Skip to content

Pull requests: ROCm/aiter

New pull request New
382 Open 3,510 Closed
382 Open 3,510 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

tuned moe config
#4287 opened Jul 19, 2026 by amd-hhashemi Contributor Loading…
1 task
5
[opt][gfx1250][ep] Add TDM deep-prefetch BF16 prefill for qk_norm_rope
#4286 opened Jul 18, 2026 by jli-melchior Contributor Loading…
1 task
[opt][gfx1250] Add TDM deep-prefetch BF16 prefill for qk_norm_rope
#4281 opened Jul 17, 2026 by jli-melchior Contributor Loading…
1 task
1
Flydsl smem migration
#4280 opened Jul 17, 2026 by xudoyuan Contributor • Draft
1 task
1
mhc bf16 compute optimize on gfx12xx
#4279 opened Jul 17, 2026 by junhaha666 Contributor • Draft
1
[gfx1250][perf][moe]Optimize prefill perf
#4278 opened Jul 17, 2026 by lalala-sh Contributor Loading…
1 task done
1
Dev/gugu fix
#4277 opened Jul 17, 2026 by yadaish Contributor Loading…
1 task
1
Add MiniMax-M3 model in Aiter - ATOM DI CI
#4275 opened Jul 17, 2026 by gyohuangxin Member • Draft
1
Add Kimi-K2.6 model in Aiter - vLLM DI CI
#4276 opened Jul 17, 2026 by gyohuangxin Member • Draft
1
Add MiniMax-M3 model in Aiter - vLLM DI CI
#4274 opened Jul 17, 2026 by gyohuangxin Member • Draft
1
[FlyDSL] Add a strided-batched variant (BMM) of the A8W8 blockscale bpreshuffle GEMM for the gfx1250
#4273 opened Jul 17, 2026 by xiangM99 Member Loading…
1 task
@xiangM99
1
Enable E2E SGLang inference on gfx1250
#4270 opened Jul 17, 2026 by akao-amd • Draft
1 task
2
[Perf] Tune DeepSeek V4 fused MoE for C32/C64 sparse decode
#4269 opened Jul 16, 2026 by Fangzhou-Ai Loading…
6 tasks done
[Triton] Add fused AdaLN-Zero (layernorm + scale/shift) kernel
#4268 opened Jul 16, 2026 by sushildubey171 Loading…
5 tasks done
2
[config] retune moe for GLM-5.2 fp4
#4266 opened Jul 16, 2026 by gbyu-amd Contributor Loading…
1 task
1
[Triton] Enable conv2d Triton kernels for gfx1100 (RDNA3) and gfx1151 (RDNA3.5)
#4265 opened Jul 16, 2026 by saeid-rostami Contributor Loading…
1 task
2
[gfx942][opus] Reduce quad inline asm and improve BF16 splitK tuning
#4263 opened Jul 16, 2026 by yifehuan Contributor • Draft
1
[MoE] Add tuned config for MiMo-V2.5-Pro prefill on MI300X
#4261 opened Jul 16, 2026 by apinge Loading…
1 task
3
CI(atom-di): track ATOM main + pin cases to 1p1d via --case
#4257 opened Jul 16, 2026 by JiaoliangYu Contributor Loading…
1
fix(triton): support paged MQA logits on gfx1201
#4255 opened Jul 16, 2026 by liminfei-amd Loading…
1
Mxfp8 gemm
#4254 opened Jul 16, 2026 by solinzby1 Contributor Loading…
1 task
6
[Triton] [Gluon] [GFX12] DSV4 Pro EP tunning (=TP1 shape)
#4253 opened Jul 16, 2026 by k50112113 Contributor Loading…
1
[FIX] Expert Map Parallel
#4252 opened Jul 15, 2026 by amirumoAMD Contributor Loading…
1 task done
1
Adapt prefill gdn
#4250 opened Jul 15, 2026 by IzacharyI Loading…
1 task
[TRITON] fused clamped-alpha SwiGLU gate activation (MiniMax-M3 / GPT-OSS)
#4249 opened Jul 15, 2026 by Chi-Chu319 Contributor Loading…
@Chi-Chu319
1
ProTip! Add no:assignee to see everything that’s not assigned.