deepseek-ai/DeepGEMM
1
0
Fork 0
mirror of https://github.com/deepseek-ai/DeepGEMM.git synced 2026-11-05 00:47:26 +00:00
Code Issues 162 Projects Packages Wiki Activity

[GH-ISSUE #92] AssertionError: best_smem_config is not None #50

New issue
Closed
opened 2026-10-26 21:11:14 +00:00 by bsakaguchi · 1 comment
bsakaguchi commented 2026-10-26 21:11:14 +00:00
Owner
Copy link

Originally created by @hewr2010 on GitHub (Apr 30, 2025).
Original GitHub issue: https://github.com/deepseek-ai/DeepGEMM/issues/92

A recent update breaks MLA forward_absorb

With DeepSeek-V3, we have q_nope [S,128,128] and q_nope_val [128,ceil(S/256)*256,128] after fp8 quant. We have k=q_nope_val.shape[2]=128 and this line would lead to empty stage_candidates

Originally created by @hewr2010 on GitHub (Apr 30, 2025). Original GitHub issue: https://github.com/deepseek-ai/DeepGEMM/issues/92 A recent update breaks MLA [forward_absorb](https://github.com/sgl-project/sglang/blob/v0.4.5.post3/python/sglang/srt/models/deepseek_v2.py#L708) * https://github.com/deepseek-ai/DeepGEMM/commit/d374456787c With DeepSeek-V3, we have `q_nope [S,128,128]` and `q_nope_val [128,ceil(S/256)*256,128]` after fp8 quant. We have `k=q_nope_val.shape[2]=128` and [this line](https://github.com/deepseek-ai/DeepGEMM/commit/d374456787c51ad9c4b3c5dbc3668ea3ad9bb9c2#diff-d07651f3d119035344e6b5ca25f1e6825d2a4029b3b5f843e2986269da2db2a9R133) would lead to empty `stage_candidates`
bsakaguchi commented 2026-10-26 21:11:16 +00:00
Author
Owner
Copy link

@LyricZhao commented on GitHub (May 7, 2025):

Fixed in github.com/deepseek-ai/DeepGEMM@daec8fd2fc, thanks for feedback! Close it now.

<!-- gh-comment-id:2856929593 --> @LyricZhao commented on GitHub (May 7, 2025): Fixed in https://github.com/deepseek-ai/DeepGEMM/commit/daec8fd2fc04c27ec141075e0f0c257e178ac72c, thanks for feedback! Close it now.
Sign in to join this conversation.
No Branch/Tag specified
Branches Tags
main
public-sync-20261014
0930-release
nv_dev
fix-mega-moe-fence
public-release-2609
public-release-2607
public-release-2606
mega-update
public-release-260416
ko3n1g/ci/build-wheels
nv_dev_f8e8fb5
nv_dev_950d31a
nv_dev_67fc648
nv_dev_c491439
nv_dev_4ff3f54
v2.1.1.post3
v2.1.1.post2
v2.1.1.post1
v2.1.1
v2.1.0
Labels
Clear labels   
pull-request
Mirrored from GitHub Pull Request

No labels
pull-request
Milestone
Clear milestone
No items
No milestone
Projects
Clear projects
No items
No project
Assignees
Clear assignees
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
deepseek-ai/DeepGEMM#50
No description provided.