 2)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 256)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_21: int32 = (cse_var_1 + 19)
+              compute_5[cse_var_21] = (compute_5[cse_var_21] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 3)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 256)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_22: int32 = (cse_var_1 + 20)
+              compute_5[cse_var_22] = (compute_5[cse_var_22] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 4)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 256)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_23: int32 = (cse_var_1 + 21)
+              compute_5[cse_var_23] = (compute_5[cse_var_23] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 5)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 256)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_24: int32 = (cse_var_1 + 22)
+              compute_5[cse_var_24] = (compute_5[cse_var_24] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 6)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 256)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_25: int32 = (cse_var_1 + 23)
+              compute_5[cse_var_25] = (compute_5[cse_var_25] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 7)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 256)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_26: int32 = (cse_var_1 + 24)
+              compute_5[cse_var_26] = (compute_5[cse_var_26] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 8)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 256)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_27: int32 = (cse_var_1 + 25)
+              compute_5[cse_var_27] = (compute_5[cse_var_27] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 9)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 256)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_28: int32 = (cse_var_1 + 26)
+              compute_5[cse_var_28] = (compute_5[cse_var_28] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 10)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 256)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_29: int32 = (cse_var_1 + 27)
+              compute_5[cse_var_29] = (compute_5[cse_var_29] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 11)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 256)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_30: int32 = (cse_var_1 + 28)
+              compute_5[cse_var_30] = (compute_5[cse_var_30] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 12)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 256)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_31: int32 = (cse_var_1 + 29)
+              compute_5[cse_var_31] = (compute_5[cse_var_31] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 13)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 256)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_32: int32 = (cse_var_1 + 30)
+              compute_5[cse_var_32] = (compute_5[cse_var_32] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 14)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 256)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_33: int32 = (cse_var_1 + 31)
+              compute_5[cse_var_33] = (compute_5[cse_var_33] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 15)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 256)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_34: int32 = (cse_var_1 + 32)
+              compute_5[cse_var_34] = (compute_5[cse_var_34] + (placeholder_1[((placeholder_3[cse_var_2]*16) + (elem_idx*16))]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 512)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_35: int32 = (cse_var_1 + 33)
+              compute_5[cse_var_35] = (compute_5[cse_var_35] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 1)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 512)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_36: int32 = (cse_var_1 + 34)
+              compute_5[cse_var_36] = (compute_5[cse_var_36] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 2)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 512)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_37: int32 = (cse_var_1 + 35)
+              compute_5[cse_var_37] = (compute_5[cse_var_37] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 3)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 512)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_38: int32 = (cse_var_1 + 36)
+              compute_5[cse_var_38] = (compute_5[cse_var_38] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 4)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 512)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_39: int32 = (cse_var_1 + 37)
+              compute_5[cse_var_39] = (compute_5[cse_var_39] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 5)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 512)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_40: int32 = (cse_var_1 + 38)
+              compute_5[cse_var_40] = (compute_5[cse_var_40] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 6)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 512)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_41: int32 = (cse_var_1 + 39)
+              compute_5[cse_var_41] = (compute_5[cse_var_41] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 7)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 512)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_42: int32 = (cse_var_1 + 40)
+              compute_5[cse_var_42] = (compute_5[cse_var_42] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 8)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 512)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_43: int32 = (cse_var_1 + 41)
+              compute_5[cse_var_43] = (compute_5[cse_var_43] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 9)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 512)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_44: int32 = (cse_var_1 + 42)
+              compute_5[cse_var_44] = (compute_5[cse_var_44] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 10)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 512)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_45: int32 = (cse_var_1 + 43)
+              compute_5[cse_var_45] = (compute_5[cse_var_45] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 11)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 512)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_46: int32 = (cse_var_1 + 44)
+              compute_5[cse_var_46] = (compute_5[cse_var_46] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 12)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 512)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_47: int32 = (cse_var_1 + 45)
+              compute_5[cse_var_47] = (compute_5[cse_var_47] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 13)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 512)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_48: int32 = (cse_var_1 + 46)
+              compute_5[cse_var_48] = (compute_5[cse_var_48] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 14)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 512)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_49: int32 = (cse_var_1 + 47)
+              compute_5[cse_var_49] = (compute_5[cse_var_49] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 15)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 512)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_50: int32 = (cse_var_1 + 48)
+              compute_5[cse_var_50] = (compute_5[cse_var_50] + (placeholder_1[((placeholder_3[cse_var_2]*16) + (elem_idx*16))]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 768)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_51: int32 = (cse_var_1 + 49)
+              compute_5[cse_var_51] = (compute_5[cse_var_51] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 1)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 768)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_52: int32 = (cse_var_1 + 50)
+              compute_5[cse_var_52] = (compute_5[cse_var_52] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 2)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 768)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_53: int32 = (cse_var_1 + 51)
+              compute_5[cse_var_53] = (compute_5[cse_var_53] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 3)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 768)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_54: int32 = (cse_var_1 + 52)
+              compute_5[cse_var_54] = (compute_5[cse_var_54] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 4)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 768)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_55: int32 = (cse_var_1 + 53)
+              compute_5[cse_var_55] = (compute_5[cse_var_55] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 5)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 768)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_56: int32 = (cse_var_1 + 54)
+              compute_5[cse_var_56] = (compute_5[cse_var_56] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 6)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 768)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_57: int32 = (cse_var_1 + 55)
+              compute_5[cse_var_57] = (compute_5[cse_var_57] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 7)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 768)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_58: int32 = (cse_var_1 + 56)
+              compute_5[cse_var_58] = (compute_5[cse_var_58] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 8)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 768)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_59: int32 = (cse_var_1 + 57)
+              compute_5[cse_var_59] = (compute_5[cse_var_59] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 9)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 768)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_60: int32 = (cse_var_1 + 58)
+              compute_5[cse_var_60] = (compute_5[cse_var_60] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 10)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 768)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_61: int32 = (cse_var_1 + 59)
+              compute_5[cse_var_61] = (compute_5[cse_var_61] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 11)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 768)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_62: int32 = (cse_var_1 + 60)
+              compute_5[cse_var_62] = (compute_5[cse_var_62] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 12)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 768)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_63: int32 = (cse_var_1 + 61)
+              compute_5[cse_var_63] = (compute_5[cse_var_63] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 13)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 768)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_64: int32 = (cse_var_1 + 62)
+              compute_5[cse_var_64] = (compute_5[cse_var_64] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 14)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 768)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_65: int32 = (cse_var_1 + 63)
+              compute_5[cse_var_65] = (compute_5[cse_var_65] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 15)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 768)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_66: int32 = (cse_var_1 + 64)
+              compute_5[cse_var_66] = (compute_5[cse_var_66] + (placeholder_1[((placeholder_3[cse_var_2]*16) + (elem_idx*16))]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1024)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_67: int32 = (cse_var_1 + 65)
+              compute_5[cse_var_67] = (compute_5[cse_var_67] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 1)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1024)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_68: int32 = (cse_var_1 + 66)
+              compute_5[cse_var_68] = (compute_5[cse_var_68] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 2)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1024)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_69: int32 = (cse_var_1 + 67)
+              compute_5[cse_var_69] = (compute_5[cse_var_69] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 3)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1024)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_70: int32 = (cse_var_1 + 68)
+              compute_5[cse_var_70] = (compute_5[cse_var_70] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 4)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1024)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_71: int32 = (cse_var_1 + 69)
+              compute_5[cse_var_71] = (compute_5[cse_var_71] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 5)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1024)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_72: int32 = (cse_var_1 + 70)
+              compute_5[cse_var_72] = (compute_5[cse_var_72] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 6)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1024)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_73: int32 = (cse_var_1 + 71)
+              compute_5[cse_var_73] = (compute_5[cse_var_73] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 7)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1024)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_74: int32 = (cse_var_1 + 72)
+              compute_5[cse_var_74] = (compute_5[cse_var_74] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 8)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1024)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_75: int32 = (cse_var_1 + 73)
+              compute_5[cse_var_75] = (compute_5[cse_var_75] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 9)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1024)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_76: int32 = (cse_var_1 + 74)
+              compute_5[cse_var_76] = (compute_5[cse_var_76] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 10)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1024)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_77: int32 = (cse_var_1 + 75)
+              compute_5[cse_var_77] = (compute_5[cse_var_77] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 11)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1024)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_78: int32 = (cse_var_1 + 76)
+              compute_5[cse_var_78] = (compute_5[cse_var_78] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 12)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1024)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_79: int32 = (cse_var_1 + 77)
+              compute_5[cse_var_79] = (compute_5[cse_var_79] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 13)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1024)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_80: int32 = (cse_var_1 + 78)
+              compute_5[cse_var_80] = (compute_5[cse_var_80] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 14)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1024)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_81: int32 = (cse_var_1 + 79)
+              compute_5[cse_var_81] = (compute_5[cse_var_81] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 15)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1024)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_82: int32 = (cse_var_1 + 80)
+              compute_5[cse_var_82] = (compute_5[cse_var_82] + (placeholder_1[((placeholder_3[cse_var_2]*16) + (elem_idx*16))]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1280)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_83: int32 = (cse_var_1 + 81)
+              compute_5[cse_var_83] = (compute_5[cse_var_83] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 1)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1280)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_84: int32 = (cse_var_1 + 82)
+              compute_5[cse_var_84] = (compute_5[cse_var_84] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 2)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1280)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_85: int32 = (cse_var_1 + 83)
+              compute_5[cse_var_85] = (compute_5[cse_var_85] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 3)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1280)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_86: int32 = (cse_var_1 + 84)
+              compute_5[cse_var_86] = (compute_5[cse_var_86] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 4)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1280)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_87: int32 = (cse_var_1 + 85)
+              compute_5[cse_var_87] = (compute_5[cse_var_87] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 5)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1280)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_88: int32 = (cse_var_1 + 86)
+              compute_5[cse_var_88] = (compute_5[cse_var_88] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 6)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1280)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_89: int32 = (cse_var_1 + 87)
+              compute_5[cse_var_89] = (compute_5[cse_var_89] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 7)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1280)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_90: int32 = (cse_var_1 + 88)
+              compute_5[cse_var_90] = (compute_5[cse_var_90] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 8)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1280)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_91: int32 = (cse_var_1 + 89)
+              compute_5[cse_var_91] = (compute_5[cse_var_91] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 9)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1280)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_92: int32 = (cse_var_1 + 90)
+              compute_5[cse_var_92] = (compute_5[cse_var_92] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 10)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1280)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_93: int32 = (cse_var_1 + 91)
+              compute_5[cse_var_93] = (compute_5[cse_var_93] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 11)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1280)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_94: int32 = (cse_var_1 + 92)
+              compute_5[cse_var_94] = (compute_5[cse_var_94] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 12)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1280)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_95: int32 = (cse_var_1 + 93)
+              compute_5[cse_var_95] = (compute_5[cse_var_95] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 13)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1280)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_96: int32 = (cse_var_1 + 94)
+              compute_5[cse_var_96] = (compute_5[cse_var_96] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 14)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1280)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_97: int32 = (cse_var_1 + 95)
+              compute_5[cse_var_97] = (compute_5[cse_var_97] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 15)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1280)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_98: int32 = (cse_var_1 + 96)
+              compute_5[cse_var_98] = (compute_5[cse_var_98] + (placeholder_1[((placeholder_3[cse_var_2]*16) + (elem_idx*16))]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1536)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_99: int32 = (cse_var_1 + 97)
+              compute_5[cse_var_99] = (compute_5[cse_var_99] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 1)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1536)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_100: int32 = (cse_var_1 + 98)
+              compute_5[cse_var_100] = (compute_5[cse_var_100] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 2)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1536)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_101: int32 = (cse_var_1 + 99)
+              compute_5[cse_var_101] = (compute_5[cse_var_101] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 3)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1536)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_102: int32 = (cse_var_1 + 100)
+              compute_5[cse_var_102] = (compute_5[cse_var_102] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 4)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1536)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_103: int32 = (cse_var_1 + 101)
+              compute_5[cse_var_103] = (compute_5[cse_var_103] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 5)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1536)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_104: int32 = (cse_var_1 + 102)
+              compute_5[cse_var_104] = (compute_5[cse_var_104] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 6)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1536)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_105: int32 = (cse_var_1 + 103)
+              compute_5[cse_var_105] = (compute_5[cse_var_105] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 7)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1536)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_106: int32 = (cse_var_1 + 104)
+              compute_5[cse_var_106] = (compute_5[cse_var_106] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 8)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1536)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_107: int32 = (cse_var_1 + 105)
+              compute_5[cse_var_107] = (compute_5[cse_var_107] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 9)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1536)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_108: int32 = (cse_var_1 + 106)
+              compute_5[cse_var_108] = (compute_5[cse_var_108] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 10)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1536)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_109: int32 = (cse_var_1 + 107)
+              compute_5[cse_var_109] = (compute_5[cse_var_109] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 11)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1536)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_110: int32 = (cse_var_1 + 108)
+              compute_5[cse_var_110] = (compute_5[cse_var_110] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 12)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1536)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_111: int32 = (cse_var_1 + 109)
+              compute_5[cse_var_111] = (compute_5[cse_var_111] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 13)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1536)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_112: int32 = (cse_var_1 + 110)
+              compute_5[cse_var_112] = (compute_5[cse_var_112] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 14)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1536)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_113: int32 = (cse_var_1 + 111)
+              compute_5[cse_var_113] = (compute_5[cse_var_113] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 15)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1536)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_114: int32 = (cse_var_1 + 112)
+              compute_5[cse_var_114] = (compute_5[cse_var_114] + (placeholder_1[((placeholder_3[cse_var_2]*16) + (elem_idx*16))]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1792)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_115: int32 = (cse_var_1 + 113)
+              compute_5[cse_var_115] = (compute_5[cse_var_115] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 1)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1792)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_116: int32 = (cse_var_1 + 114)
+              compute_5[cse_var_116] = (compute_5[cse_var_116] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 2)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1792)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_117: int32 = (cse_var_1 + 115)
+              compute_5[cse_var_117] = (compute_5[cse_var_117] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 3)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1792)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_118: int32 = (cse_var_1 + 116)
+              compute_5[cse_var_118] = (compute_5[cse_var_118] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 4)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1792)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_119: int32 = (cse_var_1 + 117)
+              compute_5[cse_var_119] = (compute_5[cse_var_119] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 5)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1792)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_120: int32 = (cse_var_1 + 118)
+              compute_5[cse_var_120] = (compute_5[cse_var_120] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 6)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1792)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_121: int32 = (cse_var_1 + 119)
+              compute_5[cse_var_121] = (compute_5[cse_var_121] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 7)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1792)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_122: int32 = (cse_var_1 + 120)
+              compute_5[cse_var_122] = (compute_5[cse_var_122] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 8)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1792)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_123: int32 = (cse_var_1 + 121)
+              compute_5[cse_var_123] = (compute_5[cse_var_123] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 9)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1792)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_124: int32 = (cse_var_1 + 122)
+              compute_5[cse_var_124] = (compute_5[cse_var_124] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 10)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1792)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_125: int32 = (cse_var_1 + 123)
+              compute_5[cse_var_125] = (compute_5[cse_var_125] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 11)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1792)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_126: int32 = (cse_var_1 + 124)
+              compute_5[cse_var_126] = (compute_5[cse_var_126] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 12)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1792)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_127: int32 = (cse_var_1 + 125)
+              compute_5[cse_var_127] = (compute_5[cse_var_127] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 13)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1792)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_128: int32 = (cse_var_1 + 126)
+              compute_5[cse_var_128] = (compute_5[cse_var_128] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 14)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1792)], 0f32)))
+            }
+            if @tir.likely((elem_idx &lt; (placeholder_3[(cse_var_2 + 1)] - placeholder_3[cse_var_2])), dtype=bool) {
+              let cse_var_129: int32 = (cse_var_1 + 127)
+              compute_5[cse_var_129] = (compute_5[cse_var_129] + (placeholder_1[(((placeholder_3[cse_var_2]*16) + (elem_idx*16)) + 15)]*max(placeholder[((((floordiv(i0.outer.i1.outer.fused, 32)*8192) + (i.outer.inner*2048)) + placeholder_2[(placeholder_3[cse_var_2] + elem_idx)]) + 1792)], 0f32)))
             }
           }
         }
       }
-      for (i0.inner: int32, 0, 128) {
-        for (i1.inner: int32, 0, 32) {
-          let cse_var_4: int32 = (((i0.inner*512) + (i0.outer.i1.outer.fused*32)) + i1.inner)
-          compute[cse_var_4] = max((compute_5[((i0.inner*32) + i1.inner)] + placeholder_4[cse_var_4]), 0f32)
-        }
+      for (i0.inner: int32, 0, 32) {
+        let cse_var_130: int32 = (((floordiv(i0.outer.i1.outer.fused, 32)*16384) + (i0.inner*512)) + (floormod(i0.outer.i1.outer.fused, 32)*16))
+        compute[ramp(cse_var_130, 1, 16)] = max((compute_5[ramp((i0.inner*16), 1, 16)] + placeholder_4[ramp(cse_var_130, 1, 16)]), broadcast(0f32, 16))
       }
     }
   }
@@ -679,7 +1310,7 @@ layout transformation, parallelization, vectorization, unrolling, and operator f
 <span class="p">)</span>
 </pre></div>
 </div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Execution time of this operator: 2.279 ms
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Execution time of this operator: 2.721 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 ed98d8878..352aa0cb8 100644
--- a/docs/how_to/tune_with_autotvm/sg_execution_times.html
+++ b/docs/how_to/tune_with_autotvm/sg_execution_times.html
@@ -322,7 +322,7 @@
             
   <div class="section" id="computation-times">
 <span id="sphx-glr-how-to-tune-with-autotvm-sg-execution-times"></span><h1>Computation times<a class="headerlink" href="#computation-times" title="Permalink to this headline">¶</a></h1>
-<p><strong>00:45.604</strong> total execution time for <strong>how_to_tune_with_autotvm</strong> files:</p>
+<p><strong>00:45.336</strong> total execution time for <strong>how_to_tune_with_autotvm</strong> files:</p>
 <table class="docutils align-default">
 <colgroup>
 <col style="width: 84%" />
@@ -331,7 +331,7 @@
 </colgroup>
 <tbody>
 <tr class="row-odd"><td><p><a class="reference internal" href="tune_conv2d_cuda.html#sphx-glr-how-to-tune-with-autotvm-tune-conv2d-cuda-py"><span class="std std-ref">Tuning High Performance Convolution on NVIDIA GPUs</span></a> (<code class="docutils literal notranslate"><span class="pre">tune_conv2d_cuda.py</span></code>)</p></td>
-<td><p>00:45.569</p></td>
+<td><p>00:45.301</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-even"><td><p><a class="reference internal" href="tune_relay_x86.html#sphx-glr-how-to-tune-with-autotvm-tune-relay-x86-py"><span class="std std-ref">Auto-tuning a Convolutional Network for x86 CPU</span></a> (<code class="docutils literal notranslate"><span class="pre">tune_relay_x86.py</span></code>)</p></td>
diff --git a/docs/how_to/tune_with_autotvm/tune_conv2d_cuda.html b/docs/how_to/tune_with_autotvm/tune_conv2d_cuda.html
index e02413f1c..2f4bea286 100644
--- a/docs/how_to/tune_with_autotvm/tune_conv2d_cuda.html
+++ b/docs/how_to/tune_with_autotvm/tune_conv2d_cuda.html
@@ -1431,8 +1431,8 @@ No: 8   GFLOPS: 0.00/0.00       result: Traceback (most recent call last):
 TimeoutError
 
         [(&#39;tile_f&#39;, [-1, 2, 1, 64]), (&#39;tile_y&#39;, [-1, 1, 1, 7]), (&#39;tile_x&#39;, [-1, 1, 7, 1]), (&#39;tile_rc&#39;, [-1, 1, 4]), (&#39;tile_ry&#39;, [-1, 3, 1]), (&#39;tile_rx&#39;, [-1, 1, 3]), (&#39;auto_unroll_max_step&#39;, 1500), (&#39;unroll_explicit&#39;, 0)],None,4909501
-No: 9   GFLOPS: 120.19/120.19   result: MeasureResult(costs=(0.0019261711071428573,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.9952189922332764, timestamp=1658794515.151304)       [(&#39;tile_f&#39;, [-1, 1, 4, 8]), (&#39;tile_y&#39;, [-1, 7, 1, 1]), (&#39;tile_x&#39;, [-1, 1, 1, 1]), (&#39;tile_rc&#39;, [-1, 2, 2]), (&#39;tile_ry&#39;, [-1, 1, 3]), (&#39;tile_rx&#39;, [-1, 1, 3]), (&#39;auto_unroll_max_step&#39;, 1500), (&#39;unroll_explicit&#39;, 0)],None,5072689
-No: 10  GFLOPS: 0.00/120.19     result: Traceback (most recent call last):
+No: 9   GFLOPS: 80.74/80.74     result: MeasureResult(costs=(0.002867357257142857,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.8408534526824951, timestamp=1658794692.6519725)       [(&#39;tile_f&#39;, [-1, 1, 4, 8]), (&#39;tile_y&#39;, [-1, 7, 1, 1]), (&#39;tile_x&#39;, [-1, 1, 1, 1]), (&#39;tile_rc&#39;, [-1, 2, 2]), (&#39;tile_ry&#39;, [-1, 1, 3]), (&#39;tile_rx&#39;, [-1, 1, 3]), (&#39;auto_unroll_max_step&#39;, 1500), (&#39;unroll_explicit&#39;, 0)],None,5072689
+No: 10  GFLOPS: 0.00/80.74      result: Traceback (most recent call last):
   File &quot;/workspace/python/tvm/autotvm/measure/measure_methods.py&quot;, line 588, in __call__
     func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
   File &quot;/workspace/python/tvm/autotvm/measure/measure_methods.py&quot;, line 540, in _build_func_common
@@ -1555,8 +1555,8 @@ Traceback (most recent call last):
   File &quot;/workspace/python/tvm/autotvm/measure/measure_methods.py&quot;, line 871, in verify_pass
     raise InstantiationError(&quot;Skipped because of invalid gpu kernel&quot;)
 tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel        [(&#39;tile_f&#39;, [-1, 4, 4, 8]), (&#39;tile_y&#39;, [-1, 1, 1, 1]), (&#39;tile_x&#39;, [-1, 1, 1, 7]), (&#39;tile_rc&#39;, [-1, 64, 2]), (&#39;tile_ry&#39;, [-1, 1, 3]), (&#39;tile_rx&#39;, [-1, 1, 3]), (&#39;auto_unroll_max_step&#39;, 1500), (&#39;unroll_explicit&#39;, 0)],None,5092711
-No: 11  GFLOPS: 261.18/261.18   result: MeasureResult(costs=(0.0008863828011049723,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.4688620567321777, timestamp=1658794516.0824132)      [(&#39;tile_f&#39;, [-1, 8, 2, 1]), (&#39;tile_y&#39;, [-1, 7, 1, 1]), (&#39;tile_x&#39;, [-1, 1, 7, 1]), (&#39;tile_rc&#39;, [-1, 2, 1]), (&#39;tile_ry&#39;, [-1, 3, 1]), (&#39;tile_rx&#39;, [-1, 3, 1]), (&#39;auto_unroll_max_step&#39;, 1500), (&#39;unroll_explicit&#39;, 0)],None,4264713
-No: 12  GFLOPS: 0.00/261.18     result: Traceback (most recent call last):
+No: 11  GFLOPS: 259.82/259.82   result: MeasureResult(costs=(0.0008910134088397789,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.4664301872253418, timestamp=1658794693.5753386)      [(&#39;tile_f&#39;, [-1, 8, 2, 1]), (&#39;tile_y&#39;, [-1, 7, 1, 1]), (&#39;tile_x&#39;, [-1, 1, 7, 1]), (&#39;tile_rc&#39;, [-1, 2, 1]), (&#39;tile_ry&#39;, [-1, 3, 1]), (&#39;tile_rx&#39;, [-1, 3, 1]), (&#39;auto_unroll_max_step&#39;, 1500), (&#39;unroll_explicit&#39;, 0)],None,4264713
+No: 12  GFLOPS: 0.00/259.82     result: Traceback (most recent call last):
   File &quot;/workspace/python/tvm/autotvm/measure/measure_methods.py&quot;, line 588, in __call__
     func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
   File &quot;/workspace/python/tvm/autotvm/measure/measure_methods.py&quot;, line 540, in _build_func_common
@@ -1679,7 +1679,7 @@ Traceback (most recent call last):
   File &quot;/workspace/python/tvm/autotvm/measure/measure_methods.py&quot;, line 871, in verify_pass
     raise InstantiationError(&quot;Skipped because of invalid gpu kernel&quot;)
 tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel        [(&#39;tile_f&#39;, [-1, 128, 1, 2]), (&#39;tile_y&#39;, [-1, 1, 7, 1]), (&#39;tile_x&#39;, [-1, 1, 1, 1]), (&#39;tile_rc&#39;, [-1, 1, 256]), (&#39;tile_ry&#39;, [-1, 1, 1]), (&#39;tile_rx&#39;, [-1, 1, 1]), (&#39;auto_unroll_max_step&#39;, 0), (&#39;unroll_explicit&#39;, 0)],None,183542
-No: 13  GFLOPS: 0.00/261.18     result: Traceback (most recent call last):
+No: 13  GFLOPS: 0.00/259.82     result: Traceback (most recent call last):
   File &quot;/workspace/python/tvm/autotvm/measure/measure_methods.py&quot;, line 588, in __call__
     func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
   File &quot;/workspace/python/tvm/autotvm/measure/measure_methods.py&quot;, line 540, in _build_func_common
@@ -1802,7 +1802,7 @@ Traceback (most recent call last):
   File &quot;/workspace/python/tvm/autotvm/measure/measure_methods.py&quot;, line 871, in verify_pass
     raise InstantiationError(&quot;Skipped because of invalid gpu kernel&quot;)
 tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel        [(&#39;tile_f&#39;, [-1, 4, 8, 8]), (&#39;tile_y&#39;, [-1, 1, 7, 1]), (&#39;tile_x&#39;, [-1, 1, 1, 1]), (&#39;tile_rc&#39;, [-1, 1, 64]), (&#39;tile_ry&#39;, [-1, 1, 1]), (&#39;tile_rx&#39;, [-1, 3, 1]), (&#39;auto_unroll_max_step&#39;, 512), (&#39;unroll_explicit&#39;, 0)],None,2482196
-No: 14  GFLOPS: 0.00/261.18     result: Traceback (most recent call last):
+No: 14  GFLOPS: 0.00/259.82     result: Traceback (most recent call last):
   File &quot;/workspace/python/tvm/autotvm/measure/measure_methods.py&quot;, line 588, in __call__
     func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
   File &quot;/workspace/python/tvm/autotvm/measure/measure_methods.py&quot;, line 540, in _build_func_common
@@ -1925,9 +1925,9 @@ Traceback (most recent call last):
   File &quot;/workspace/python/tvm/autotvm/measure/measure_methods.py&quot;, line 871, in verify_pass
     raise InstantiationError(&quot;Skipped because of invalid gpu kernel&quot;)
 tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel        [(&#39;tile_f&#39;, [-1, 64, 1, 4]), (&#39;tile_y&#39;, [-1, 1, 7, 1]), (&#39;tile_x&#39;, [-1, 1, 1, 7]), (&#39;tile_rc&#39;, [-1, 4, 2]), (&#39;tile_ry&#39;, [-1, 1, 3]), (&#39;tile_rx&#39;, [-1, 1, 3]), (&#39;auto_unroll_max_step&#39;, 1500), (&#39;unroll_explicit&#39;, 1)],None,10306226
-No: 15  GFLOPS: 5.29/261.18     result: MeasureResult(costs=(0.043745694,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.8309924602508545, timestamp=1658794520.5886626)        [(&#39;tile_f&#39;, [-1, 2, 2, 8]), (&#39;tile_y&#39;, [-1, 1, 1, 7]), (&#39;tile_x&#39;, [-1, 7, 1, 1]), (&#39;tile_rc&#39;, [-1, 4, 8]), (&#39;tile_ry&#39;, [-1, 1, 1]), (&#39;tile_rx&#39;, [-1, 1, 1]), (&#39;auto_unroll_max_step&#39;, 0), (&#39;unroll_explicit&#39;, 1)],None,5330964
-No: 16  GFLOPS: 3.34/261.18     result: MeasureResult(costs=(0.06941130225,), error_no=MeasureErrorNo.NO_ERROR, all_cost=4.529698848724365, timestamp=1658794521.8308659)       [(&#39;tile_f&#39;, [-1, 8, 4, 4]), (&#39;tile_y&#39;, [-1, 1, 1, 7]), (&#39;tile_x&#39;, [-1, 1, 1, 7]), (&#39;tile_rc&#39;, [-1, 4, 1]), (&#39;tile_ry&#39;, [-1, 1, 3]), (&#39;tile_rx&#39;, [-1, 1, 1]), (&#39;auto_unroll_max_step&#39;, 512), (&#39;unroll_explicit&#39;, 0)],None,2140058
-No: 17  GFLOPS: 0.00/261.18     result: Traceback (most recent call last):
+No: 15  GFLOPS: 5.29/259.82     result: MeasureResult(costs=(0.0437411905,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.825692892074585, timestamp=1658794698.0947344)        [(&#39;tile_f&#39;, [-1, 2, 2, 8]), (&#39;tile_y&#39;, [-1, 1, 1, 7]), (&#39;tile_x&#39;, [-1, 7, 1, 1]), (&#39;tile_rc&#39;, [-1, 4, 8]), (&#39;tile_ry&#39;, [-1, 1, 1]), (&#39;tile_rx&#39;, [-1, 1, 1]), (&#39;auto_unroll_max_step&#39;, 0), (&#39;unroll_explicit&#39;, 1)],None,5330964
+No: 16  GFLOPS: 3.34/259.82     result: MeasureResult(costs=(0.0693275815,), error_no=MeasureErrorNo.NO_ERROR, all_cost=4.529665470123291, timestamp=1658794699.329122) [(&#39;tile_f&#39;, [-1, 8, 4, 4]), (&#39;tile_y&#39;, [-1, 1, 1, 7]), (&#39;tile_x&#39;, [-1, 1, 1, 7]), (&#39;tile_rc&#39;, [-1, 4, 1]), (&#39;tile_ry&#39;, [-1, 1, 3]), (&#39;tile_rx&#39;, [-1, 1, 1]), (&#39;auto_unroll_max_step&#39;, 512), (&#39;unroll_explicit&#39;, 0)],None,2140058
+No: 17  GFLOPS: 0.00/259.82     result: Traceback (most recent call last):
   File &quot;/workspace/python/tvm/autotvm/measure/measure_methods.py&quot;, line 142, in build
     res = future.result()
   File &quot;/usr/lib/python3.7/concurrent/futures/_base.py&quot;, line 435, in result
@@ -1945,8 +1945,8 @@ No: 17  GFLOPS: 0.00/261.18     result: Traceback (most recent call last):
 TimeoutError
 
         [(&#39;tile_f&#39;, [-1, 2, 2, 1]), (&#39;tile_y&#39;, [-1, 1, 7, 1]), (&#39;tile_x&#39;, [-1, 7, 1, 1]), (&#39;tile_rc&#39;, [-1, 4, 16]), (&#39;tile_ry&#39;, [-1, 3, 1]), (&#39;tile_rx&#39;, [-1, 1, 3]), (&#39;auto_unroll_max_step&#39;, 1500), (&#39;unroll_explicit&#39;, 1)],None,10195251
-No: 18  GFLOPS: 27.94/261.18    result: MeasureResult(costs=(0.008284331857142857,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.269599199295044, timestamp=1658794532.8574402)        [(&#39;tile_f&#39;, [-1, 4, 8, 4]), (&#39;tile_y&#39;, [-1, 1, 1, 1]), (&#39;tile_x&#39;, [-1, 1, 1, 1]), (&#39;tile_rc&#39;, [-1, 1, 4]), (&#39;tile_ry&#39;, [-1, 3, 1]), (&#39;tile_rx&#39;, [-1, 3, 1]), (&#39;auto_unroll_max_step&#39;, 0), (&#39;unroll_explicit&#39;, 1)],None,6068603
-No: 19  GFLOPS: 0.00/261.18     result: Traceback (most recent call last):
+No: 18  GFLOPS: 28.14/259.82    result: MeasureResult(costs=(0.008227899714285714,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.2651729583740234, timestamp=1658794710.3470612)       [(&#39;tile_f&#39;, [-1, 4, 8, 4]), (&#39;tile_y&#39;, [-1, 1, 1, 1]), (&#39;tile_x&#39;, [-1, 1, 1, 1]), (&#39;tile_rc&#39;, [-1, 1, 4]), (&#39;tile_ry&#39;, [-1, 3, 1]), (&#39;tile_rx&#39;, [-1, 3, 1]), (&#39;auto_unroll_max_step&#39;, 0), (&#39;unroll_explicit&#39;, 1)],None,6068603
+No: 19  GFLOPS: 0.00/259.82     result: Traceback (most recent call last):
   File &quot;/workspace/python/tvm/autotvm/measure/measure_methods.py&quot;, line 588, in __call__
     func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
   File &quot;/workspace/python/tvm/autotvm/measure/measure_methods.py&quot;, line 540, in _build_func_common
@@ -2069,7 +2069,7 @@ Traceback (most recent call last):
   File &quot;/workspace/python/tvm/autotvm/measure/measure_methods.py&quot;, line 871, in verify_pass
     raise InstantiationError(&quot;Skipped because of invalid gpu kernel&quot;)
 tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel        [(&#39;tile_f&#39;, [-1, 16, 4, 8]), (&#39;tile_y&#39;, [-1, 1, 7, 1]), (&#39;tile_x&#39;, [-1, 7, 1, 1]), (&#39;tile_rc&#39;, [-1, 4, 128]), (&#39;tile_ry&#39;, [-1, 1, 3]), (&#39;tile_rx&#39;, [-1, 1, 3]), (&#39;auto_unroll_max_step&#39;, 0), (&#39;unroll_explicit&#39;, 1)],None,6956993
-No: 20  GFLOPS: 0.00/261.18     result: Traceback (most recent call last):
+No: 20  GFLOPS: 0.00/259.82     result: Traceback (most recent call last):
   File &quot;/workspace/python/tvm/autotvm/measure/measure_methods.py&quot;, line 588, in __call__
     func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
   File &quot;/workspace/python/tvm/autotvm/measure/measure_methods.py&quot;, line 540, in _build_func_common
@@ -2232,7 +2232,7 @@ and measure running time.</p>
 Best config:
 [(&#39;tile_f&#39;, [-1, 8, 2, 1]), (&#39;tile_y&#39;, [-1, 7, 1, 1]), (&#39;tile_x&#39;, [-1, 1, 7, 1]), (&#39;tile_rc&#39;, [-1, 2, 1]), (&#39;tile_ry&#39;, [-1, 3, 1]), (&#39;tile_rx&#39;, [-1, 3, 1]), (&#39;auto_unroll_max_step&#39;, 1500), (&#39;unroll_explicit&#39;, 0)],None,4264713
 Finish loading 20 records
-Time cost of this operator: 0.001283
+Time cost of this operator: 0.001279
 </pre></div>
 </div>
 <div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-tune-with-autotvm-tune-conv2d-cuda-py">
diff --git a/docs/how_to/work_with_microtvm/micro_autotune.html b/docs/how_to/work_with_microtvm/micro_autotune.html
index 34eade602..ff7ee7271 100644
--- a/docs/how_to/work_with_microtvm/micro_autotune.html
+++ b/docs/how_to/work_with_microtvm/micro_autotune.html
@@ -578,10 +578,10 @@ the tuned operator.</p>
 ########## Build without Autotuning ##########
 Node Name                                     Ops                                           Time(us)  Time(%)  Shape              Inputs  Outputs  Measurements(us)
 ---------                                     ---                                           --------  -------  -----              ------  -------  ----------------
-tvmgen_default_fused_nn_contrib_conv2d_NCHWc  tvmgen_default_fused_nn_contrib_conv2d_NCHWc  309.9     98.71    (1, 2, 10, 10, 3)  2       1        [309.9]
-tvmgen_default_fused_layout_transform_1       tvmgen_default_fused_layout_transform_1       3.078     0.98     (1, 6, 10, 10)     1       1        [3.078]
-tvmgen_default_fused_layout_transform         tvmgen_default_fused_layout_transform         0.97      0.309    (1, 1, 10, 10, 3)  1       1        [0.97]
-Total_time                                    -                                             313.948   -        -                  -       -        -
+tvmgen_default_fused_nn_contrib_conv2d_NCHWc  tvmgen_default_fused_nn_contrib_conv2d_NCHWc  311.4     98.726   (1, 2, 10, 10, 3)  2       1        [311.4]
+tvmgen_default_fused_layout_transform_1       tvmgen_default_fused_layout_transform_1       3.06      0.97     (1, 6, 10, 10)     1       1        [3.06]
+tvmgen_default_fused_layout_transform         tvmgen_default_fused_layout_transform         0.957     0.303    (1, 1, 10, 10, 3)  1       1        [0.957]
+Total_time                                    -                                             315.417   -        -                  -       -        -
 </pre></div>
 </div>
 </div>
@@ -634,10 +634,10 @@ Total_time                                    -
 ########## Build with Autotuning ##########
 Node Name                                     Ops                                           Time(us)  Time(%)  Shape              Inputs  Outputs  Measurements(us)
 ---------                                     ---                                           --------  -------  -----              ------  -------  ----------------
-tvmgen_default_fused_nn_contrib_conv2d_NCHWc  tvmgen_default_fused_nn_contrib_conv2d_NCHWc  80.438    96.718   (1, 6, 10, 10, 1)  2       1        [80.438]
-tvmgen_default_fused_layout_transform_1       tvmgen_default_fused_layout_transform_1       1.766     2.123    (1, 6, 10, 10)     1       1        [1.766]
-tvmgen_default_fused_layout_transform         tvmgen_default_fused_layout_transform         0.964     1.159    (1, 1, 10, 10, 3)  1       1        [0.964]
-Total_time                                    -                                             83.167    -        -                  -       -        -
+tvmgen_default_fused_nn_contrib_conv2d_NCHWc  tvmgen_default_fused_nn_contrib_conv2d_NCHWc  225.2     98.614   (1, 1, 10, 10, 6)  2       1        [225.2]
+tvmgen_default_fused_layout_transform_1       tvmgen_default_fused_layout_transform_1       2.186     0.957    (1, 6, 10, 10)     1       1        [2.186]
+tvmgen_default_fused_layout_transform         tvmgen_default_fused_layout_transform         0.978     0.428    (1, 1, 10, 10, 3)  1       1        [0.978]
+Total_time                                    -                                             228.364   -        -                  -       -        -
 </pre></div>
 </div>
 <div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-work-with-microtvm-micro-autotune-py">
diff --git a/docs/how_to/work_with_microtvm/micro_train.html b/docs/how_to/work_with_microtvm/micro_train.html
index a3d184c86..325893872 100644
--- a/docs/how_to/work_with_microtvm/micro_train.html
+++ b/docs/how_to/work_with_microtvm/micro_train.html
@@ -510,7 +510,7 @@ take about <strong>2 minutes</strong> to download the Stanford Cars, while COCO
 <a href="https://docs.python.org/3/library/shutil.html#shutil.move" title="shutil.move" class="sphx-glr-backref-module-shutil sphx-glr-backref-type-py-function"><span class="n">shutil</span><span class="o">.</span><span class="n">move</span></a><span class="p">(</span><span class="sa">f</span><span class="s2">&quot;</span><span class="si">{</span><a href="https://docs.python.org/3/library/stdtypes.html#str" title="builtins.str" class="sphx-glr-backref-module-builtins sphx-glr-backref-typ [...]
 </pre></div>
 </div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>&#39;/tmp/tmpp0m4t4w1/images/random&#39;
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>&#39;/tmp/tmp8hc9tzj6/images/random&#39;
 </pre></div>
 </div>
 </div>
@@ -570,8 +570,8 @@ objects to other stuff? We can display some examples from our datasets using <co
     <span class="n">plt</span><span class="o">.</span><span class="n">axis</span><span class="p">(</span><span class="s2">&quot;off&quot;</span><span class="p">)</span>
 </pre></div>
 </div>
-<img src="../../_images/sphx_glr_micro_train_001.png" srcset="../../_images/sphx_glr_micro_train_001.png" alt="[1.0, 0.0], [1.0, 0.0], [1.0, 0.0], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0], [1.0, 0.0], [0.0, 1.0], [1.0, 0.0]" class = "sphx-glr-single-img"/><div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>/tmp/tmpp0m4t4w1/images/target contains 8144 images
-/tmp/tmpp0m4t4w1/images/random contains 5000 images
+<img src="../../_images/sphx_glr_micro_train_001.png" srcset="../../_images/sphx_glr_micro_train_001.png" alt="[1.0, 0.0], [1.0, 0.0], [1.0, 0.0], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0], [1.0, 0.0], [0.0, 1.0], [1.0, 0.0]" class = "sphx-glr-single-img"/><div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>/tmp/tmp8hc9tzj6/images/target contains 8144 images
+/tmp/tmp8hc9tzj6/images/random contains 5000 images
 </pre></div>
 </div>
 </div>
@@ -683,13 +683,13 @@ the time on our validation set).</p>
 </pre></div>
 </div>
 <div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Epoch 1/3
-328/328 - 55s - loss: 0.2405 - accuracy: 0.9187 - val_loss: 0.1651 - val_accuracy: 0.9468
+328/328 - 55s - loss: 0.2231 - accuracy: 0.9217 - val_loss: 0.1474 - val_accuracy: 0.9524
 Epoch 2/3
-328/328 - 52s - loss: 0.1058 - accuracy: 0.9609 - val_loss: 0.1311 - val_accuracy: 0.9573
+328/328 - 52s - loss: 0.1028 - accuracy: 0.9599 - val_loss: 0.1210 - val_accuracy: 0.9596
 Epoch 3/3
-328/328 - 52s - loss: 0.0699 - accuracy: 0.9730 - val_loss: 0.1189 - val_accuracy: 0.9619
+328/328 - 52s - loss: 0.0690 - accuracy: 0.9739 - val_loss: 0.1083 - val_accuracy: 0.9637
 
-&lt;keras.callbacks.History object at 0x7f855e3d0110&gt;
+&lt;keras.callbacks.History object at 0x7fbc00a109d0&gt;
 </pre></div>
 </div>
 </div>
@@ -951,7 +951,7 @@ as intended.</p>
 <p>From here, we could modify the model to read live images from the camera - we have another
 Arduino tutorial for how to do that <a class="reference external" href="https://github.com/guberti/tvm-arduino-demos/tree/master/examples/person_detection">on GitHub</a>. Alternatively, we could also
 <a class="reference external" href="https://tvm.apache.org/docs/how_to/work_with_microtvm/micro_autotune.html">use TVM’s autotuning capabilities</a> to dramatically improve the model’s performance.</p>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 5 minutes  9.984 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 5 minutes  8.039 seconds)</p>
 <div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-work-with-microtvm-micro-train-py">
 <div class="sphx-glr-download sphx-glr-download-python docutils container">
 <p><a class="reference download internal" download="" href="../../_downloads/b52cec46baf4f78d6bcd94cbe269c8a6/micro_train.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">micro_train.py</span></code></a></p>
diff --git a/docs/how_to/work_with_microtvm/sg_execution_times.html b/docs/how_to/work_with_microtvm/sg_execution_times.html
index cf7aba526..8a672d60c 100644
--- a/docs/how_to/work_with_microtvm/sg_execution_times.html
+++ b/docs/how_to/work_with_microtvm/sg_execution_times.html
@@ -322,7 +322,7 @@
             
   <div class="section" id="computation-times">
 <span id="sphx-glr-how-to-work-with-microtvm-sg-execution-times"></span><h1>Computation times<a class="headerlink" href="#computation-times" title="Permalink to this headline">¶</a></h1>
-<p><strong>05:55.731</strong> total execution time for <strong>how_to_work_with_microtvm</strong> files:</p>
+<p><strong>05:54.504</strong> total execution time for <strong>how_to_work_with_microtvm</strong> files:</p>
 <table class="docutils align-default">
 <colgroup>
 <col style="width: 83%" />
@@ -331,15 +331,15 @@
 </colgroup>
 <tbody>
 <tr class="row-odd"><td><p><a class="reference internal" href="micro_train.html#sphx-glr-how-to-work-with-microtvm-micro-train-py"><span class="std std-ref">Training Vision Models for microTVM on Arduino</span></a> (<code class="docutils literal notranslate"><span class="pre">micro_train.py</span></code>)</p></td>
-<td><p>05:09.984</p></td>
+<td><p>05:08.039</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-even"><td><p><a class="reference internal" href="micro_autotune.html#sphx-glr-how-to-work-with-microtvm-micro-autotune-py"><span class="std std-ref">Autotuning with microTVM</span></a> (<code class="docutils literal notranslate"><span class="pre">micro_autotune.py</span></code>)</p></td>
-<td><p>00:42.489</p></td>
+<td><p>00:43.166</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-odd"><td><p><a class="reference internal" href="micro_tflite.html#sphx-glr-how-to-work-with-microtvm-micro-tflite-py"><span class="std std-ref">microTVM with TFLite Models</span></a> (<code class="docutils literal notranslate"><span class="pre">micro_tflite.py</span></code>)</p></td>
-<td><p>00:03.257</p></td>
+<td><p>00:03.297</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-even"><td><p><a class="reference internal" href="micro_ethosu.html#sphx-glr-how-to-work-with-microtvm-micro-ethosu-py"><span class="std std-ref">Running TVM on bare metal Arm(R) Cortex(R)-M55 CPU and Ethos(TM)-U55 NPU with CMSIS-NN</span></a> (<code class="docutils literal notranslate"><span class="pre">micro_ethosu.py</span></code>)</p></td>
diff --git a/docs/how_to/work_with_relay/sg_execution_times.html b/docs/how_to/work_with_relay/sg_execution_times.html
index 2a80f4cef..a95b0504c 100644
--- a/docs/how_to/work_with_relay/sg_execution_times.html
+++ b/docs/how_to/work_with_relay/sg_execution_times.html
@@ -322,7 +322,7 @@
             
   <div class="section" id="computation-times">
 <span id="sphx-glr-how-to-work-with-relay-sg-execution-times"></span><h1>Computation times<a class="headerlink" href="#computation-times" title="Permalink to this headline">¶</a></h1>
-<p><strong>00:42.004</strong> total execution time for <strong>how_to_work_with_relay</strong> files:</p>
+<p><strong>00:41.929</strong> total execution time for <strong>how_to_work_with_relay</strong> files:</p>
 <table class="docutils align-default">
 <colgroup>
 <col style="width: 84%" />
@@ -331,19 +331,19 @@
 </colgroup>
 <tbody>
 <tr class="row-odd"><td><p><a class="reference internal" href="using_pipeline_executor.html#sphx-glr-how-to-work-with-relay-using-pipeline-executor-py"><span class="std std-ref">Using Pipeline Executor in Relay</span></a> (<code class="docutils literal notranslate"><span class="pre">using_pipeline_executor.py</span></code>)</p></td>
-<td><p>00:30.456</p></td>
+<td><p>00:30.380</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-even"><td><p><a class="reference internal" href="using_external_lib.html#sphx-glr-how-to-work-with-relay-using-external-lib-py"><span class="std std-ref">Using External Libraries in Relay</span></a> (<code class="docutils literal notranslate"><span class="pre">using_external_lib.py</span></code>)</p></td>
-<td><p>00:09.916</p></td>
+<td><p>00:09.840</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-odd"><td><p><a class="reference internal" href="build_gcn.html#sphx-glr-how-to-work-with-relay-build-gcn-py"><span class="std std-ref">Building a Graph Convolutional Network</span></a> (<code class="docutils literal notranslate"><span class="pre">build_gcn.py</span></code>)</p></td>
-<td><p>00:01.626</p></td>
+<td><p>00:01.701</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-even"><td><p><a class="reference internal" href="using_relay_viz.html#sphx-glr-how-to-work-with-relay-using-relay-viz-py"><span class="std std-ref">Use Relay Visualizer to Visualize Relay</span></a> (<code class="docutils literal notranslate"><span class="pre">using_relay_viz.py</span></code>)</p></td>
-<td><p>00:00.007</p></td>
+<td><p>00:00.008</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 </tbody>
diff --git a/docs/how_to/work_with_schedules/intrin_math.html b/docs/how_to/work_with_schedules/intrin_math.html
index 12504f571..a16966d99 100644
--- a/docs/how_to/work_with_schedules/intrin_math.html
+++ b/docs/how_to/work_with_schedules/intrin_math.html
@@ -517,7 +517,7 @@ The following example customizes CUDA lowering rule for <code class="code docuti
 <a href="../../reference/api/python/ir.html#tvm.ir.register_intrin_lowering" title="tvm.ir.register_intrin_lowering" class="sphx-glr-backref-module-tvm-ir sphx-glr-backref-type-py-function"><span class="n">register_intrin_lowering</span></a><span class="p">(</span><span class="s2">&quot;tir.exp&quot;</span><span class="p">,</span> <span class="n">target</span><span class="o">=</span><span class="s2">&quot;cuda&quot;</span><span class="p">,</span> <span class="n">f</span><span class="o">= [...]
 </pre></div>
 </div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>&lt;function my_cuda_math_rule at 0x7f8556c6ce60&gt;
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>&lt;function my_cuda_math_rule at 0x7fbb85bdd050&gt;
 </pre></div>
 </div>
 <p>Register the rule to TVM with override option to override existing rule.
diff --git a/docs/how_to/work_with_schedules/sg_execution_times.html b/docs/how_to/work_with_schedules/sg_execution_times.html
index 47dabba17..535336123 100644
--- a/docs/how_to/work_with_schedules/sg_execution_times.html
+++ b/docs/how_to/work_with_schedules/sg_execution_times.html
@@ -322,7 +322,7 @@
             
   <div class="section" id="computation-times">
 <span id="sphx-glr-how-to-work-with-schedules-sg-execution-times"></span><h1>Computation times<a class="headerlink" href="#computation-times" title="Permalink to this headline">¶</a></h1>
-<p><strong>00:04.042</strong> total execution time for <strong>how_to_work_with_schedules</strong> files:</p>
+<p><strong>00:04.090</strong> total execution time for <strong>how_to_work_with_schedules</strong> files:</p>
 <table class="docutils align-default">
 <colgroup>
 <col style="width: 83%" />
@@ -331,27 +331,27 @@
 </colgroup>
 <tbody>
 <tr class="row-odd"><td><p><a class="reference internal" href="intrin_math.html#sphx-glr-how-to-work-with-schedules-intrin-math-py"><span class="std std-ref">Intrinsics and Math Functions</span></a> (<code class="docutils literal notranslate"><span class="pre">intrin_math.py</span></code>)</p></td>
-<td><p>00:01.893</p></td>
+<td><p>00:01.894</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-even"><td><p><a class="reference internal" href="tensorize.html#sphx-glr-how-to-work-with-schedules-tensorize-py"><span class="std std-ref">Use Tensorize to Leverage Hardware Intrinsics</span></a> (<code class="docutils literal notranslate"><span class="pre">tensorize.py</span></code>)</p></td>
-<td><p>00:00.922</p></td>
+<td><p>00:00.966</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-odd"><td><p><a class="reference internal" href="reduction.html#sphx-glr-how-to-work-with-schedules-reduction-py"><span class="std std-ref">Reduction</span></a> (<code class="docutils literal notranslate"><span class="pre">reduction.py</span></code>)</p></td>
-<td><p>00:00.528</p></td>
+<td><p>00:00.531</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-even"><td><p><a class="reference internal" href="scan.html#sphx-glr-how-to-work-with-schedules-scan-py"><span class="std std-ref">Scan and Recurrent Kernel</span></a> (<code class="docutils literal notranslate"><span class="pre">scan.py</span></code>)</p></td>
-<td><p>00:00.516</p></td>
+<td><p>00:00.514</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-odd"><td><p><a class="reference internal" href="extern_op.html#sphx-glr-how-to-work-with-schedules-extern-op-py"><span class="std std-ref">External Tensor Functions</span></a> (<code class="docutils literal notranslate"><span class="pre">extern_op.py</span></code>)</p></td>
-<td><p>00:00.100</p></td>
+<td><p>00:00.101</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-even"><td><p><a class="reference internal" href="schedule_primitives.html#sphx-glr-how-to-work-with-schedules-schedule-primitives-py"><span class="std std-ref">Schedule Primitives in TVM</span></a> (<code class="docutils literal notranslate"><span class="pre">schedule_primitives.py</span></code>)</p></td>
-<td><p>00:00.041</p></td>
+<td><p>00:00.042</p></td>
 <td><p>0.0 MB</p></td>
 </tr>
 <tr class="row-odd"><td><p><a class="reference internal" href="tedd.html#sphx-glr-how-to-work-with-schedules-tedd-py"><span class="std std-ref">Use Tensor Expression Debug Display (TEDD) for Visualization</span></a> (<code class="docutils literal notranslate"><span class="pre">tedd.py</span></code>)</p></td>
diff --git a/docs/how_to/work_with_schedules/tensorize.html b/docs/how_to/work_with_schedules/tensorize.html
index 39b0af49a..6309f8764 100644
--- a/docs/how_to/work_with_schedules/tensorize.html
+++ b/docs/how_to/work_with_schedules/tensorize.html
@@ -572,7 +572,7 @@ The importing needs to happen before the tensorized GEMV being executed.</p>
              C: Buffer(C_2: Pointer(float32), float32, [524288], [])}
   buffer_map = {A_1: A, B_1: B, C_1: C}
   preflattened_buffer_map = {A_1: A_3: Buffer(A_2, float32, [1024, 64], []), B_1: B_3: Buffer(B_2, float32, [512, 64], []), C_1: C_3: Buffer(C_2, float32, [1024, 512], [])} {
-  attr [IterVar(i: int32, (nullptr), &quot;DataPar&quot;, &quot;&quot;)] &quot;pragma_import_llvm&quot; = &quot;; ModuleID = &#39;/tmp/tmp1q_31ci7/input0.cc&#39;\nsource_filename = \&quot;/tmp/tmp1q_31ci7/input0.cc\&quot;\ntarget datalayout = \&quot;e-m:e-i64:64-f80:128-n8:16:32:64-S128\&quot;\ntarget triple = \&quot;x86_64-pc-linux-gnu\&quot;\n\n; Function Attrs: noinline nounwind optnone uwtable\ndefine dso_local i32 @gemv_update(float*, float*, float*, i32, i32, i32) #0 {\n  %7 = allo [...]
+  attr [IterVar(i: int32, (nullptr), &quot;DataPar&quot;, &quot;&quot;)] &quot;pragma_import_llvm&quot; = &quot;; ModuleID = &#39;/tmp/tmp2orpuey9/input0.cc&#39;\nsource_filename = \&quot;/tmp/tmp2orpuey9/input0.cc\&quot;\ntarget datalayout = \&quot;e-m:e-i64:64-f80:128-n8:16:32:64-S128\&quot;\ntarget triple = \&quot;x86_64-pc-linux-gnu\&quot;\n\n; Function Attrs: noinline nounwind optnone uwtable\ndefine dso_local i32 @gemv_update(float*, float*, float*, i32, i32, i32) #0 {\n  %7 = allo [...]
   for (i, 0, 1024) {
     for (j.outer: int32, 0, 32) {
       @tir.call_extern(&quot;gemv_update&quot;, @tir.tvm_access_ptr(@tir.type_annotation(, dtype=float32), C_2, ((i*512) + (j.outer*16)), 16, 2, dtype=handle), @tir.tvm_access_ptr(@tir.type_annotation(, dtype=float32), A_2, (i*64), 64, 1, dtype=handle), @tir.tvm_access_ptr(@tir.type_annotation(, dtype=float32), B_2, (j.outer*1024), 1024, 1, dtype=handle), 16, 64, 64, dtype=int32)
diff --git a/docs/reference/api/python/auto_scheduler.html b/docs/reference/api/python/auto_scheduler.html
index 9db5b9de9..a6a3c0127 100644
--- a/docs/reference/api/python/auto_scheduler.html
+++ b/docs/reference/api/python/auto_scheduler.html
@@ -1597,7 +1597,7 @@ history states as starting point to perform Evolutionary Search).</p></li>
 
 <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>
@@ -1881,7 +1881,7 @@ Candidates:
 
 <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">
diff --git a/docs/reference/api/typedoc/classes/bytestreamreader.html b/docs/reference/api/typedoc/classes/bytestreamreader.html
index f2d99cd95..644c73bb7 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/21d54f988/web/src/rpc_server.ts#L43">rpc_server.ts:43</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/rpc_server.ts#L43">rpc_server.ts:43</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/rpc_server.ts#L42">rpc_server.ts:42</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/rpc_server.ts#L63">rpc_server.ts:63</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/rpc_server.ts#L49">rpc_server.ts:49</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/rpc_server.ts#L57">rpc_server.ts:57</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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 6e058bbd0..2a7acf446 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/21d54f988/web/src/memory.ts#L223">memory.ts:223</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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">&lt;</span><a href="../interfaces/disposable.html" class="tsd-signature-type">Disposable</a><span class="tsd-signature-symbol">&gt;</span><span class="tsd-signature-symbol"> = []</span></div>
 					<aside class="tsd-sources">
 						<ul>
-							<li>Defined in <a href="https://github.com/apache/tvm/blob/21d54f988/web/src/memory.ts#L208">memory.ts:208</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/memory.ts#L312">memory.ts:312</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/memory.ts#L284">memory.ts:284</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/memory.ts#L388">memory.ts:388</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/memory.ts#L376">memory.ts:376</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/memory.ts#L267">memory.ts:267</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/memory.ts#L243">memory.ts:243</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/memory.ts#L321">memory.ts:321</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/memory.ts#L252">memory.ts:252</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/memory.ts#L359">memory.ts:359</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/memory.ts#L342">memory.ts:342</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/memory.ts#L350">memory.ts:350</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/memory.ts#L326">memory.ts:326</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/memory.ts#L363">memory.ts:363</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/memory.ts#L346">memory.ts:346</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/memory.ts#L334">memory.ts:334</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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 5e9b054f6..b7d761e32 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/21d54f988/web/src/runtime.ts#L262">runtime.ts:262</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L260">runtime.ts:260</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L258">runtime.ts:258</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L262">runtime.ts:262</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L279">runtime.ts:279</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L270">runtime.ts:270</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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 19217c51c..1b054da1b 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/21d54f988/web/src/runtime.ts#L202">runtime.ts:202</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L200">runtime.ts:200</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L198">runtime.ts:198</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L223">runtime.ts:223</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L230">runtime.ts:230</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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 aeb67469b..7f88ad41e 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/21d54f988/web/src/environment.ts#L86">environment.ts:86</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/environment.ts#L70">environment.ts:70</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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"> =&gt; </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/21d54f988/web/src/environment.ts#L69">environment.ts:69</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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">&lt;</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">&gt;</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/21d54f988/web/src/environment.ts#L78">environment.ts:78</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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">&lt;</span><span class="tsd-signature-type">number</span><span class="tsd-signature-symbol">&gt;</span><span class="tsd-signature-symbol"> = []</span></div>
 					<aside class="tsd-sources">
 						<ul>
-							<li>Defined in <a href="https://github.com/apache/tvm/blob/21d54f988/web/src/environment.ts#L84">environment.ts:84</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/environment.ts#L105">environment.ts:105</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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 d12c0b535..3e1fbbf4a 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/21d54f988/web/src/runtime.ts#L49">runtime.ts:49</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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">&lt;</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">&gt;</span></div>
 					<aside class="tsd-sources">
 						<ul>
-							<li>Defined in <a href="https://github.com/apache/tvm/blob/21d54f988/web/src/runtime.ts#L46">runtime.ts:46</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L45">runtime.ts:45</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L44">runtime.ts:44</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L47">runtime.ts:47</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L76">runtime.ts:76</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L66">runtime.ts:66</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L84">runtime.ts:84</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L95">runtime.ts:95</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L72">runtime.ts:72</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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 ee1001423..537ab9804 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/21d54f988/web/src/runtime.ts#L583">runtime.ts:583</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L579">runtime.ts:579</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L654">runtime.ts:654</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L597">runtime.ts:597</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L631">runtime.ts:631</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L644">runtime.ts:644</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L621">runtime.ts:621</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L609">runtime.ts:609</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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 a4905b973..0592f6726 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/21d54f988/web/src/runtime.ts#L692">runtime.ts:692</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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">&lt;</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">&gt;</span></div>
 					<aside class="tsd-sources">
 						<ul>
-							<li>Defined in <a href="https://github.com/apache/tvm/blob/21d54f988/web/src/runtime.ts#L684">runtime.ts:684</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L683">runtime.ts:683</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L932">runtime.ts:932</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L994">runtime.ts:994</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L924">runtime.ts:924</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L732">runtime.ts:732</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L952">runtime.ts:952</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L816">runtime.ts:816</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L1033">runtime.ts:1033</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L846">runtime.ts:846</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L750">runtime.ts:750</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L1013">runtime.ts:1013</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L789">runtime.ts:789</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L914">runtime.ts:914</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L1140">runtime.ts:1140</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/web/src/runtime.ts#L1140">runtime.ts:1140</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/21d54f988/web/src/runtime.ts#L740">runtime.ts:740</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L868">runtime.ts:868</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L857">runtime.ts:857</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L940">runtime.ts:940</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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 68aed0999..398f32ef9 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/21d54f988/web/src/memory.ts#L40">memory.ts:40</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/memory.ts#L32">memory.ts:32</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/memory.ts#L33">memory.ts:33</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/memory.ts#L154">memory.ts:154</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/memory.ts#L90">memory.ts:90</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/memory.ts#L97">memory.ts:97</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/memory.ts#L74">memory.ts:74</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/memory.ts#L81">memory.ts:81</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/memory.ts#L104">memory.ts:104</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/memory.ts#L132">memory.ts:132</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/memory.ts#L145">memory.ts:145</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/memory.ts#L60">memory.ts:60</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/memory.ts#L67">memory.ts:67</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/memory.ts#L53">memory.ts:53</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/memory.ts#L114">memory.ts:114</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/memory.ts#L124">memory.ts:124</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/memory.ts#L175">memory.ts:175</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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 a857822e6..3ecfd5191 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/21d54f988/web/src/runtime.ts#L504">runtime.ts:504</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L502">runtime.ts:502</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L516">runtime.ts:516</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L530">runtime.ts:530</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L561">runtime.ts:561</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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 4a02a617f..7dd2c3f94 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/21d54f988/web/src/runtime.ts#L304">runtime.ts:304</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L297">runtime.ts:297</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L293">runtime.ts:293</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L289">runtime.ts:289</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L291">runtime.ts:291</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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">&lt;</span><span class="tsd-signature-type">number</span><span class="tsd-signature-symbol">&gt;</span></div>
 					<aside class="tsd-sources">
 						<ul>
-							<li>Defined in <a href="https://github.com/apache/tvm/blob/21d54f988/web/src/runtime.ts#L295">runtime.ts:295</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L370">runtime.ts:370</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L414">runtime.ts:414</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L355">runtime.ts:355</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L474">runtime.ts:474</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L443">runtime.ts:443</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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 30d4b3231..20c0428d9 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/21d54f988/web/src/runtime.ts#L158">runtime.ts:158</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L157">runtime.ts:157</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L165">runtime.ts:165</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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 7063fb528..99259e012 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/21d54f988/web/src/rpc_server.ts#L92">rpc_server.ts:92</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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"> =&gt; </span><span class="tsd-signature-type">Record</span><span class="tsd-signature-symbol">&lt;</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/21d54f988/web/src/rpc_server.ts#L82">rpc_server.ts:82</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/rpc_server.ts#L78">rpc_server.ts:78</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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"> =&gt; </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/21d54f988/web/src/rpc_server.ts#L81">rpc_server.ts:81</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/rpc_server.ts#L79">rpc_server.ts:79</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/rpc_server.ts#L80">rpc_server.ts:80</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/rpc_server.ts#L77">rpc_server.ts:77</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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 0281f602b..8851e2159 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/21d54f988/web/src/runtime.ts#L145">runtime.ts:145</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L145">runtime.ts:145</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/runtime.ts#L143">runtime.ts:143</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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 aea1c3cc5..ed118e559 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/21d54f988/web/src/webgpu.ts#L57">webgpu.ts:57</a></li>
+									<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/webgpu.ts#L50">webgpu.ts:50</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/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/21d54f988/web/src/webgpu.ts#L51">webgpu.ts:51</a></li>
+							<li>Defined in <a href="https://github.com/apache/tvm/blob/19e5ec657/web/src/webgpu.ts#L51">webgpu.ts:51</a></li>
 						</ul>
... 1719 lines suppressed ...