MLU devices : Checks if mlu is available via an cndev-based check which won't trigger the drivers and leave mlu

[huismiling]

What does this PR do?

MLU devices : Checks if mlu is available via an cndev-based check which won't trigger the drivers and leave mlu

Before submitting

• This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
• Did you read the contributor guideline,
  Pull Request section?
• Was this discussed/approved via a Github issue or the forum? Please add a link
  to it if that's the case.
• Did you make sure to update the documentation with your changes? Here are the
  documentation guidelines, and
  here are tips on formatting docstrings.
• Did you write any new necessary tests?

Who can review?

Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.

[huismiling]

@ArthurZucker
Hi, Could you help to merge this PR ?

[ArthurZucker]

Hey is there any doc regarding PYTORCH_CNDEV_BASED_MLU_CHECK anywhere? Or are we just using / creating is for this?

[HuggingFaceDocBuilderDev]

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.

[huismiling]

Hey is there any doc regarding PYTORCH_CNDEV_BASED_MLU_CHECK anywhere? Or are we just using / creating is for this?

@ArthurZucker
Hi, Just using / creating PYTORCH_CNDEV_BASED_MLU_CHECK is OK.
Like nvml, CNDEV is a cambricon library for MLUs, . It will not affect pytorch but torch_mlu.

nvml-based and cndev-based check has been used in accelerate.
huggingface/accelerate#3187

https://github.com/huggingface/accelerate/blob/ba7ab93f5e688466ea56908ea3b056fae2f9a023/src/accelerate/utils/imports.py#L116

def is_cuda_available():
    """
    Checks if `cuda` is available via an `nvml-based` check which won't trigger the drivers and leave cuda
    uninitialized.
    """
    with patch_environment(PYTORCH_NVML_BASED_CUDA_CHECK="1"):
        available = torch.cuda.is_available()

    return available

https://github.com/huggingface/accelerate/blob/ba7ab93f5e688466ea56908ea3b056fae2f9a023/src/accelerate/utils/imports.py#L322

def is_mlu_available(check_device=False):
    """
    Checks if `mlu` is available via an `cndev-based` check which won't trigger the drivers and leave mlu
    uninitialized.
    """
    if importlib.util.find_spec("torch_mlu") is None:
        return False

    import torch_mlu  # noqa: F401

    with patch_environment(PYTORCH_CNDEV_BASED_MLU_CHECK="1"):
        available = torch.mlu.is_available()

    return available

[huismiling]

@ArthurZucker
Hi, is this PR okay to merge ? Is there anything else I can help ?