🧗 Add Ascend NPU support for vLLM server #3286
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
ji-huazhong
commented
Apr 12, 2025
| @@ -114,7 +124,7 @@ def update_named_param(self, name: str, dtype: torch.dtype, shape: Sequence[int] | |||
| weight = torch.empty(shape, dtype=dtype, device=self.device) | |||
|
|
|||
| # Use NCCL to broadcast the updated weights from the client (src) to all workers. | |||
| self.pynccl_comm.broadcast(weight, src=self.client_rank, stream=torch.cuda.current_stream()) | |||
There was a problem hiding this comment.
We don't need to explicitly pass the stream to pynccl_comm, as pynccl_comm will get the current stream internally
https://github.com/vllm-project/vllm/blob/68bb122eb458a42169dd9d1772a7b32b84b2be95/vllm/distributed/device_communicators/pynccl.py#L207
ji-huazhong
commented
Apr 12, 2025
| @@ -212,7 +215,7 @@ def init_communicator(self): | |||
|
|
|||
| # Set up the communication group for weight broadcasting | |||
| pg = StatelessProcessGroup.create(host=self.host, port=self.group_port, rank=self.rank, world_size=world_size) | |||
| self.pynccl_comm = PyNcclCommunicator(pg, device="cuda:0") | |||
There was a problem hiding this comment.
Here we remove the hardcoded "cuda" to support third-party accelerators
qgallouedec
reviewed
Apr 15, 2025
qgallouedec
approved these changes
Apr 15, 2025
|
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. |
|
For the record: I can't test for Ascend NPU, but this PR introduces no regression for cuda |
qgallouedec
changed the title
shirinyamani
pushed a commit
that referenced
this pull request
Apr 16, 2025
Co-authored-by: Quentin Gallouédec <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
What does this PR do?
closes #3211
Note: This PR needs to be used with the main branch of vllm and vllm-ascend to work properly
Before submitting
Pull Request section?
to it if that's the case.
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.
cc @qgallouedec