fix(sglang): derive free page indices on CPU#384
Open
shipiyouniao wants to merge 1 commit into
Open
Conversation
Contributor
There was a problem hiding this comment.
Pull request overview
This PR addresses issue #382 by ensuring the SGLang elastic paged allocator computes unique page IDs on the CPU before converting them to Python integers for kvcached_allocator.free(), reducing allocator bookkeeping coupling to CUDA execution behavior.
Changes:
- Move
free_indexto CPU before derivingpage_idsviatorch.uniqueinElasticPagedTokenToKVPoolAllocator.free(). - Adjust the CPU-side conversion of
page_idsinto Python indices passed tokvcached_allocator.free(). - Add a regression test to assert the CPU move happens before
torch.uniqueand that the allocator receives expected page IDs.
Reviewed changes
Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.
| File | Description |
|---|---|
| tests/test_sglang_allocator_free_cpu.py | Adds a unit test verifying free_index is moved to CPU before unique page-id derivation and that the freed page IDs are correct. |
| kvcached/integration/sglang/patches.py | Updates the SGLang allocator free path to derive unique page IDs on CPU prior to Python conversion and freeing. |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
Comment on lines
289
to
293
| page_ids = torch.unique(free_index_cpu // self.page_size) | ||
| try: | ||
| indices: list[int] = page_ids.cpu().numpy().tolist() | ||
| indices: list[int] = page_ids.numpy().tolist() | ||
| except Exception: | ||
| indices = list(page_ids) |
shipiyouniao
force-pushed
the
fix/sglang-free-page-indices-cpu
branch
2 times, most recently
from
6259195 to
26c7de3
Compare
shipiyouniao
force-pushed
the
fix/sglang-free-page-indices-cpu
branch
from
26c7de3 to
3494825
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.
Summary
Fixes #382.
Move
ElasticPagedTokenToKVPoolAllocator.free()page-id derivation to CPU before converting free indices into Python page ids forkvcached_allocator.free().This keeps allocator bookkeeping out of the CUDA execution path and avoids relying on GPU tensor conversion behavior in the SGLang free path.
Validation
python -m pytest tests/test_sglang_allocator_free_cpu.py -qpython -m compileall -q kvcached/integration/sglang tests/test_sglang_allocator_free_cpu.pygit diff --check