Skip to content

fix(qwen35): reclaim prefill scratch after allocation failure #731

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
Kaden-Schutt merged 1 commit into from
Sep 9, 2026
Merged

fix(qwen35): reclaim prefill scratch after allocation failure #731

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions CHANGELOG.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7,6 +7,7 @@
- DFlash prompt-cache repair on terminal overshoot (`RepairForTerminal`) (#695).
- Template-aware primer splice (#692).
- Transactional DFlash constructors with emitter rollback (#691).
- Qwen35 prefill scratch construction retains owning tensors until publication, reclaiming partial allocations on failure so an immediate retry can reuse them.
- MQ-V2 prefill admit rule (#690).
- gfx1100 DFlash launch fusion and split-K residual tiers (#702 body, S1–S8).
- Dense-TP prefill chunking equals arch batch × tp (#725).
Expand Down
4 changes: 2 additions & 2 deletions crates/hipfire-arch-qwen35/map.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -37,7 +37,7 @@ _Generated by `scripts/check-crate-maps.py` from the tree — do not edit inside
| [`src/mtp_spec.rs`](src/mtp_spec.rs) | 3,827 | 33 | 12 |
| [`src/mtp_speculator.rs`](src/mtp_speculator.rs) | 522 | 3 | 0 |
| [`src/paro_moe.rs`](src/paro_moe.rs) | 222 | 0 | 0 |
| [`src/qwen35/batch.rs`](src/qwen35/batch.rs) | 1,698 | 16 | 0 |
| [`src/qwen35/batch.rs`](src/qwen35/batch.rs) | 1,833 | 16 | 1 |
| [`src/qwen35/config.rs`](src/qwen35/config.rs) | 1,643 | 41 | 21 |
| [`src/qwen35/ep_batch.rs`](src/qwen35/ep_batch.rs) | 4,805 | 20 | 7 |
| [`src/qwen35/forward.rs`](src/qwen35/forward.rs) | 6,251 | 31 | 12 |
Expand Down Expand Up @@ -97,6 +97,6 @@ _Generated by `scripts/check-crate-maps.py` from the tree — do not edit inside

### Totals

- 29 modules · 58,634 lines · 440 public items · 192 tests · 11 examples
- 29 modules · 58,769 lines · 440 public items · 193 tests · 11 examples

<!-- crate-map:generated:end -->
Loading
Loading