Clean up some unused build env variables (#158599)
* Parameter build-with-debug isn't needed, it isn't even passed into Docker. Debug build is detected via the build environment name
* AWS_DEFAULT_REGION is a leftover from ARC and isn't used anywhere in .ci/pytorch nor .github
Signed-off-by: Huy Do <huydhn@gmail.com>
Pull Request resolved: #158599
Approved by: https://github.com/cyyever, https://github.com/ZainRizvi
ghstack dependencies: #158598

enable_caching_generated_triton_templates=True by default (#158592)
Got some risk, but good to catch issues if there is any, easy to revert single flag flip.
Pull Request resolved: #158592
Approved by: https://github.com/eellison

Revert "Add warning about removed sm50 and sm60 arches (#158301)"
This reverts commit fb731fe.
Reverted #158301 on behalf of https://github.com/facebook-github-bot due to Diff reverted internally ([comment](#158301 (comment)))

[dynamo] Support more basic output types for `nonstrict_trace` (#157969)
Fixes #157397 and improves the user-facing error message for remaining
unsupported cases.
Pull Request resolved: #157969
Approved by: https://github.com/zou3519

[AOTI] windows package load dev (#158671)
changes:
1. add extract file fail handler for Windows develop.
2. normalize more file paths.
Pull Request resolved: #158671
Approved by: https://github.com/angelayi

Revert "[Easy] Show some clear error when torch.ops.load_library fail…
…s. (#157524)"
This reverts commit 555f356.
Reverted #157524 on behalf of https://github.com/wdvr due to reverting for now to reopen the discussion ([comment](#157524 (comment)))

[cca] [c10d] Refactor CUDAEventCache into separate files (#158616)
Summary:
Refactored CUDAEventCache from ProcessGroupNCCL.hpp/.cpp into dedicated header and implementation files for better code organization and maintainability.
Split out CUDAEventCache into:
- New header file: CUDAEventCache.hpp
- New implementation file: CUDAEventCache.cpp
- Updated build_variables.bzl to include the new file
This change improves code maintainability, readability, and follows better code organization practices.
---
> Generated by [Confucius Code Assist (CCA)](https://www.internalfb.com/wiki/Confucius/Analect/Shared_Analects/Confucius_Code_Assist_(CCA)/)
[Session](https://www.internalfb.com/confucius?session_id=61b9029a-636b-11f0-9d9a-f1bcc55be1ce&tab=Chat), [Trace](https://www.internalfb.com/confucius?session_id=61b9029a-636b-11f0-9d9a-f1bcc55be1ce&tab=Trace)
Test Plan:
Verified build with:
```
buck build //caffe2/test/distributed:c10d
```
---
> Generated by [Confucius Code Assist (CCA)](https://www.internalfb.com/wiki/Confucius/Analect/Shared_Analects/Confucius_Code_Assist_(CCA)/)
[Session](https://www.internalfb.com/confucius?session_id=61b9029a-636b-11f0-9d9a-f1bcc55be1ce&tab=Chat), [Trace](https://www.internalfb.com/confucius?session_id=61b9029a-636b-11f0-9d9a-f1bcc55be1ce&tab=Trace)
Pull Request resolved: #158616
Approved by: https://github.com/fduwjj

Build domain libraries on the build job (#158600)
By setting the name of the domain libraries to build via `BUILD_ADDITIONAL_PACKAGES` environment variable, the build job will build them and make them available as artifacts in the same way as the PyTorch CI wheel. To ensure that this doesn't break CI, the test job will still build them as usual if the wheels are not there.  Building dependencies like FBGEMM on the test job is bad, especially for GPU jobs, because it leave the GPU resource idle
Fixes #152024
Signed-off-by: Huy Do <huydhn@gmail.com>
Pull Request resolved: #158600
Approved by: https://github.com/yangw-dev
ghstack dependencies: #158598, #158599

Update
[ghstack-poisoned]

Update
[ghstack-poisoned]