[feat] Enable MLA chunked prefill and KV cache reuse on SM121#15347
Open
CodersAcademy006 wants to merge 2 commits into
Open
[feat] Enable MLA chunked prefill and KV cache reuse on SM121#15347CodersAcademy006 wants to merge 2 commits into
CodersAcademy006 wants to merge 2 commits into
Conversation
Contributor
📝 WalkthroughWalkthroughIn ChangesMLA SM121 allowlist expansion
Estimated code review effort🎯 1 (Trivial) | ⏱️ ~2 minutes 🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches🧪 Generate unit tests (beta)
Comment |
Signed-off-by: Srijan Upadhyay <srjnupadhyay@gmail.com>
65577f3 to
156e45b
Compare
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
I updated the SM version check logic in
py_executor_creator.pyto allow MLA chunked prefill and KV cache block reuse on SM121 (Blackwell) architectures. Specifically, I added SM121 (121) to the validation lists for bothenable_block_reuseandenable_chunked_contextchecks, preventing the executor from automatically disabling these features on Blackwell GPUs.This resolves issue #15344, where these optimization features were being disabled on SM121 devices because the Python-side validator was missing SM121 in its allowlist. The underlying C++ kernels already support SM121, so this change enables full compatibility for MLA optimizations on this hardware.