Summary:
Usage explanation will be in the release note runbook.
This allows to generate diffs like:
```
Processing torch.nn
Things that were added:
{'quantizable', 'ChannelShuffle', 'LazyConvTranspose2d', 'LazyConv2d', 'LazyConvTranspose3d', 'LazyConv1d', 'GaussianNLLLoss', 'LazyConv3d', 'PixelUnshuffle', 'UninitializedParameter', 'LazyLinear', 'LazyConvTranspose1d'}
Things that were removed:
set()
```
This can then be shared with module owners along with the commits to help them validate that the namespace changes for their submodule is as expected.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/51685
Reviewed By: zhangguanheng66
Differential Revision: D26260258
Pulled By: albanD
fbshipit-source-id: 40e40f86314e17246899d01ffa4b2631e93b52f7
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/51521
* Add loop & if node to the list of nodes that could produce sequence type output.
* Switch from `[]` to `at()` to avoid segfault of out of range access.
Test Plan: Imported from OSS
Reviewed By: pbelevich
Differential Revision: D26203112
Pulled By: SplitInfinity
fbshipit-source-id: e990eeed933124b195be0be159271e33fb485063
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/51517
Fix get/set attributes when getting/setting a model parameter.
This PR also fixes inplace ops in If blocks.
Test Plan: Imported from OSS
Reviewed By: pbelevich
Differential Revision: D26203116
Pulled By: SplitInfinity
fbshipit-source-id: bed6ee6dd92b5b43febc8c584a6872290f8fe33f
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/50911
Need to replace dtype of export created scalars from float to double. (In torch implicit conversion logic, python numbers are double)
Test case skipped in CI due to that current CI job env does not have CUDA support.
Test Plan: Imported from OSS
Reviewed By: pbelevich
Differential Revision: D26050889
Pulled By: SplitInfinity
fbshipit-source-id: 1fdde23a68d4793e6b9a82840acc213e5c3aa760
Summary:
Handle sequence output shape and type inference.
This PR fixes value type of sequence outputs. Prior to this, all model sequence type outputs were unfolded for ONNX models.
This PR also enable shape inference for sequence outputs to represent the dynamic shape of these values.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/46542
Reviewed By: ezyang
Differential Revision: D24924236
Pulled By: bzinodev
fbshipit-source-id: 506e70a38cfe31069191d7f40fc6375239c6aafe
Summary:
Update pass that handles prim::ListUnpack in peephole file, so that it also covers the case when input to the node is of ListType.
Fixes https://github.com/pytorch/pytorch/issues/45816
Pull Request resolved: https://github.com/pytorch/pytorch/pull/46264
Reviewed By: mrshenli
Differential Revision: D24566070
Pulled By: bzinodev
fbshipit-source-id: 32555487054f6a7fe02cc17c66bcbe81ddf9623e
Summary:
First commit contains the initial code from Richard's branch.
Second commit are the changes that I made during the writing process
Third commit is the update to support category/topic pair for each commit
Pull Request resolved: https://github.com/pytorch/pytorch/pull/47360
Reviewed By: ejguan
Differential Revision: D24741003
Pulled By: albanD
fbshipit-source-id: d0fcc6765968dc1732d8a515688d11372c7e653d
Summary:
Currently, no test reports are uploaded to CI because the paths for the `onnx` runs are incorrect. This PR attempts to change that.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/47315
Reviewed By: malfet
Differential Revision: D24727607
Pulled By: janeyx99
fbshipit-source-id: f6d91698fdb15a39e01ef812032d4cd30621f864
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/46383
The old `USE_METAL` is actually being used by Caffe2. Here we introduce a new macro to enable metal in pytorch.
ghstack-source-id: 114499392
Test Plan:
- Circle CI
- The Person Segmentation model works
Reviewed By: linbinyu
Differential Revision: D24322018
fbshipit-source-id: 4e5548afba426b49f314366d89b18ba0c7e745ca
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/46112
### Summary
This PR adds the support of running torchscript models on iOS GPU via Metal (Inference only). The feature is currently in prototype state, API changes are expected. The tutorial and the documents will be added once it goes to beta.
allow-large-files
- Users API
```
auto module = torch::jit::load(model);
module.eval();
at::Tensor input = at::ones({1,3,224,224}, at::ScalarType::Float).metal();
auto output = module.forward({input}).toTensor().cpu();
```
- Supported Models
- Person Segmentation v106 (FB Internal)
- Mobilenetv2
- Supported Operators
- aten::conv2d
- aten::addmm
- aten::add.Tensor
- aten::sub.Tensor
- aten::mul.Tensor
- aten::relu
- aten::hardtanh
- aten::hardtanh_
- aten::sigmoid
- aten::max_pool2d
- aten::adaptive_avg_pool2d
- aten::reshape
- aten::t
- aten::view
- aten::log_softmax.int
- aten::upsample_nearest2d.vec
- Supported Devices
- Apple A9 and above
- iOS 10.2 and above
- CMake scripts
- `IOS_ARCH=arm64 ./scripts/build_ios.sh -DUSE_METAL=ON`
### Test Plan
- Circle CI
ghstack-source-id: 114155638
Test Plan:
1. Sandcastle CI
2. Circle CI
Reviewed By: dreiss
Differential Revision: D23236555
fbshipit-source-id: 98ffc48b837e308bc678c37a9a5fd8ae72d11625
Summary:
* Support propagating `dim_param` in ONNX by encoding as `ShapeSymbol` in `SymbolicShape` of outputs. If export is called with `dynamic_axes` provided, shape inference will start with these axes set as dynamic.
* Add new test file `test_pytorch_onnx_shape_inference.py`, reusing all test cases from `test_pytorch_onnx_onnxruntime.py`, but focus on validating shape for all nodes in graph. Currently this is not enabled in the CI, since there are still quite some existing issues and corner cases to fix. The test is default to run only at opset 12.
* Bug fixes, such as div, _len, and peephole.cpp passes for PackPadded, and LogSoftmaxCrossEntropy.
* This PR depends on existing PR such as 44332.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/44920
Reviewed By: eellison
Differential Revision: D23958398
Pulled By: bzinodev
fbshipit-source-id: 00479d9bd19c867d526769a15ba97ec16d56e51d
Summary:
There is a module called `2to3` which you can target for future specifically to remove these, the directory of `caffe2` has the most redundant imports:
```2to3 -f future -w caffe2```
Pull Request resolved: https://github.com/pytorch/pytorch/pull/45033
Reviewed By: seemethere
Differential Revision: D23808648
Pulled By: bugra
fbshipit-source-id: 38971900f0fe43ab44a9168e57f2307580d36a38
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/44651
Adds pruning for our anaconda channels (pytorch-nightly, pytorch-test)
into our CI pipeline so that it gets run on a more consistent basis.
Signed-off-by: Eli Uriegas <eliuriegas@fb.com>
Test Plan: Imported from OSS
Reviewed By: walterddr
Differential Revision: D23692851
Pulled By: seemethere
fbshipit-source-id: fa69b506b73805bf2ffbde75d221aef1ee3f753e
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/43564
Static dispatch was originally introduced for mobile selective build.
Since we have added selective build support for dynamic dispatch and
tested it in FB production for months, we can deprecate static dispatch
to reduce the complexity of the codebase.
Test Plan: Imported from OSS
Reviewed By: ezyang
Differential Revision: D23324452
Pulled By: ljk53
fbshipit-source-id: d2970257616a8c6337f90249076fca1ae93090c7
Summary:
build_android.sh should check PYTHON environment variable before trying to use default python executable.
Even in that case, try to pick python3 over python2 when available.
Closes https://github.com/pytorch/pytorch/issues/41795
Pull Request resolved: https://github.com/pytorch/pytorch/pull/41927
Reviewed By: seemethere
Differential Revision: D22696850
Pulled By: malfet
fbshipit-source-id: be236c2baf54a1cd111e55ee7743cdc93cb6b9d7
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/39341
This PR introduces neon backend for vec256 class for float datatype.
For now only aarch64 is enabled due to few issues with enabling in
aarch32 bit.
Test Plan:
vec256_test
Imported from OSS
Differential Revision: D21822399
fbshipit-source-id: 3851c4336d93d1c359c85b38cf19904f82bc7b8d
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/40059
This benchmark is added specifically for mobile to see if compiler is
autovectorizing and thus we have no advantage of neon backend for vec256
for add op.
Test Plan:
CI
Imported from OSS
Differential Revision: D22055146
fbshipit-source-id: 43ba6c4ae57c6f05d84887c2750ce21ae1b0f0b5
Summary:
This re-applies D21232894 (b9d3869df3) and D22162524, plus updates jni_deps in a few places
to avoid breaking host JNI tests.
Test Plan: `buck test @//fbandroid/mode/server //fbandroid/instrumentation_tests/com/facebook/caffe2:host-test`
Reviewed By: xcheng16
Differential Revision: D22199952
fbshipit-source-id: df13eef39c01738637ae8cf7f581d6ccc88d37d5
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/37243
*** Why ***
As it stands, we have two thread pool solutions concurrently in use in PyTorch mobile: (1) the open source pthreadpool library under third_party, and (2) Caffe2's implementation of pthreadpool under caffe2/utils/threadpool. Since the primary use-case of the latter has been to act as a drop-in replacement for the third party version so as to enable integration and usage from within NNPACK and QNNPACK, Caffe2's implementation is intentionally written to the exact same interface as the third party version.
The original argument in favor of C2's implementation has been improved performance as a result of using spin locks, as opposed to relinquishing the thread's time slot and putting it to sleep - a less expensive operation up to a point. That seems to have given C2's implementation the upper hand in performance, hence justifying the added maintenance complexity, until the third party version improved in parallel surpassing the efficiency of C2's implementation as I have verified in benchmarks. With that advantage gone, there is no reason to continue using C2's implementation in PyTorch mobile either from the perspective of performance or code hygiene. As a matter of fact, there is considerable performance benefit to be had as a result of using the third party version as it currently stands.
This is a tricky change though, mainly because in order to avoid potential performance regressions, of which I have witnessed none but just in abundance of caution, we have decided to continue using the internal C2's implementation whenever building for Caffe2. Again, this is mainly to avoid potential performance regressions in production C2 use cases even if doing so results in reduced performance as far as I can tell.
So to summarize, today, and as it currently stands, we are using C2's implementation for (1) NNPACK, (2) PyTorch QNNPACK, and (3) ATen parallel_for on mobile builds, while using the third party version of pthreadpool for XNNPACK as XNNPACK does not provide any build options to link against an external implementation unlike NNPACK and QNNPACK do.
The goal of this PR then, is to unify all usage on mobile to the third party implementation both for improved performance and better code hygiene. This applies to PyTorch's use of NNPACK, QNNPACK, XNNPACK, and mobile's implementation of ATen parallel_for, all getting routed to the
exact same third party implementation in this PR.
Considering that NNPACK, QNNPACK, and XNNPACK are not mobile specific, these benefits carry over to non-mobile builds of PyTorch (but not Caffe2) as well. The implementation of ATen parallel_for on non-mobile builds remains unchanged.
*** How ***
This is where things get tricky.
A good deal of the build system complexity in this PR arises from our desire to maintain C2's implementation intact for C2's use.
pthreadpool is a C library with no concept of namespaces, which means two copies of the library cannot exist in the same binary or symbol collision will occur violating ODR. This means that somehow, and based on some condition, we must decide on the choice of a pthreadpool implementation. In practice, this has become more complicated as a result of all the possible combinations that USE_NNPACK, USE_QNNPACK, USE_PYTORCH_QNNPACK, USE_XNNPACK, USE_SYSTEM_XNNPACK, USE_SYSTEM_PTHREADPOOL and other variables can result in. Having said that, I have done my best in this PR to surgically cut through this complexity in a way that minimizes the side effects, considering the significance of the performance we are leaving on the table, yet, as a result of this combinatorial explosion explained above I cannot guarantee that every single combination will work as expected on the first try. I am heavily relying on CI to find any issues as local testing can only go that far.
Having said that, this PR provides a simple non mobile-specific C++ thread pool implementation on top of pthreadpool, namely caffe2::PThreadPool that automatically routes to C2's implementation or the third party version depending on the build configuration. This simplifies the logic at the cost of pushing the complexity to the build scripts. From there on, this thread pool is used in aten parallel_for, and NNPACK and family, again, routing all usage of threading to C2 or third party pthreadpool depending on the build configuration.
When it is all said or done, the layering will look like this:
a) aten::parallel_for, uses
b) caffe2::PThreadPool, which uses
c) pthreadpool C API, which delegates to
c-1) third_party implementation of pthreadpool if that's what the build has requested, and the rabbit hole ends here.
c-2) C2's implementation of pthreadpool if that's what the build has requested, which itself delegates to
c-2-1) caffe2::ThreadPool, and the rabbit hole ends here.
NNPACK, and (PyTorch) QNNPACK directly hook into (c). They never go through (b).
Differential Revision: D21232894
Test Plan: Imported from OSS
Reviewed By: dreiss
Pulled By: AshkanAliabadi
fbshipit-source-id: 8b3de86247fbc3a327e811983e082f9d40081354
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/39587
Example of using direct linking to pytorch_jni library from aar and updating android/README.md with the tutorial how to do it.
Adding `nativeBuild` dimension to `test_app`, using direct aar dependencies, as headers packaging is not landed yet, excluding `nativeBuild` from building by default for CI.
Additional change to `scripts/build_pytorch_android.sh`:
Skipping clean task here as android gradle plugin 3.3.2 exteralNativeBuild has problems with it when abiFilters are specified.
Will be returned back in the following diffs with upgrading of gradle and android gradle plugin versions.
Test Plan: Imported from OSS
Differential Revision: D22118945
Pulled By: IvanKobzarev
fbshipit-source-id: 31c54b49b1f262cbe5f540461d3406f74851db6c
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/39588
Before this diff we used c++_static linking.
Users will dynamically link to libpytorch_jni.so and have at least one more their own shared library that probably uses stl library.
We must have not more than one stl per app. ( https://developer.android.com/ndk/guides/cpp-support#one_stl_per_app )
To have only one stl per app changing ANDROID_STL way to c++_shared, that will add libc++_shared.so to packaging.
Test Plan: Imported from OSS
Differential Revision: D22118031
Pulled By: IvanKobzarev
fbshipit-source-id: ea1e5085ae207a2f42d1fa9f6ab8ed0a21768e96
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/39868
### Summary
why disable NNPACK on iOS
- To stay consistency with our internal version
- It's currently blocking some external users due to its lack support of x86 architecture
- https://github.com/pytorch/pytorch/issues/32040
- https://discuss.pytorch.org/t/undefined-symbols-for-architecture-x86-64-for-libtorch-in-swift-unit-test/84552/6
- NNPACK uses fast convolution algorithms (FFT, winograd) to reduce the computational complexity of convolutions with large kernel size. The algorithmic speedup is limited to specific conv params which are unlikely to appear in mobile networks.
- Since XNNPACK has been enabled, it performs much better than NNPACK on depthwise-separable convolutions which is the algorithm being used by most of mobile computer vision networks.
### Test Plan
- CI Checks
Test Plan: Imported from OSS
Differential Revision: D22087365
Pulled By: xta0
fbshipit-source-id: 89a959b0736c1f8703eff10723a8fbd02357fd4a
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/39999
Cleaned up the android build scripts. Consolidated common functions into
common.sh. Also made a few minor fixes:
- We should trust build_android.sh doing right about reusing existing
`build_android_$abi` directory;
- We should clean up `pytorch_android/src/main/jniLibs/` to remove
broken symbolic links in case custom abi list changes since last build;
Test Plan: Imported from OSS
Differential Revision: D22036926
Pulled By: ljk53
fbshipit-source-id: e93915ee4f195111b6171cdabc667fa0135d5195
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/39188
Extracting Vulkan_LIBS and Vulkan_INCLUDES setup from `cmake/Dependencies.cmake` to `cmake/VulkanDependencies.cmake` and reuse it in android/pytorch_android/CMakeLists.txt
Adding control to build with Vulkan setting env variable `USE_VULKAN` for `scripts/build_android.sh` `scripts/build_pytorch_android.sh`
We do not use Vulkan backend in pytorch_android, but with this build option we can track android aar change with `USE_VULKAN` added.
Currently it is 88Kb.
Test Plan: Imported from OSS
Differential Revision: D21770892
Pulled By: IvanKobzarev
fbshipit-source-id: a39433505fdcf43d3b524e0fe08062d5ebe0d872
Summary:
* Does a basic upload of release candidates to an extra folder within our
S3 bucket.
* Refactors AWS promotion to allow for easier development of restoration
of backups
Backup restoration usage:
```
RESTORE_FROM=v1.6.0-rc3 restore-backup.sh
```
Requires:
* AWS credentials to upload / download stuff
* Anaconda credentials to upload
Pull Request resolved: https://github.com/pytorch/pytorch/pull/38690
Differential Revision: D21691033
Pulled By: seemethere
fbshipit-source-id: 31118814db1ca701c55a3cb0bc32caa1e77a833d
Summary:
Files that were named the same within the anaconda repository, i.e.
pytorch_1.5.0-cpu.bz2, were found to be clobbering each other,
especially amongst different platforms.
This lead to similarly named packages for different platforms to not get
promoted.
This also adds "--skip" to our anaconda upload so that we don't end up
overwriting our releases just in case this script gets run twice.
Also, conda search ends up erroring out if it doesn't find anything for
the current platform being searched for so we should just continue
forward if we don't find anything since we want to be able to use this
script for all of the packages we support which also do not release
packages for all of the same platforms. (torchtext for example only has
"noarch")
This should also probably be back-ported to the `release/1.5` branch since this changeset was used to release `v1.5.0`
Signed-off-by: Eli Uriegas <eliuriegas@fb.com>
Pull Request resolved: https://github.com/pytorch/pytorch/pull/37089
Differential Revision: D21184768
Pulled By: seemethere
fbshipit-source-id: dbe12d74df593b57405b178ddb2375691e128a49
Summary:
Adds a new promotion pipeline for both our wheel packages hosted on S3
as well as our conda packages hosted on anaconda.
Promotion is only run on tags that that match the following regex:
/v[0-9]+(\.[0-9]+)*/
Example:
v1.5.0
The promotion pipeline is also only run after a manual approval from
someone within the CircleCI security context "org-member"
> NOTE: This promotion pipeline does not cover promotion of packages that
> are published to PyPI, this is an intentional choice as those
> packages cannot be reverted after they have been published.
TODO: Write a proper testing pipeline for this
Signed-off-by: Eli Uriegas <eliuriegas@fb.com>
Pull Request resolved: https://github.com/pytorch/pytorch/pull/34993
Differential Revision: D20539497
Pulled By: seemethere
fbshipit-source-id: 104772d3c3898d77a24ef9bf25f7dbd2496613df
Summary:
How this actually works:
1. Get's a list of URLs from anaconda for pkgs to download, most
likely from pytorch-test
2. Download all of those packages locally in a temp directory
3. Upload all of those packages, with a dry run upload by default
This, along with https://github.com/pytorch/pytorch/issues/34500 basically completes the scripting work for the eventual promotion pipeline.
Currently testing with:
```
TEST_WITHOUT_GIT_TAG=1 TEST_PYTORCH_PROMOTE_VERSION=1.4.0 PYTORCH_CONDA_FROM=pytorch scripts/release/promote/conda_to_conda.sh
```
Signed-off-by: Eli Uriegas <eliuriegas@fb.com>
Pull Request resolved: https://github.com/pytorch/pytorch/pull/34659
Differential Revision: D20432687
Pulled By: seemethere
fbshipit-source-id: c2a99f6cbc6a7448e83e666cde11d6875aeb878e
