Thanks to visit codestin.com
Credit goes to github.com

Skip to content

Pull requests: ROCm/aiter

New pull request New
367 Open 2,842 Closed
367 Open 2,842 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

Add HIP fused mHC post-pre path
#3536 opened Jun 4, 2026 by LiuYinfeng01 Contributor Loading…
1 task
1
Add Radeon GPU CI smoke test
#3535 opened Jun 4, 2026 by vivienfanghuagood Contributor Loading…
1 task
1
fix(moe-tune): bound registration-barrier deadlock + harden tune() reporting; add opt-in tuning depth
#3532 opened Jun 4, 2026 by jhinpan Loading…
7
add torch compile guard
#3530 opened Jun 4, 2026 by ahmed-bsod Contributor • Draft
1
[tune] GLM-4.7-FP8 FMOE configs for EP=4 + fused shared expert (MI355x)
#3529 opened Jun 4, 2026 by omirosh Contributor Loading…
1 task
2
Fix dsv4_rotate(Hadamard) and top_k_per_row oon wave32 ci:atom
#3528 opened Jun 4, 2026 by junhaha666 Contributor Loading…
1 task
4
Route MXFP4 1-stage decode through fused quant+sort path
#3527 opened Jun 4, 2026 by akii96 Contributor • Draft
1
Fused softmax-topk-sort HIP kernel for MoE decode dispatch ci:vllm
#3526 opened Jun 4, 2026 by akii96 Contributor • Draft
1
Autotune MoE sorting dispatch: runtime kernel selection between oneshot and multi-phase paths
#3525 opened Jun 3, 2026 by akii96 Contributor • Draft
1
Fix OPUS warning on gfx1250
#3524 opened Jun 3, 2026 by Boss2002n Contributor Loading…
6
ci(sglang-downstream): add GLM-5.1-MXFP4 accuracy gate (TP2) ci:sglang
#3523 opened Jun 3, 2026 by sunway513 Collaborator Loading…
3
Integrate DS R1 GroupedTopk + Sigmoid Routing Into DS Routing
#3522 opened Jun 3, 2026 by amirumoAMD Contributor Loading…
1 task done
1
Tune Kimi2.5 A8W8 BPreshuffle GEMMs
#3520 opened Jun 3, 2026 by XiaobingSuper Contributor Loading…
2
fix(moe): route SwiGLU MXFP4 unshuffled weights to CK-Tile instead of CK2stages
#3518 opened Jun 3, 2026 by srinivamd Loading…
2
[OPUS] tensor_load_to_lds support (tcopy) for gfx1250
#3517 opened Jun 3, 2026 by demonsan Contributor Loading…
1
Fix: add missing end_sync barrier in cross_device_reduce_1stage
#3514 opened Jun 3, 2026 by zovonoir Contributor Loading…
5 of 6 tasks
1
3
[Flydsl] flydsl opt based prefill gdn hip
#3513 opened Jun 3, 2026 by huizzhan Contributor • Draft
1 task
1
enable DeepSeek-V3.2 accuracy in sglang downstream test ci:sglang
#3510 opened Jun 3, 2026 by bingxche Contributor Loading…
1 task
1
[module_pa_ragged] dead code
#3507 opened Jun 3, 2026 by amd-ruitang3 Contributor Loading…
1 task
2
test(moe_2stage): add gpt-oss legacy cases with strict accuracy check
#3505 opened Jun 3, 2026 by zhiding512 Contributor Loading…
3
Moe a8w4 optimization for decode
#3504 opened Jun 3, 2026 by lburzawa Contributor Loading…
1 task done
1
[feat](pa_mqa_logits) Support gfx1250 in deepgemm fp8 paged MQA logits
#3501 opened Jun 3, 2026 by charlieguo1106 Loading…
1
[gfx12] mla v3
#3500 opened Jun 2, 2026 by feifei14119 Contributor Loading…
1 task
5
fix(triton/decode): honor real paged KV block stride (support non-contiguous cache)
#3498 opened Jun 2, 2026 by lorri-rao Loading…
3
mxfp4fp8 fmha gfx950
#3495 opened Jun 2, 2026 by jcaraban Contributor Loading…
1 task
1
ProTip! Find all pull requests that aren't related to any open issues with -linked:issue.