[Perf] Optimize perf of Qwen3 #1245
Closed
Conversation
This file contains hidden or 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
Contributor
|
Can you add e2e testing? I want to try it out. |
|
This pull request has conflicts, please resolve those before we can evaluate the pull request. |
Signed-off-by: rjg-lyh <1318825571@qq.com>
MengqingCao
reviewed
Jun 26, 2025
|
|
||
|
|
||
| class AddRMSNormQuant(RMSNorm): | ||
| """Root mean square normalization. |
Collaborator
There was a problem hiding this comment.
Please update the comment
| self.post_attention_layernorm = RMSNorm(config.hidden_size, | ||
| eps=config.rms_norm_eps) | ||
| else: | ||
| from vllm_ascend.quantization.quant_config import AscendQuantConfig |
Collaborator
There was a problem hiding this comment.
It seems the mainly changes on CustomQwen3DecoderLayer is the AddRMSNormQuant layer. I prefer to inheiret from Qwen3DecoderLayer and add the logic of AddRMSNormQuant. This could make the optimization point clearly and reduce redundant code
| import torch_npu | ||
|
|
||
| if residual is not None: | ||
| x, _, residual = torch_npu.npu_add_rms_norm_quant(x, residual, self.weight, |
Collaborator
There was a problem hiding this comment.
QQ: what does "add" mean here?
|
This pull request has conflicts, please resolve those before we can evaluate the pull request. |
Collaborator
|
Yes, we should avoid to add big paste code in vllm-ascend. Please also paste perf results here. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
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Outdated suggestions cannot be applied.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 this PR does / why we need it?
Optimize the performance of Qwen3 model by registering a custom model.
Does this PR introduce any user-facing change?
No.
How was this patch tested?
CI passed with existing test.