{{ message }}
Repository navigation
docs: propose Qwen3.5 prefill/decode serving on Ascend - #1537
Open
wangzhao-11a wants to merge 1 commit into
Open
wangzhao-11a wants to merge 1 commit into
wangzhao-11a wants to merge 1 commit into
Conversation
rtp-llm-review-bot
added
auto-lgtm:needs-approval
Collaborator
自动审查进度
更新时间：2026-10-21 20:53 北京时间。运行期间约每 5 分钟更新；阶段任务数会随核验展开，不能当作整轮完成百分比。 |
rtp-llm-review-bot
added
review:delayed
wangzhao-11a
force-pushed
the
codex/qwen35-ascend-pd-design
branch
from
October 21, 2026 09:12
727aa21 to
c32ed72
Compare
rtp-llm-review-bot
added
review:running
rtp-llm-review-bot
added
review:running
Collaborator
rtp-llm-review-bot
added
review:blocked
This branch has not been deployed
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
This proposal defines single-host Qwen3.5-35B-A3B BF16 PD serving on Ascend 950, starting with two TP=1 instances. It records the audited main commit, separates reusable model/cache/PD infrastructure from the unmerged Ascend foundation, and specifies complete hybrid-state transfer, UB memory lifetime, prefix reuse, independent graph modes and acceptance gates.
The proposal does not claim Ascend runtime or performance validation. Three- and four-card ratios remain part of the design and require their own hardware acceptance. No unmerged community implementation is included.
Validation: checked source references and current PR status against upstream; reviewed the feature-matrix arithmetic and BF16 full-attention KV estimates;
git diff --check.