Skip to content

Pull requests: ROCm/aiter

New pull request New
328 Open 2,681 Closed
328 Open 2,681 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

[GLUON] Moe a8w4 optimizations
#3317 opened May 23, 2026 by lburzawa Contributor Loading…
1 task done
1
[ck gemm a8w8 blockscale] shape-aware kernel selection heuristic for untuned shapes
#3316 opened May 22, 2026 by eppaneamd Contributor Loading…
6
Align flydsl buffer resource size usage
#3314 opened May 22, 2026 by coderfeli Collaborator • Draft
1
Zhimding/fix aot moe pad 0522
#3313 opened May 22, 2026 by coderfeli Collaborator Loading…
10
[MOE] Add fused dynamic MXFP8 quant + moe_sort HIP path
#3312 opened May 22, 2026 by yadaish Contributor • Draft
1 task
1
[Triton-Gluon-MLA-GFX950] add decode kernel for nhead<=16
#3311 opened May 22, 2026 by Dewei-Wang-sh Contributor Loading…
3 tasks done
@Dewei-Wang-sh
9
add asmjit AOT kernels for qwen35/Hunyuan3
#3309 opened May 22, 2026 by tingqli Loading…
1 task
14
Replace fmha with a new kernel
#3308 opened May 22, 2026 by JohnNikolay84 Contributor Loading…
1 task
1
[Triton][CDNA4] Optimize gluon blockscale a8w8 gemm kernel
#3307 opened May 21, 2026 by lijinpei-amd Contributor Loading…
1
mla: add fp8 qh32 seqlen=1 persistent kernel support on gfx950
#3304 opened May 21, 2026 by alexioslyrakis-amd Contributor Loading…
1 task done
1
PA: route uniform shuffled-FP8 MTP decode (qlen 2/3/4) to pa_fwd_asm
#3302 opened May 21, 2026 by vstakhov-amd Loading…
5 tasks done
1
CI: add MoE perf regression check (bench_moe)
#3300 opened May 21, 2026 by zhiding512 Contributor Loading…
1 task
7
[moe][perf] tune qwen235b a16w16 moe config
#3298 opened May 21, 2026 by lalala-sh Contributor Loading…
1 task done
1
add pageattention with sliding window
#3297 opened May 21, 2026 by ChengYao-amd Loading…
1 task
1
repro(pa-asm): standalone reproducer for fp8 PA OOB at bs=128, qlen=3
#3295 opened May 21, 2026 by yhl-amd Loading…
3 tasks
8
[Triton] [ATOM] MXFP8 GEMM and A8W4 MOE optimization
#3292 opened May 20, 2026 by k50112113 Contributor Loading…
1
Workaround custom_all_reduce issue (By Chengzhe Fan @zx3xyy)
#3290 opened May 20, 2026 by zjing14 Contributor Loading…
2
Fix flaky dropout tests
#3289 opened May 20, 2026 by micmelesse Contributor Loading…
1 task
2
Add Kimi K2.5/K2.6 FP4 fused MoE tunings for TP2 (inter_dim=1024, 385 experts, top-9)
#3287 opened May 20, 2026 by xaguilar-amd Contributor Loading…
1
[Triton] [ATOM] DSV4 mxfp8 GEMM
#3286 opened May 20, 2026 by k50112113 Contributor • Draft
1
Add GLM-4.7-FP8 tuned/untuned BF16 GEMM configs (gfx950)
#3285 opened May 20, 2026 by omirosh Contributor Loading…
2 tasks
2
Tune gfx950 batched GEMM configs and benchmarks
#3283 opened May 20, 2026 by nidal567 Contributor • Draft
1 task done
14
Jim/dev/mi400 mha bwd
#3281 opened May 20, 2026 by slippedJim Contributor Loading…
1 task
4
[FlyDSL][MOE] Enable a8w8 blockscale moe splitk in flydsl
#3280 opened May 20, 2026 by lalala-sh Contributor Loading…
1 task done
6
[FLYDSL] Add mixed slicek-splitk policies for flydsl hgemm
#3279 opened May 20, 2026 by xytpai Contributor Loading…
1
ProTip! Find all pull requests that aren't related to any open issues with -linked:issue.