Skip to content

Pull requests: ROCm/aiter

New pull request New
156 Open 1,976 Closed
156 Open 1,976 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

feat: add INT8/INT4 quantization support for 2-stage ASM MoE kernels
#2340 opened Mar 19, 2026 by amd-zfyu Loading…
3 tasks
feat: parallel JIT warmup for faster server startup
#2338 opened Mar 18, 2026 by sunway513 • Draft
3 of 4 tasks
1
GFX1250 Gluon MoE A4W4 Kernel
#2337 opened Mar 18, 2026 by farlukas • Draft
7 of 8 tasks
1
fix(gemm_a8w8_bpreshuffle): pass splitK/KBatch to CK kernels
#2335 opened Mar 18, 2026 by AviralGoelAMD Loading…
9
[CI][CK_TILE] Update CK and fix fmha_fwd arg init
#2334 opened Mar 18, 2026 by hyoon1 Loading…
1 task
5
update moe_smooth_quant dispatch and v2 supports block_m is a multiple of 16
#2333 opened Mar 18, 2026 by junhaha666 Loading…
1 task
4
PR to main
#2332 opened Mar 18, 2026 by Boss2002n • Draft
1 task
1
replace ck_tile type covert by opus cast
#2331 opened Mar 18, 2026 by yzhou103 Loading…
1 task
11
[OPUS] Add finfo class for float-valued type properties
#2330 opened Mar 18, 2026 by carlushuang Loading…
3
Refactor topk softmax asm bind
#2327 opened Mar 18, 2026 by yzhou103 Loading…
1 task
6
[HIP] fused split GDR decode
#2326 opened Mar 18, 2026 by huizzhan • Draft
1 task
1
add tuned config for qwen3.5 fp8,a8w8 blockscale gemm ci:sglang ci:vllm
#2324 opened Mar 18, 2026 by zovonoir Loading…
5
CK mha bwd: add sink attention score gradient support
#2321 opened Mar 18, 2026 by LJ-underdog • Draft
1 task done
6
mla ps fp8 mode get_meta_param avoid kv_tail_len < max_seqlen_q and fix nhead=128 reduce mgc
#2319 opened Mar 18, 2026 by minmengdie Loading…
1 task
4
Fix MoE dispatch for Quark W4A6 models (MXFP4 weights with QuantType.No)
#2318 opened Mar 18, 2026 by vecheruk-amd Loading…
1 task
[Triton] Fix bench_mha
#2317 opened Mar 18, 2026 by micmelesse Loading…
1 task
1
Add MPerBlock=128 tile size for blockscale FP8 MoE kernels
#2314 opened Mar 17, 2026 by ChuanLi1101 Loading…
2 tasks
3
Add blockPerCu support for CKTile GEMMs and CKTile MOE tuning ci:atom
#2313 opened Mar 17, 2026 by amd-yashagar Loading…
7
Refactor MoE ASM ops: remove PyTorch/pybind dependency, use C ABI + c…
#2308 opened Mar 17, 2026 by zufayu Loading…
1 task
6
Fix gfx950 triton test failures: invalid JSON config and tight tolerances
#2307 opened Mar 17, 2026 by azaidy Loading…
1
[TRITON] Gluon extend-attention kernel for gfx950
#2306 opened Mar 17, 2026 by realvideogame2 Loading…
12 of 16 tasks
ffm_tests_gfx1250: reduced triton_tests to ~25% high-value cases, gfx…
#2304 opened Mar 16, 2026 by Boss2002n • Draft
1 task
Improve batch prefill test coverage
#2302 opened Mar 16, 2026 by poyenc Loading…
3 tasks done
@poyenc
3
gfx1250 C++ JIT build support
#2301 opened Mar 16, 2026 by sunway513 Loading…
4 tasks
13
Made a new branch to add necessary files for batched_gemm_a16w8_block…
#2297 opened Mar 16, 2026 by nidal567 Loading…
1
ProTip! What’s not been updated in a month: updated:<2026-02-18.