Skip to content

Pull requests: ROCm/aiter

New pull request New
177 Open 1,885 Closed
177 Open 1,885 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

optimize the fa fwd and bwd kernels
#2265 opened Mar 12, 2026 by scxiao Loading…
1 task done
1
[CI] Flash Attention CK CI
#2264 opened Mar 12, 2026 by micmelesse • Draft
1 task
1
Introduce asm fmoe kernels that do not require bf16->fp8 quantization
#2262 opened Mar 12, 2026 by JohnNikolay84 Loading…
1 task
@JohnNikolay84
1
tune ep8 fp4 dsv3 fmoe
#2261 opened Mar 12, 2026 by inkcherry • Draft
1 task
3
[TRITON] Update config of MHA PE forward kernel on gfx950 ci:atom ci:sglang ci:vllm enhancement New feature or request triton
#2260 opened Mar 12, 2026 by brunomazzottiamd Loading…
1 task done
@brunomazzottiamd
2
Add performance parity tests for AITER kernels
#2258 opened Mar 12, 2026 by ChuanLi1101 Loading…
4 tasks
1
[FLYDSL] Add GDR decode kernels
#2257 opened Mar 12, 2026 by xytpai Loading…
1
refactor: use ctypes binding
#2255 opened Mar 12, 2026 by amd-ruitang3 Loading…
1 task
7
Capture git errors in case git doesn't exist
#2254 opened Mar 11, 2026 by t2h-fb Loading…
fix fmoe_int8_g1u0_a16
#2251 opened Mar 11, 2026 by junhaha666 Loading…
1 task
4
bwd v3 remove ck dependency
#2250 opened Mar 11, 2026 by JaxChen29 Loading…
1 task
[MXFP4] Patch fp4_utils.py rounding logic following #975
#2249 opened Mar 11, 2026 by Knarf04 Loading…
1
2
[Kernel] Add 2-stage ASM MoE INT8 kernels
#2248 opened Mar 11, 2026 by zufayu Loading…
7
Added HSA header flag passthrough to compile.py
#2245 opened Mar 10, 2026 by Micky774 Loading…
1 task done
5
[Fix] result after topK becomes 0 caused by error propagation to topK
#2244 opened Mar 10, 2026 by kensclin Loading…
1 task
2
[gfx1201] Add tuned kernel configs and FP8 attention support for AMD RDNA4 (Radeon AI PRO R9700)
#2242 opened Mar 10, 2026 by vllmellm • Draft
1 task
[COMMS] Fused allreduce, residual add, rms, quant, gemm
#2238 opened Mar 10, 2026 by micmelesse Loading…
1 task
19
[WIP] Enable fp8 attention for triton unified attention
#2235 opened Mar 10, 2026 by weitliao • Draft
3 tasks
3
Enable accuracy test for all models and add threshold checking ci:atom
#2234 opened Mar 10, 2026 by gyohuangxin Loading…
4 tasks
4
[TRITON] Moe a8w4 gluon gfx1250
#2228 opened Mar 9, 2026 by lburzawa Loading…
1 task done
2
Add Triton fallback for fused_rope_rms (QKNorm+RoPE)
#2227 opened Mar 9, 2026 by sunway513 Loading…
2 tasks
6
Fix error checking in aiter_hip_common.h
#2225 opened Mar 9, 2026 by draganmladjenovic Loading…
1 task
1
Add Quantype.per_1x128 support to fused MoE testing
#2224 opened Mar 9, 2026 by jakki-amd Loading…
1 task done
1
[TRITON] Improve RDNA config selection for unified attention
#2223 opened Mar 9, 2026 by skysnow2001 Loading…
1 task
2
[CI] Flash Attention RDNA CI
#2222 opened Mar 9, 2026 by micmelesse Loading…
1 task done
2
ProTip! Type g i on any issue or pull request to go back to the issue listing page.