[Release-Only] Use triton pypi package for 2.1.0 release

[huydhn]

This should fix pytorch/pytorch#109221 (comment) in a safe way by just replacing pytorch-triton with triton package for release.

• torch-2.1.0+cu121.with.pypi.cudnn-cp310-cp310-linux_x86_64.whl should have:

Metadata-Version: 2.1
Name: torch
Version: 2.1.0+cu121.with.pypi.cudnn
...
Requires-Dist: filelock
Requires-Dist: typing-extensions
Requires-Dist: sympy
Requires-Dist: networkx
Requires-Dist: jinja2
Requires-Dist: fsspec
Requires-Dist: nvidia-cuda-nvrtc-cu12 (==12.1.105) ; platform_system == "Linux" and platform_machine == "x86_64"
Requires-Dist: nvidia-cuda-runtime-cu12 (==12.1.105) ; platform_system == "Linux" and platform_machine == "x86_64"
Requires-Dist: nvidia-cuda-cupti-cu12 (==12.1.105) ; platform_system == "Linux" and platform_machine == "x86_64"
Requires-Dist: nvidia-cudnn-cu12 (==8.9.2.26) ; platform_system == "Linux" and platform_machine == "x86_64"
Requires-Dist: nvidia-cublas-cu12 (==12.1.3.1) ; platform_system == "Linux" and platform_machine == "x86_64"
Requires-Dist: nvidia-cufft-cu12 (==11.0.2.54) ; platform_system == "Linux" and platform_machine == "x86_64"
Requires-Dist: nvidia-curand-cu12 (==10.3.2.106) ; platform_system == "Linux" and platform_machine == "x86_64"
Requires-Dist: nvidia-cusolver-cu12 (==11.4.5.107) ; platform_system == "Linux" and platform_machine == "x86_64"
Requires-Dist: nvidia-cusparse-cu12 (==12.1.0.106) ; platform_system == "Linux" and platform_machine == "x86_64"
Requires-Dist: nvidia-nccl-cu12 (==2.18.1) ; platform_system == "Linux" and platform_machine == "x86_64"
Requires-Dist: nvidia-nvtx-cu12 (==12.1.105) ; platform_system == "Linux" and platform_machine == "x86_64"
Requires-Dist: triton (==2.1.0) ; platform_system == "Linux" and platform_machine == "x86_64"
Provides-Extra: dynamo
Requires-Dist: jinja2 ; extra == 'dynamo'
Provides-Extra: opt-einsum
Requires-Dist: opt-einsum (>=3.3) ; extra == 'opt-einsum'

On the other hand, pytorch-triton-rocm package stay the same for ROCm release.

[malfet]

I think one better delete this line altogether, as dependency should be trition==2.1.0, but TRITON_VERSION would be 2.1.0+e6216047b8

[huydhn]

I think one better delete this line altogether, as dependency should be triton==2.1.0, but TRITON_VERSION would be 2.1.0+e6216047b8

The TRITON_VERSION is fine and will be the correct 2.1.0 for release. The only concern I have is that I could only find triton=2.1.0 set here pytorch/pytorch#108423 for CUDA 12.1 (https://github.com/pytorch/pytorch/actions/runs/6162117011). So, i'm trying to understand which one set it for 11.8

[huydhn]

Note: 12.1 should be fine because it has triton from pytorch/pytorch#108423, but I'm a bit unclear about 11.8, which doesn't set pytorch_extra_install_requirements variable. torch-2.1.0+cu118-cp310-cp310-linux_x86_64.whl looks like this atm:

Metadata-Version: 2.1
Name: torch
Version: 2.1.0+cu118
...
Requires-Dist: filelock
Requires-Dist: typing-extensions
Requires-Dist: sympy
Requires-Dist: networkx
Requires-Dist: jinja2
Requires-Dist: fsspec
Requires-Dist: pytorch-triton (==2.1.0)
Provides-Extra: dynamo
Requires-Dist: jinja2 ; extra == 'dynamo'
Provides-Extra: opt-einsum
Requires-Dist: opt-einsum (>=3.3) ; extra == 'opt-einsum'

[malfet]

LGTM!