Skip to content

Pull requests: warpfront/hipfire

New pull request New
19 Open 435 Closed
19 Open 435 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

fix(generate): seed sequential AR sampler per request instead of fixed 0x13579BDF
#627 opened Aug 23, 2026 by ghazni101 Contributor • Draft
Multi-slot VL support; quantizer F16-fallback fix; slot engine VRAM + context-cap fixes
#626 opened Aug 22, 2026 by ghazni101 Contributor • Draft
perf(loader): UMA-gated eviction + parallel warmer + zero-copy uploads — 2-3.4x faster dense loads, ~6x faster A3B MoE loads on dGPUs; fix sequential sampler seed
#617 opened Aug 22, 2026 by ghazni101 Contributor • Draft
1
feat(gfx11): add optional CK prefill sidecar and prebuilt bundle
#616 opened Aug 22, 2026 by HUSRCF Contributor • Draft
13 of 14 tasks
fix(windows): daemon discovery, diag compiler probe, and quantize build
#615 opened Aug 21, 2026 by creazyboyone Loading…
8 of 22 tasks
4
fix(quantize): gemma4 GGUF config + tensor-name translation
#613 opened Aug 21, 2026 by noctrex Loading…
Fix hipfire-quantize compilation on Windows
#612 opened Aug 21, 2026 by noctrex Loading…
perf(gfx1100): batch-tiled B=2 WMMA GEMM variants for prefill (+14% 4B / +29% 27B, p<0.01)
#611 opened Aug 20, 2026 by ghazni101 Contributor • Draft
quant: Ornith 1.5 35B-A3B — SKU, MTP sidecar, VL, and gfx11 qt44 kernels meta:wip Open, but WIP PRs
#610 opened Aug 20, 2026 by nwoolmer Collaborator Loading…
feat(serve): OpenAI client parity on the multi-slot path — tools, reasoning_effort, sampling, usage, tool-iteration reuse
#609 opened Aug 20, 2026 by alpineQ Contributor Loading…
8 tasks done
fix(serve): multi-slot engine fits 24 GB cards — skip GDN S-tape, chunk prefill scratch, apply device visibility
#608 opened Aug 20, 2026 by alpineQ Contributor Loading…
9 tasks done
fix(config): register serve.multi_slot_slots, multi_slot_ctx and multi_slot_prefill_chunk
#607 opened Aug 20, 2026 by alpineQ Contributor Loading…
6 tasks done
fix(scripts): close unterminated quote that broke serve_concurrency_gate.sh parsing
#606 opened Aug 20, 2026 by alpineQ Contributor Loading…
4 tasks done
quant: absorb #597 Bonsai (qt40/41), add MQ4 v2 (qt=44), MQ4C (qt=45), DFlash2, and the product ladder
#599 opened Aug 18, 2026 by Kaden-Schutt Collaborator Loading…
quant: add Bonsai TQ2G128 (qt=40) and BQ1G128 (qt=41)
#597 opened Aug 17, 2026 by nwoolmer Collaborator Loading…
Fixed character encoding for arch-specific generate loops.
#595 opened Aug 16, 2026 by fgheorghe • Draft
Fix tool calls being silently dropped on DeepSeek V4 and Cohere2MoE
#593 opened Aug 16, 2026 by fgheorghe • Draft
feat(ds4): add DeepSeek-V4-Flash-0731 — routed-expert paging, +28.8% cache budget, and an experimental prefetch adapter meta:wip Open, but WIP PRs
#563 opened Aug 3, 2026 by nwoolmer Collaborator • Draft
4
refactor: unify device mesh, model parallelism, and generation dispatch
#527 opened Jul 12, 2026 by fivetide Collaborator • Draft
14 of 42 tasks
5
ProTip! Follow long discussions with comments:>50.