Skip to content

Pull requests: ROCm/aiter

New pull request New
563 Open 4,012 Closed
563 Open 4,012 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

[UT] Support a4w4 in test_mega_moe
#5052 opened Aug 27, 2026 by yanboshao Collaborator Loading…
1 task
1
[Triton/Gluon] [gfx950] add an optimized prefill fp8_mqa_logits for H64D128
#5048 opened Aug 27, 2026 by Dewei-Wang-sh Contributor Loading…
6
[HIP] [JIT] fp8_paged_mqa_logits: hand-written gfx950 decode indexer kernel
#5047 opened Aug 27, 2026 by sumin-hong Loading…
1 task done
6
[HIP] [JIT] fp8_mqa_logits: hand-written gfx950 prefill indexer kernel
#5046 opened Aug 27, 2026 by sumin-hong Loading…
1 task done
3
[CK] [FlyDSL] Retune GLM5.2 mxfp4 MoE and fix a scale-view cache leak
#5045 opened Aug 27, 2026 by XiaobingSuper Contributor Loading…
11
[ASM] [HIP] [gfx1250]asm mha bf16 hd192x128
#5043 opened Aug 27, 2026 by shay-li77 Contributor Loading…
1 task
9
[HIP] fix(topk): correct lambda arity in ob::radix_kernel / last_filter_kernel
#5042 opened Aug 27, 2026 by JohnQinAMD Contributor Loading…
4
[FlyDSL] [gfx1250] Add batched a8w8 mxscale_128 gemm
#5041 opened Aug 27, 2026 by aoli26 Contributor Loading…
1 task done
5
[Triton/Gluon] Moe routing optimizations
#5038 opened Aug 27, 2026 by lburzawa Contributor Loading…
1 task done
1
[Triton/Gluon] MOE a8w4 cudagraph updates
#5037 opened Aug 26, 2026 by lburzawa Contributor Loading…
1 task done
1
[Triton/Attention] Clamp unified_attention 3D config to the LDS budget on gfx1201
#5035 opened Aug 26, 2026 by lobo235 • Draft
1
2
[HIP] [Bugfix] Fix DSV4 FP4 KV-cache scattered row writes
#5034 opened Aug 26, 2026 by AMD-yanfeiwang Contributor Loading…
1
[Triton/Gluon] Tune MoE GEMM A8W8 ci:triton-300x ci:triton-355
#5033 opened Aug 26, 2026 by nidal567 Contributor Loading…
1 task done
@nidal567
1
[Triton/Gluon] Add bench for gluon/triton mxfp8 gemm
#5029 opened Aug 26, 2026 by k50112113 Contributor Loading…
1
[Triton/Gluon] Tune MoE GEMM A8W8 blockscale ci:triton-300x ci:triton-355
#5028 opened Aug 26, 2026 by nidal567 Contributor Loading…
1 task done
@nidal567
1
[ROCm][Perf] Add head_dim 512 + weightless V-norm to fused_qk_norm_rope_cache_pts_quant_shuffle
#5027 opened Aug 26, 2026 by mpashkovskii • Draft
1
[FlyDSL] jdbmm backward pass
#5025 opened Aug 26, 2026 by SamiAario-AMD • 2/2 Loading…
7 of 8 tasks
@SamiAario-AMD
20
[CK] Add MHA forward tuning scripts and kernel-info dumping utit…
#5024 opened Aug 26, 2026 by huishi-hs Loading…
1 task
1
[Triton/Gluon] Consolidate and reorganize ops/triton utils
#5023 opened Aug 26, 2026 by Boss2002n Contributor Loading…
9
[Triton/Gluon] Migrate the MHC tuned configs to the nested layout
#5021 opened Aug 26, 2026 by Boss2002n Contributor Loading…
7
[Triton/Gluon] Migrate conv configs to the nested arch/backend layout
#5018 opened Aug 26, 2026 by Boss2002n Contributor Loading…
10
[FlyDSL] Add TopK
#5011 opened Aug 26, 2026 by lirui927 Contributor • Draft
1 task
7
[Triton/Gluon] Support caller-defined padding cache slot in fused MLA writer
#5010 opened Aug 26, 2026 by tanth47 Loading…
2
[HIP] [Feature] Radix-select top-k for wide ungrouped MoE routers ci:all
#5009 opened Aug 26, 2026 by fanxingran Loading…
1 task
3
[FlyDSL] One-stage split-K for the a8w8 preshuffle GEMM ci:atom
#5007 opened Aug 26, 2026 by XiaobingSuper Contributor Loading…
21
ProTip! Add no:assignee to see everything that’s not assigned.