add channels last support for thnn_conv2d (non-dilated) #49582

mingfeima · 2020-12-18T02:20:28Z

Stack from ghstack:

add channels last support for FractionalMaxPool2d and FractionalMaxPool3d #61338 add channels last support for FractionalMaxPool2d on CPU
opitimze ConvTransposedND with mkldnn float32 and bfloat16 on CPU #58348 opitimze ConvTransposedND with mkldnn float32 and bfloat16 on CPU
enable BFloat16 mkldnn_convolution on both contiguous and channels last memory format #55864 enable BFloat16 mkldnn_convolution on both contiguous and channels last memory format
add channels last (2d) support for mkldnn_convolution #55584 add channels last (2d) support for mkldnn_convolution
add channels last support for ConvTranspose2d #51185 add channels last support for ConvTranspose2d
add channels last support for PixelShuffle and PixelUnshuffle #50573 add channels last support for PixelShuffle and PixelUnshuffle
add channels last support for ChannelShuffle #50247 add channels last support for ChannelShuffle
add channel last support for MaxUnpool2d #49984 add channel last support for MaxUnpool2d
add channels last for GroupNorm #49821 add channels last for GroupNorm
add channels last support for thnn_conv2d (non-dilated) #49582 add channels last support for thnn_conv2d (non-dilated)

Differential Revision: D26007050

[ghstack-poisoned]

facebook-github-bot · 2020-12-18T02:20:42Z

💊 CI failures summary and remediations

As of commit 7279dd8 (more details on the Dr. CI page and at hud.pytorch.org/pr/49582):

1/2 failures introduced in this PR
1/2 tentatively recognized as flaky ❄️
- Click here to rerun these jobs

🕵️ 1 new failure recognized by patterns

The following CI failures do not appear to be due to upstream breakages:

pytorch_linux_xenial_py3_clang5_asan_test1 (1/1)

Step: "Run tests" (full log | diagnosis details | 🔁 rerun)

Jul 08 02:53:15 SUMMARY: UndefinedBehaviorSanit.../jenkins/workspace/aten/src/ATen/Utils.cpp:20:3 in

Jul 08 02:53:15     #9 0x555efbea28f2 in PyEval_EvalCode /home/builder/ktietz/cos6/ci_cos6/python_1622833237666/work/Python/ceval.c:731
Jul 08 02:53:15     #10 0x555efbf0acd5 in run_mod /home/builder/ktietz/cos6/ci_cos6/python_1622833237666/work/Python/pythonrun.c:1025
Jul 08 02:53:15     #11 0x555efbf0cd5d in PyRun_StringFlags /home/builder/ktietz/cos6/ci_cos6/python_1622833237666/work/Python/pythonrun.c:949
Jul 08 02:53:15     #12 0x555efbf0cdbb in PyRun_SimpleStringFlags /home/builder/ktietz/cos6/ci_cos6/python_1622833237666/work/Python/pythonrun.c:445
Jul 08 02:53:15     #13 0x555efbf0d926 in run_command /home/builder/ktietz/cos6/ci_cos6/python_1622833237666/work/Modules/main.c:301
Jul 08 02:53:15     #14 0x555efbf0d926 in Py_Main /home/builder/ktietz/cos6/ci_cos6/python_1622833237666/work/Modules/main.c:749
Jul 08 02:53:15     #15 0x555efbe47196 in main /home/builder/ktietz/cos6/ci_cos6/python_1622833237666/work/Programs/python.c:69
Jul 08 02:53:15     #16 0x7f42f008183f in __libc_start_main /build/glibc-S7Ft5T/glibc-2.23/csu/../csu/libc-start.c:291
Jul 08 02:53:15     #17 0x555efbed733d in _start (/opt/conda/bin/python3.6+0x1a733d)
Jul 08 02:53:15 
Jul 08 02:53:15 SUMMARY: UndefinedBehaviorSanitizer: undefined-behavior /var/lib/jenkins/workspace/aten/src/ATen/Utils.cpp:20:3 in 
Jul 08 02:53:15 + retcode=1
Jul 08 02:53:15 + set -e
Jul 08 02:53:15 + return 1
Jul 08 02:53:15 + [[ pytorch-linux-xenial-py3-clang5-asan-test1 == *-NO_AVX-* ]]
Jul 08 02:53:15 + [[ pytorch-linux-xenial-py3-clang5-asan-test1 == *-NO_AVX2-* ]]
Jul 08 02:53:15 + '[' -n https://github.com/pytorch/pytorch/pull/49582 ']'
Jul 08 02:53:15 + [[ pytorch-linux-xenial-py3-clang5-asan-test1 != *coverage* ]]
Jul 08 02:53:15 ++ mktemp
Jul 08 02:53:15 + DETERMINE_FROM=/tmp/tmp.CufbEAbXYp
Jul 08 02:53:15 + file_diff_from_base /tmp/tmp.CufbEAbXYp

❄️ 1 failure tentatively classified as flaky

but reruns have not yet been triggered to confirm:

pytorch_linux_xenial_cuda11_1_cudnn8_py3_gcc7_test2 (1/1)

Step: "Run tests" (full log | diagnosis details | 🔁 rerun) ❄️

Jul 08 05:23:59 unknown file: Failure

Jul 08 05:23:59 frame #7: build/bin/test_api() [0xc0b8c5]
Jul 08 05:23:59 frame #8: build/bin/test_api() [0xc0bb65]
Jul 08 05:23:59 frame #9: testing::internal::UnitTestImpl::RunAllTests() + 0xbf9 (0xc0cba9 in build/bin/test_api)
Jul 08 05:23:59 frame #10: testing::UnitTest::Run() + 0x8f (0xc0ce4f in build/bin/test_api)
Jul 08 05:23:59 frame #11: main + 0xc8 (0x5833a8 in build/bin/test_api)
Jul 08 05:23:59 frame #12: __libc_start_main + 0xf0 (0x7f7f36387840 in /lib/x86_64-linux-gnu/libc.so.6)
Jul 08 05:23:59 frame #13: _start + 0x29 (0x5b9a19 in build/bin/test_api)
Jul 08 05:23:59 " thrown in the test body.
Jul 08 05:23:59 [  FAILED  ] IntegrationTest.MNIST_CUDA (3 ms)
Jul 08 05:23:59 [ RUN      ] IntegrationTest.MNISTBatchNorm_CUDA
Jul 08 05:23:59 unknown file: Failure
Jul 08 05:23:59 C++ exception with description "Error opening images file at test/cpp/api/mnist/train-images-idx3-ubyte
Jul 08 05:23:59 Exception raised from read_images at /var/lib/jenkins/workspace/torch/csrc/api/src/data/datasets/mnist.cpp:67 (most recent call first):
Jul 08 05:23:59 frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >) + 0x6b (0x7f7f4ff34c6b in /var/lib/jenkins/workspace/build/lib/libc10.so)
Jul 08 05:23:59 frame #1: c10::detail::torchCheckFail(char const*, char const*, unsigned int, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&) + 0xce (0x7f7f4ff3047e in /var/lib/jenkins/workspace/build/lib/libc10.so)
Jul 08 05:23:59 frame #2: <unknown function> + 0x4215952 (0x7f7f54595952 in /var/lib/jenkins/workspace/build/lib/libtorch_cpu.so)
Jul 08 05:23:59 frame #3: torch::data::datasets::MNIST::MNIST(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, torch::data::datasets::MNIST::Mode) + 0x46 (0x7f7f545969f6 in /var/lib/jenkins/workspace/build/lib/libtorch_cpu.so)
Jul 08 05:23:59 frame #4: IntegrationTest_MNISTBatchNorm_CUDA_Test::TestBody() + 0x9d6 (0x7843b6 in build/bin/test_api)
Jul 08 05:23:59 frame #5: void testing::internal::HandleExceptionsInMethodIfSupported<testing::Test, void>(testing::Test*, void (testing::Test::*)(), char const*) + 0x4a (0xc1489a in build/bin/test_api)
Jul 08 05:23:59 frame #6: build/bin/test_api() [0xc0b2d6]
Jul 08 05:23:59 frame #7: build/bin/test_api() [0xc0b8c5]

Preview docs built from this PR

This comment was automatically generated by Dr. CI (expand for details).

Follow this link to opt-out of these comments for your Pull Requests.

Please report bugs/suggestions to the (internal) Dr. CI Users group.

Click here to manually regenerate this comment.

ghstack-source-id: 8044e97 Pull Request resolved: #49582

mingfeima · 2020-12-21T11:37:07Z

Initiative of this patch

Due to performance or functionality reasons, nn.Conv2d will choose mkldnn_convolution or thnn_conv2d on CPU depending on input tensor size and kernel size, refer to use_mkldnn definition. So in order to claim that nn.Conv2d supports channels last memory format on CPU, need to make sure the underlying implementations support CL format. mkldnn_convolution CL support will be covered in later patches, this PR aims at providing CL support for thnn_conv2d.

Performance

Measured Conv2d test configs #40610 on single core of Intel(R) Xeon(R) Gold 6248 CPU @ 2.50GHz, mkldnn path is manually disabled with torch._C._set_mkldnn_enabled(False) so as to test thnn_conv2d:

Notice that on grouped conv CL performs worse than Contiguous in the following table, but grouped conv will end up in mkldnn with default setting (mkldnn has much better perf with group conv than the gemm based algorithm)

input size	output channels	kernel	stride	padding	groups	before (nchw)	after (nchw)	after (nhwc)
[1,3,224,224]	64	7	2	3	1	4.443	4.435	4.488
[1,64,56,56]	128	1	1	0	1	0.337	0.337	0.360
[1,128,56,56]	128	3	1	1	32	1.784	1.798	9.012
[1,128,56,56]	256	1	1	0	1	1.117	1.116	1.155
[1,64,56,56]	256	1	1	0	1	0.773	0.774	0.783
[1,256,56,56]	128	1	1	0	1	0.994	1.006	1.084
[1,128,56,56]	128	3	1	1	32	1.787	1.784	9.013
[1,128,56,56]	256	1	1	0	1	1.117	1.117	1.159
[1,256,56,56]	128	1	1	0	1	0.994	0.995	1.085
[1,128,56,56]	128	3	1	1	32	1.783	1.786	9.004
[1,128,56,56]	256	1	1	0	1	1.123	1.115	1.157
[1,256,56,56]	256	1	1	0	1	2.057	2.059	2.136
[1,256,56,56]	256	3	2	1	32	4.083	4.101	4.186
[1,256,28,28]	512	1	1	0	1	1.059	1.057	1.073
[1,256,56,56]	512	1	2	0	1	1.395	1.400	1.149
[1,512,28,28]	256	1	1	0	1	1.048	1.050	1.109
[1,256,28,28]	256	3	1	1	32	1.379	1.383	2.936
[1,256,28,28]	512	1	1	0	1	1.057	1.056	1.072
[1,512,28,28]	256	1	1	0	1	1.048	1.051	1.106
[1,256,28,28]	256	3	1	1	32	1.376	1.381	2.942
[1,256,28,28]	512	1	1	0	1	1.059	1.065	1.073
[1,512,28,28]	256	1	1	0	1	1.049	1.056	1.110
[1,256,28,28]	256	3	1	1	32	1.377	1.387	2.933
[1,256,28,28]	512	1	1	0	1	1.058	1.056	1.072
[1,512,28,28]	512	1	1	0	1	2.014	2.008	2.059
[1,512,28,28]	512	3	2	1	32	2.263	2.284	1.583
[1,512,14,14]	1024	1	1	0	1	1.200	1.199	1.136
[1,512,28,28]	1024	1	2	0	1	1.356	1.359	1.175
[1,1024,14,14]	512	1	1	0	1	1.239	1.235	1.164
[1,512,14,14]	512	3	1	1	32	1.380	1.395	1.301
[1,512,14,14]	1024	1	1	0	1	1.192	1.190	1.140
[1,1024,14,14]	512	1	1	0	1	1.234	1.235	1.163
[1,512,14,14]	512	3	1	1	32	1.375	1.394	1.303
[1,512,14,14]	1024	1	1	0	1	1.192	1.195	1.143
[1,1024,14,14]	512	1	1	0	1	1.240	1.235	1.165
[1,512,14,14]	512	3	1	1	32	1.384	1.394	1.309
[1,512,14,14]	1024	1	1	0	1	1.195	1.189	1.134
[1,1024,14,14]	512	1	1	0	1	1.234	1.231	1.164
[1,512,14,14]	512	3	1	1	32	1.380	1.399	1.301
[1,512,14,14]	1024	1	1	0	1	1.196	1.188	1.141
[1,1024,14,14]	512	1	1	0	1	1.234	1.237	1.168
[1,512,14,14]	512	3	1	1	32	1.379	1.388	1.303
[1,512,14,14]	1024	1	1	0	1	1.192	1.192	1.140
[1,1024,14,14]	512	1	1	0	1	1.234	1.234	1.166
[1,512,14,14]	512	3	1	1	32	1.381	1.389	1.302
[1,512,14,14]	1024	1	1	0	1	1.189	1.194	1.138
[1,1024,14,14]	512	1	1	0	1	1.235	1.235	1.161
[1,512,14,14]	512	3	1	1	32	1.379	1.397	1.312
[1,512,14,14]	1024	1	1	0	1	1.198	1.191	1.143
[1,1024,14,14]	512	1	1	0	1	1.236	1.233	1.163
[1,512,14,14]	512	3	1	1	32	1.379	1.391	1.306
[1,512,14,14]	1024	1	1	0	1	1.191	1.190	1.137
[1,1024,14,14]	512	1	1	0	1	1.236	1.232	1.168
[1,512,14,14]	512	3	1	1	32	1.375	1.389	1.303
[1,512,14,14]	1024	1	1	0	1	1.194	1.195	1.136
[1,1024,14,14]	512	1	1	0	1	1.243	1.241	1.161
[1,512,14,14]	512	3	1	1	32	1.379	1.398	1.305
[1,512,14,14]	1024	1	1	0	1	1.195	1.189	1.141
[1,1024,14,14]	512	1	1	0	1	1.237	1.232	1.163
[1,512,14,14]	512	3	1	1	32	1.379	1.391	1.304
[1,512,14,14]	1024	1	1	0	1	1.193	1.188	1.136
[1,1024,14,14]	512	1	1	0	1	1.237	1.231	1.166
[1,512,14,14]	512	3	1	1	32	1.380	1.397	1.305
[1,512,14,14]	1024	1	1	0	1	1.191	1.193	1.136
[1,1024,14,14]	512	1	1	0	1	1.233	1.234	1.163
[1,512,14,14]	512	3	1	1	32	1.377	1.399	1.302
[1,512,14,14]	1024	1	1	0	1	1.194	1.191	1.140
[1,1024,14,14]	512	1	1	0	1	1.236	1.237	1.162
[1,512,14,14]	512	3	1	1	32	1.376	1.391	1.305
[1,512,14,14]	1024	1	1	0	1	1.194	1.194	1.144
[1,1024,14,14]	512	1	1	0	1	1.233	1.232	1.169
[1,512,14,14]	512	3	1	1	32	1.381	1.397	1.300
[1,512,14,14]	1024	1	1	0	1	1.191	1.189	1.140
[1,1024,14,14]	512	1	1	0	1	1.236	1.235	1.164
[1,512,14,14]	512	3	1	1	32	1.379	1.397	1.306
[1,512,14,14]	1024	1	1	0	1	1.199	1.187	1.136
[1,1024,14,14]	512	1	1	0	1	1.239	1.241	1.162
[1,512,14,14]	512	3	1	1	32	1.377	1.404	1.305
[1,512,14,14]	1024	1	1	0	1	1.192	1.189	1.136
[1,1024,14,14]	512	1	1	0	1	1.235	1.233	1.166
[1,512,14,14]	512	3	1	1	32	1.379	1.407	1.310
[1,512,14,14]	1024	1	1	0	1	1.202	1.193	1.151
[1,1024,14,14]	512	1	1	0	1	1.239	1.234	1.163
[1,512,14,14]	512	3	1	1	32	1.376	1.393	1.305
[1,512,14,14]	1024	1	1	0	1	1.194	1.199	1.134
[1,1024,14,14]	512	1	1	0	1	1.234	1.236	1.162
[1,512,14,14]	512	3	1	1	32	1.379	1.395	1.305
[1,512,14,14]	1024	1	1	0	1	1.191	1.194	1.138
[1,1024,14,14]	512	1	1	0	1	1.235	1.234	1.164
[1,512,14,14]	512	3	1	1	32	1.377	1.398	1.301
[1,512,14,14]	1024	1	1	0	1	1.192	1.191	1.137
[1,1024,14,14]	512	1	1	0	1	1.232	1.233	1.168
[1,512,14,14]	512	3	1	1	32	1.380	1.394	1.310
[1,512,14,14]	1024	1	1	0	1	1.191	1.189	1.135
[1,1024,14,14]	1024	1	1	0	1	2.336	2.333	2.280
[1,1024,14,14]	1024	3	2	1	32	1.711	1.735	1.288
[1,1024,7,7]	2048	1	1	0	1	1.620	1.615	1.746
[1,1024,14,14]	2048	1	2	0	1	1.720	1.728	1.763
[1,2048,7,7]	1024	1	1	0	1	1.630	1.629	1.763
[1,1024,7,7]	1024	3	1	1	32	1.492	1.503	1.166
[1,1024,7,7]	2048	1	1	0	1	1.620	1.614	1.749
[1,2048,7,7]	1024	1	1	0	1	1.624	1.640	1.760
[1,1024,7,7]	1024	3	1	1	32	1.489	1.510	1.172
[1,1024,7,7]	2048	1	1	0	1	1.630	1.616	1.750

[ghstack-poisoned]

ghstack-source-id: 420d9cd Pull Request resolved: #49582

[ghstack-poisoned]

Differential Revision: [D26007050](https://our.internmc.facebook.com/intern/diff/D26007050) [ghstack-poisoned]

ghstack-source-id: 0584ef4 Pull Request resolved: pytorch#49582

Differential Revision: [D26007050](https://our.internmc.facebook.com/intern/diff/D26007050) [ghstack-poisoned]

VitalyFedyunin · 2021-06-23T16:48:43Z

@VitalyFedyunin has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.

Differential Revision: [D26007050](https://our.internmc.facebook.com/intern/diff/D26007050) [ghstack-poisoned]

mingfeima · 2021-06-29T01:38:29Z

rebased!

VitalyFedyunin · 2021-06-29T18:08:51Z

@VitalyFedyunin has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.

Differential Revision: [D26007050](https://our.internmc.facebook.com/intern/diff/D26007050) [ghstack-poisoned]

mingfeima · 2021-07-09T01:04:52Z

rebased!

VitalyFedyunin · 2021-07-12T17:42:22Z

@VitalyFedyunin has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.

facebook-github-bot · 2021-07-20T19:52:35Z

@VitalyFedyunin merged this pull request in 8b88c24.

facebook-github-bot · 2021-07-22T15:31:31Z

This pull request has been reverted by b60d1b7.

VitalyFedyunin · 2021-07-22T15:40:29Z

Reverting as it might be offender to Iphone perf regression, please do nothing at thing point, might be false positive, I'm investigating.

mingfeima · 2021-11-03T01:48:03Z

@VitalyFedyunin The current code base in ConvolutionMM2d.cpp has replaced .mm() with cpublas::gemm, also use TensorAccessor to save .narrow(), which shall be able to same some time in dispatch and TensorImpl construction/deconstruction.

I will rewrite this one accordingly, and collect new performance data. @VitalyFedyunin

add channels last support for thnn_conv2d (non-dilated)

892ed9c

[ghstack-poisoned]

This was referenced Dec 18, 2020

add channels last for MaxPool2d #48917

Closed

add channels last support for AvgPool2d on CPU #48918

Closed

optimize channels last for BatchNorm2d on CPU #48919

Closed

facebook-github-bot added the cla signed label Dec 18, 2020

mingfeima mentioned this pull request Dec 18, 2020

add channels last for AdapativeMaxPool2d #48920

Closed

mingfeima added a commit that referenced this pull request Dec 18, 2020

add channels last support for thnn_conv2d (non-dilated)

ab529f4

ghstack-source-id: 8044e97 Pull Request resolved: #49582

pytorchbot added the open source label Dec 18, 2020

Update on "add channels last support for thnn_conv2d (non-dilated)"

ce08011

[ghstack-poisoned]

mingfeima added a commit that referenced this pull request Dec 21, 2020

add channels last support for thnn_conv2d (non-dilated)

4c6a701

ghstack-source-id: 420d9cd Pull Request resolved: #49582

This was referenced Dec 24, 2020

add channels last for GroupNorm #49821

Closed

add channel last support for MaxUnpool2d #49984

Closed

mingfeima added 3 commits January 1, 2021 08:55

Update on "add channels last support for thnn_conv2d (non-dilated)"

114a93c

[ghstack-poisoned]

Update on "add channels last support for thnn_conv2d (non-dilated)"

8a9a7c5

[ghstack-poisoned]

Update on "add channels last support for thnn_conv2d (non-dilated)"

7e2b955

[ghstack-poisoned]

mingfeima mentioned this pull request Jan 8, 2021

add channels last support for ChannelShuffle #50247

Closed

mingfeima mentioned this pull request Jan 15, 2021

add channels last support for PixelShuffle and PixelUnshuffle #50573

Closed

mingfeima added 5 commits January 18, 2021 10:01

Update on "add channels last support for thnn_conv2d (non-dilated)"

6aab268

[ghstack-poisoned]

Update on "add channels last support for thnn_conv2d (non-dilated)"

eae6036

[ghstack-poisoned]

Update on "add channels last support for thnn_conv2d (non-dilated)"

27cad2b

[ghstack-poisoned]

Update on "add channels last support for thnn_conv2d (non-dilated)"

ed4fac0

Differential Revision: [D26007050](https://our.internmc.facebook.com/intern/diff/D26007050) [ghstack-poisoned]

Update on "add channels last support for thnn_conv2d (non-dilated)"

f235c59

Differential Revision: [D26007050](https://our.internmc.facebook.com/intern/diff/D26007050) [ghstack-poisoned]

mingfeima mentioned this pull request Jan 27, 2021

add channels last support for ConvTranspose2d #51185

Closed

mingfeima added 5 commits February 5, 2021 11:02

Update on "add channels last support for thnn_conv2d (non-dilated)"

c1bef15

Differential Revision: [D26007050](https://our.internmc.facebook.com/intern/diff/D26007050) [ghstack-poisoned]

Update on "add channels last support for thnn_conv2d (non-dilated)"

a80f583

Differential Revision: [D26007050](https://our.internmc.facebook.com/intern/diff/D26007050) [ghstack-poisoned]

Update on "add channels last support for thnn_conv2d (non-dilated)"

2966b0b

Differential Revision: [D26007050](https://our.internmc.facebook.com/intern/diff/D26007050) [ghstack-poisoned]

Update on "add channels last support for thnn_conv2d (non-dilated)"

ab1aed8

Differential Revision: [D26007050](https://our.internmc.facebook.com/intern/diff/D26007050) [ghstack-poisoned]

Update on "add channels last support for thnn_conv2d (non-dilated)"

91b4dae

Differential Revision: [D26007050](https://our.internmc.facebook.com/intern/diff/D26007050) [ghstack-poisoned]

mingfeima mentioned this pull request May 15, 2021

opitimze ConvTransposedND with mkldnn float32 and bfloat16 on CPU #58348

Closed

dgl-intel pushed a commit to dgl-intel/pytorch that referenced this pull request May 15, 2021

add channels last support for thnn_conv2d (non-dilated)

da844e9

ghstack-source-id: 0584ef4 Pull Request resolved: pytorch#49582

mingfeima added 6 commits June 3, 2021 12:54

Update on "add channels last support for thnn_conv2d (non-dilated)"

73257fd

Differential Revision: [D26007050](https://our.internmc.facebook.com/intern/diff/D26007050) [ghstack-poisoned]

Update on "add channels last support for thnn_conv2d (non-dilated)"

cb456c6

Differential Revision: [D26007050](https://our.internmc.facebook.com/intern/diff/D26007050) [ghstack-poisoned]

Update on "add channels last support for thnn_conv2d (non-dilated)"

60e0d55

Differential Revision: [D26007050](https://our.internmc.facebook.com/intern/diff/D26007050) [ghstack-poisoned]

Update on "add channels last support for thnn_conv2d (non-dilated)"

77f7ec2

Differential Revision: [D26007050](https://our.internmc.facebook.com/intern/diff/D26007050) [ghstack-poisoned]

Update on "add channels last support for thnn_conv2d (non-dilated)"

1d739b0

Differential Revision: [D26007050](https://our.internmc.facebook.com/intern/diff/D26007050) [ghstack-poisoned]

Update on "add channels last support for thnn_conv2d (non-dilated)"

3dd38e4

Differential Revision: [D26007050](https://our.internmc.facebook.com/intern/diff/D26007050) [ghstack-poisoned]

Update on "add channels last support for thnn_conv2d (non-dilated)"

2762055

Differential Revision: [D26007050](https://our.internmc.facebook.com/intern/diff/D26007050) [ghstack-poisoned]

VitalyFedyunin approved these changes Jun 29, 2021

View reviewed changes

Update on "add channels last support for thnn_conv2d (non-dilated)"

9cb085d

Differential Revision: [D26007050](https://our.internmc.facebook.com/intern/diff/D26007050) [ghstack-poisoned]

This was referenced Jul 7, 2021

add channels last support for FractionalMaxPool2d #61337

Closed

add channels last support for FractionalMaxPool2d and FractionalMaxPool3d #61338

Closed

Update on "add channels last support for thnn_conv2d (non-dilated)"

7279dd8

Differential Revision: [D26007050](https://our.internmc.facebook.com/intern/diff/D26007050) [ghstack-poisoned]

facebook-github-bot closed this in 8b88c24 Jul 20, 2021

facebook-github-bot added the Merged label Jul 20, 2021

facebook-github-bot added the Reverted label Jul 22, 2021

facebook-github-bot deleted the gh/mingfeima/7/head branch July 24, 2021 14:17

mingfeima mentioned this pull request Nov 10, 2021

add channels last support for thnn_conv2d (non-dilated) #68101

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

add channels last support for thnn_conv2d (non-dilated) #49582

add channels last support for thnn_conv2d (non-dilated) #49582

Uh oh!

mingfeima commented Dec 18, 2020 •

edited

Loading

Uh oh!

facebook-github-bot commented Dec 18, 2020 •

edited

Loading

Uh oh!

mingfeima commented Dec 21, 2020

Uh oh!

VitalyFedyunin commented Jun 23, 2021

Uh oh!

mingfeima commented Jun 29, 2021

Uh oh!

VitalyFedyunin commented Jun 29, 2021

Uh oh!

mingfeima commented Jul 9, 2021

Uh oh!

VitalyFedyunin commented Jul 12, 2021

Uh oh!

facebook-github-bot commented Jul 20, 2021

Uh oh!

facebook-github-bot commented Jul 22, 2021

Uh oh!

VitalyFedyunin commented Jul 22, 2021

Uh oh!

mingfeima commented Nov 3, 2021 •

edited

Loading

Uh oh!

Uh oh!

add channels last support for thnn_conv2d (non-dilated) #49582

add channels last support for thnn_conv2d (non-dilated) #49582

Uh oh!

Conversation

mingfeima commented Dec 18, 2020 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

facebook-github-bot commented Dec 18, 2020 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

💊 CI failures summary and remediations

🕵️ 1 new failure recognized by patterns

pytorch_linux_xenial_py3_clang5_asan_test1 (1/1)

❄️ 1 failure tentatively classified as flaky

pytorch_linux_xenial_cuda11_1_cudnn8_py3_gcc7_test2 (1/1)

Uh oh!

mingfeima commented Dec 21, 2020

Initiative of this patch

Performance

Uh oh!

VitalyFedyunin commented Jun 23, 2021

Uh oh!

mingfeima commented Jun 29, 2021

Uh oh!

VitalyFedyunin commented Jun 29, 2021

Uh oh!

mingfeima commented Jul 9, 2021

Uh oh!

VitalyFedyunin commented Jul 12, 2021

Uh oh!

facebook-github-bot commented Jul 20, 2021

Uh oh!

facebook-github-bot commented Jul 22, 2021

Uh oh!

VitalyFedyunin commented Jul 22, 2021

Uh oh!

mingfeima commented Nov 3, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

mingfeima commented Dec 18, 2020 •

edited

Loading

facebook-github-bot commented Dec 18, 2020 •

edited

Loading

mingfeima commented Nov 3, 2021 •

edited

Loading