Release FBGEMM v0.5.0

Fix jagged_test_index_select_2d that hangs in OSS and revert skip tes…

…ts (pytorch#2036)

Summary:
Pull Request resolved: pytorch#2036

Before c++20, `std::atomic_flag` is initialized to an unspecified state, hence the loop `while (lock.test_and_set(std::memory_order_acquire)` is never broken and causes the test to hang in OSS. This diff properly initializes the `std::atomic_flag`.

Reviewed By: q10, sryap

Differential Revision: D49528661

fbshipit-source-id: ba2213cb9bf8c0abbd1e169db03f0e32dd2a7ebb

[FBGEMM][v0.5.0-rc3] Fix using package version and missing bash path

Skip PooledEmbeddingModulesTest until FailedHealthCheck is fixed (pyt…

…orch#1999)

Summary:
Pull Request resolved: pytorch#1999

Hypothesis version 6.83.2 onwards introduces
`HealthCheck.differing_executors` that causes tests
in`permute_pooled_embedding_test.py` to fail with error:

`The method PooledEmbeddingModulesTest.setUp was called from multiple different executors. This may lead to flaky tests and nonreproducible errors when replaying from database`.

Currently, we're using the latest version of hypothesis on CI:

https://github.com/pytorch/FBGEMM/actions/runs/6084855480/job/16515052387

Current hypothesis on FBCode is 6.70.1 which does not have
`HealthCheck.differing_executors`.

Reviewed By: shintaro-iwasaki

Differential Revision: D49020046

fbshipit-source-id: 8ab1350411260c771baf05efe607f91c12df2385

Add support for building FBGEMM_GPU against Python 3.11 in OSS (pytor…

…ch#1646)

Summary:
- Parallelize the FBGEMM CI builds to build and test static and shared libraries independently instead of in serial
- Move the FBGEMM CI builds to run inside Docker containers
- Add support for building FBGEMM_GPU against Python 3.11 in OSS
- Move all FBGEMM_GPU nightly and release build jobs to run inside `amazonlinux:2023` Docker container
- Assuming no build errors or resource starvation, the full OSS build process now runs under 30 minutes.

Pull Request resolved: pytorch#1646

Reviewed By: shintaro-iwasaki

Differential Revision: D44157228

Pulled By: q10

fbshipit-source-id: 6403ea9955856157785c50837b0b8e4c0cd26d53

Add support for building FBGEMM_GPU against Python 3.11 in OSS (pytor…

…ch#1646)

Summary:
- Parallelize the FBGEMM CI builds to build and test static and shared libraries independently instead of in serial
- Move the FBGEMM CI builds to run inside Docker containers
- Add support for building FBGEMM_GPU against Python 3.11 in OSS
- Move all FBGEMM_GPU nightly and release build jobs to run inside `amazonlinux:2023` Docker container
- Assuming no build errors or resource starvation, the full OSS build process now runs under 30 minutes.

Pull Request resolved: pytorch#1646

Reviewed By: shintaro-iwasaki

Differential Revision: D44157228

Pulled By: q10

fbshipit-source-id: 6403ea9955856157785c50837b0b8e4c0cd26d53

include cstdint (pytorch#1640)

Summary:
fix build with gcc-13

Pull Request resolved: pytorch#1640

Reviewed By: shintaro-iwasaki

Differential Revision: D44044422

Pulled By: q10

fbshipit-source-id: 692ec9c34f4aaf726294a2b643fbceabf8159033

Remove Local Version Identifier (pytorch#1631)

Summary:
- Remove local version identifier from build versioning
- Use CPU-only instances for CPU-only tests

Pull Request resolved: pytorch#1631

Reviewed By: shintaro-iwasaki

Differential Revision: D43915104

Pulled By: q10

fbshipit-source-id: 223845c1c38d911f650d5db926aa7f2f59561b85

release_20221214 3

Release 0.3.0