Conversation
Co-authored-by: Jingshu Chen <96519051+elmo2019@users.noreply.github.com>
hnyls2002
requested review from
ByronHsu,
Duyi-Wang,
HaiShaw,
ShangmingCai,
alphabetc1,
iforgetmyname,
ispobock,
ping1jing2,
sogalin and
whybeyoung
as code owners
September 10, 2026 22:58
hnyls2002
requested review from
huangtingwei9988,
hzh0425 and
xiezhq-hermann
as code owners
September 14, 2026 09:46
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
2 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
pop_preallocated/resume_retracted_reqsnow decide admission per request on the host, allocate every admitted request's KV in onealloc_extend/alloc_extend_swa_tailcall, and only then publish the transfer metadata. The per-requesttorch.tensor(..., device=cuda)copies that used to synchronize the scheduler stream on every arriving request are gone: lengths go up once through pinned memory (pinned_int64_pair).alloc_extend_swa_tailtakesswa_tail_lensper request instead of a singleswa_tail_lenwith abs == 1assert.Follows #38159; the DSV4 NPU allocator keeps its per-request hooks.
CI States
Latest PR Test (Base): ❌ Run #34829774543
Latest PR Test (Extra): ❌ Run #34829774291
Latest PR Test (AMD ROCm 10): ❌ Run #34829774408