[Feature] Support engine with NPU backend.

[wangjiangben-hw]

Thanks for your contribution and we appreciate it a lot. The following instructions would make your pull request more healthy and more easily get feedback. If you do not understand some items, don't worry, just make the pull request and seek help from maintainers.

Motivation

Added ascending device support in mmcv.

Modification

• The main modification points are as follows:
  1. In mmengine/device/utils.py, We've added our NPU devices, When judging whether the device is available, try to import the npu plugin package.
  2. In mmengine/dist/, We have added NPU-related distributed initialization methods.
  3. In mmengine/optim/optimizer/ and mmengine/runner/amp.py, We've added some judgment to make the mixed precision mode turn on automatically on the NPU and work properly.

BC-breaking (Optional)

None

Use cases (Optional)

• test case
  • tests/test_device/test_device.py

Checklist

• Pre-commit or other linting tools are used to fix the potential lint issues.
• The modification is covered by complete unit tests. If not, please add more unit test to ensure the correctness.
• If the modification has potential influence on downstream projects, this PR should be tested with downstream projects, like MMDet or MMCls.
• The documentation has been modified accordingly, like docstring or example tutorials.

[ZwwWayne]

Hi @wangjiangben-hw

Thanks for your kind contribution. Would you like to fix the lint issue so that we could merge this PR?
As this issue is fixed in the most recent main branch, you cam simply fix it by rebasing your code onto the most recent main.

[zhouzaida]

• Test hccl backend for dist module. https://github.com/open-mmlab/mmengine/tree/main/tests/test_dist
• Add npu method in BaseModel

mmengine/mmengine/model/base_model/base_model.py

Lines 198 to 211 in 13484aa

	def cuda(
	self,
	device: Optional[Union[int, str, torch.device]] = None,
	) -> nn.Module:
	"""Overrides this method to call :meth:`BaseDataPreprocessor.cuda`
	additionally.
	Returns:
	nn.Module: The model itself.
	"""
	if device is None or isinstance(device, int):
	device = torch.device('cuda', index=device)
	self._set_device(torch.device(device))
	return super().cuda(device)

[wangjiangben-hw]

• Test hccl backend for dist module. https://github.com/open-mmlab/mmengine/tree/main/tests/test_dist
• Add npu method in BaseModel

mmengine/mmengine/model/base_model/base_model.py

Lines 198 to 211 in 13484aa

	def cuda(
	self,
	device: Optional[Union[int, str, torch.device]] = None,
	) -> nn.Module:
	"""Overrides this method to call :meth:`BaseDataPreprocessor.cuda`
	additionally.
	Returns:
	nn.Module: The model itself.
	"""
	if device is None or isinstance(device, int):
	device = torch.device('cuda', index=device)
	self._set_device(torch.device(device))
	return super().cuda(device)

added * Add npu method in BaseModel

[HAOCHENYE]

is_npu_avalialble should also be added to docs/en/device.rst and docs/zh_cn/device.rst

[wangjiangben-hw]

is_npu_avalialble should also be added to docs/en/device.rst and docs/zh_cn/device.rst

added

[codecov]

Codecov Report

❗ No coverage uploaded for pull request base (main@276ca24). Click here to learn what that means.
Patch has no changes to coverable lines.

❗ Current head 402f555 differs from pull request most recent head 6b963f0. Consider uploading reports for the commit 6b963f0 to get more accurate results

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #572   +/-   ##
=======================================
  Coverage        ?   77.80%           
=======================================
  Files           ?      126           
  Lines           ?     9092           
  Branches        ?     1811           
=======================================
  Hits            ?     7074           
  Misses          ?     1701           
  Partials        ?      317           

Flag	Coverage Δ
unittests	77.80% <0.00%> (?)

Flags with carried forward coverage won't be shown. Click here to find out more.

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

☔ View full report at Codecov.
📢 Do you have feedback about the report comment? Let us know in this issue.