Skip to content

Pull requests: ROCm/aiter

New pull request New
399 Open 3,575 Closed
399 Open 3,575 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

[Bugfix][Triton] Avoid RDNA4 unified attention LDS overflow
#4385 opened Jul 25, 2026 by hogeheer499-commits Loading…
1
1
Repining CK to f33252ce and fix build/test errors ci:all
#4384 opened Jul 25, 2026 by JiaLuo-CAN Contributor Loading…
1 task
1
[TRITON] Add gluon support for MXFP4 quant kernel in gfx950 and gfx1250
#4383 opened Jul 24, 2026 by NimitPtl Contributor • Draft
1 task
@NimitPtl
2
[TRITON][GLUON][GFX950][DSV4] Paged Sparse Attention Gluon Kernel
#4382 opened Jul 24, 2026 by cagrikymk Contributor • Draft
2
move all configs
#4379 opened Jul 24, 2026 by Boss2002n Contributor Loading…
1 task
4
[MLA] Deterministic single-split decode option for reproducible serving (fixes #4364)
#4378 opened Jul 24, 2026 by MohitAMD Loading…
1 of 2 tasks
1
4
feat(topk): deterministic tie-break-by-token-id for sparse-MLA indexer top-k
#4376 opened Jul 24, 2026 by yhl-amd Contributor Loading…
1
[Bug Fix]fix flyDSL MoE sorting graph capture break with dp attn + ep ci:atom
#4375 opened Jul 24, 2026 by amd-weisun Contributor Loading…
6
[gfx1250][FlyDSL] Refactor GEMMs with layout-based API
#4374 opened Jul 24, 2026 by aoli26 Contributor Loading…
1 task done
8
[CK][VSA] Add thin sparse attention operator
#4373 opened Jul 24, 2026 by LiuYinfeng01 Contributor Loading…
6 tasks done
1
Implement FlyDSL version of fused_qk_norm_mrope_3d_cache_pts_quant_shuffle
#4371 opened Jul 24, 2026 by amd-meskelin • Draft
1 task
1
[MI355] add 128x128 block-scales fp8 8wave moe/gemm kernels from pyhip
#4370 opened Jul 24, 2026 by tingqli • Draft
1 task
1
[module_moe_asm] refactor and rm torch ci:all
#4368 opened Jul 24, 2026 by amd-ruitang3 Contributor Loading…
1
feat: support fp32 chunk states in GDN prefill
#4366 opened Jul 24, 2026 by junna2016 Loading…
2
[Bugfix][MLA] Gate gfx942 native qh64 fp8 decode to page_size=64 (fix page_size=1 GPU fault)
#4365 opened Jul 24, 2026 by MohitAMD Loading…
3 tasks done
1
3
[WIP] Add unit-scale FP8 KV cache support to fused_qknorm_idxrqknorm
#4362 opened Jul 23, 2026 by zijiecode • Draft
1
perf(expt_data): skip redundant TokenStart/TileStart stores in stage1
#4361 opened Jul 23, 2026 by amd-hhashemi Contributor Loading…
6
Swap the cache config for the default cases of gfx1250-GEMM-A8W8_BLOCKSCALE.json
#4359 opened Jul 23, 2026 by jpvillam-amd Loading…
1 task
@jpvillam-amd
1
Chefang mha global load
#4357 opened Jul 23, 2026 by fangche123 Contributor Loading…
1 task
4
Jeongkim/silotiger 699 gfx950 port
#4355 opened Jul 23, 2026 by JH-Leon-KIM-AMD Loading…
2 tasks done
32
feat(flydsl): Add HSTU Forward kernel
#4354 opened Jul 23, 2026 by damien-lejeune Loading…
1 task done
1
[FLYDSL] mfma16_hip GDR K5 prefill chunk_gdn_fwd_h for MI308
#4353 opened Jul 23, 2026 by huizzhan Contributor Loading…
1 task
1
fix(mla): refresh gfx950 MLA HSACO batch for large page_id
#4351 opened Jul 23, 2026 by fangche123 Contributor Loading…
1 task
2
Aiterker 112 asm ptl1
#4348 opened Jul 23, 2026 by JohnNikolay84 Contributor Loading…
1 task
@JohnNikolay84
2
Fix: add missing end_sync barrier in fused allreduce+rmsnorm kernel
#4346 opened Jul 23, 2026 by yuzho-amd Loading…
1 task
2
ProTip! Find all pull requests that aren't related to any open issues with -linked:issue.