[Hardware][Ascend]MLA for deepseek #88
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
wangxiyuan
requested changes
Feb 18, 2025
vllm_ascend/attention.py
Outdated
| kv_b_proj_weight = self.kv_b_proj.weight.reshape(self.num_heads, | ||
| self.qk_nope_head_dim + self.v_head_dim, | ||
| self.kv_lora_rank) | ||
| w_kc = kv_b_proj_weight[:, :self.qk_nope_head_dim, :].contiguous() |
There was a problem hiding this comment.
Should modify model loader to enable this on intialize stage
There was a problem hiding this comment.
Can workaround this by set this weight as attribute during runtime, this way only do slice + contiguous one time compared with this version
vllm_ascend/attention.py
Outdated
| compressType=0, calcType=0, scaleType=0, quantType=0, | ||
| inputLayout=0, outDataType=-1, attnOut=attn_output) | ||
| attn_output_t = torch_npu.npu_transpose(attn_output, (1, 0, 2), require_contiguous=True) | ||
| attn_output_t = torch_npu.npu_bmmV2(attn_output_t, w_vc, []) |
There was a problem hiding this comment.
torch.bmm can do the same maybe.
| kv = self.kv_b_proj(kv_c_normed)[0].view(num_tokens, kv_heads_num, -1) | ||
| k_nope, value = kv.split([self.qk_nope_head_dim, self.v_head_dim], dim=-1) | ||
| k_cache = torch.cat([kv_c_normed.view(num_tokens, self.num_kv_heads, -1), k_pe], dim=2) | ||
| k_pe = k_pe.repeat(1, self.num_heads, 1) |
There was a problem hiding this comment.
Can you test torch.expand here? which dose not touch global memory compared with repeat.
Signed-off-by: YHT <[email protected]>
Angazenn
pushed a commit
to Angazenn/vllm-ascend
that referenced
this pull request
Feb 21, 2025
<!-- Thanks for sending a pull request! BEFORE SUBMITTING, PLEASE READ https://docs.vllm.ai/en/latest/contributing/overview.html --> ### What this PR does / why we need it? <!-- - Please clarify what changes you are proposing. The purpose of this section is to outline the changes and how this PR fixes the issue. If possible, please consider writing useful notes for better and faster reviews in your PR. - Please clarify why the changes are needed. For instance, the use case and bug description. - Fixes # --> To adapt to the MLA structure of vLLM DeepSeek on Ascend hardware, write the AscendMLAAttentionBackendImpl class. ### Does this PR introduce _any_ user-facing change? <!-- Note that it means *any* user-facing change including all aspects such as API, interface or other behavior changes. Documentation-only updates are not considered user-facing changes. --> Users can choose to set VLLM_MLA_DISABLE to 1 or 0 to disable or enable MLA. ### How was this patch tested? <!-- CI passed with new added/existing test. If it was tested in a way different from regular unit tests, please clarify how you tested step by step, ideally copy and paste-able, so that other reviewers can test and check, and descendants can verify in the future. If tests were not added, please describe why they were not added and/or why it was difficult to add. --> Signed-off-by: YHT <[email protected]> Co-authored-by: YHT <[email protected]> Signed-off-by: angazenn <[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 this PR does / why we need it?
To adapt to the MLA structure of vLLM DeepSeek on Ascend hardware, write the AscendMLAAttentionBackendImpl class.
Does this PR introduce any user-facing change?
Users can choose to set VLLM_MLA_DISABLE to 1 or 0 to disable or enable MLA.
How was this patch tested?