Skip to content

Pull requests: ROCm/hip-ep

Pull request creation is restricted
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
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort

Pull requests list

perf(qconv): speed up both paths of the quantized 1x1 convolution
#1074 opened Sep 20, 2026 by AMDmoore Contributor Loading…
Feat/qdq conv not convert to gemm
#1073 opened Sep 20, 2026 by amd-genmingz Collaborator Draft
perf(qgemm): retune the K=32 tile from 64x64 to 128x16
#1071 opened Sep 20, 2026 by donghanbj Contributor Loading…
ci: run the #961 reland together with #599 so CI can exercise Gemma-4
#1069 opened Sep 20, 2026 by BoarQing Contributor Loading…
3 of 8 tasks
perf(allocator): reland large pinned-buffer pooling with the eviction leak fixed
#1068 opened Sep 20, 2026 by BoarQing Contributor Loading…
3 of 4 tasks
kernels: QMoE kernel optimization for baseline and MTP/EAGLE enhancement New feature or request
#1067 opened Sep 20, 2026 by penglu12-quark Contributor Loading…
4 tasks
fix(matmul_nbits): gate LDS barrier on archs that provide s_barrier
#1066 opened Sep 18, 2026 by amd-mah Loading…
3 of 4 tasks
feat(hip-to-tosa): lower OneHot and document NonZero boundary
#1065 opened Sep 18, 2026 by pragyam32 Contributor Loading…
feat(hip-to-tosa): lower Shape/Size/Identity/ConstantOfShape residues to TOSA
#1064 opened Sep 18, 2026 by pragyam32 Contributor Loading…
4 tasks
feat(hip-to-tosa): lower hip.silu and hip.swish by decomposing to TOSA
#1062 opened Sep 18, 2026 by pragyam32 Contributor Loading…
4 tasks
feat(hip-to-tosa): lower hip.grid_sample by decomposing to TOSA
#1059 opened Sep 18, 2026 by pragyam32 Contributor Loading…
4 tasks
Use rocmlirTriton for llvm-project
#1057 opened Sep 18, 2026 by ilango100 Contributor Loading…
feat(hip-to-tosa): lower hip.gelu by decomposing to TOSA
#1054 opened Sep 17, 2026 by pragyam32 Contributor Loading…
4 tasks
feat(hip-to-tosa): lower round and mod to TOSA
#1053 opened Sep 17, 2026 by nvunnam57128 Contributor Loading…
feat(hip-to-tosa): lower hip.softplus by decomposing to TOSA
#1052 opened Sep 17, 2026 by pragyam32 Contributor Loading…
4 tasks done
perf(gqa): optimize D128 sliding-window prefill
#1050 opened Sep 17, 2026 by donghanbj Contributor Loading…
perf(qgemm): optimize W4 prefill kernels
#1051 opened Sep 17, 2026 by donghanbj Contributor Loading…
feat(hip-to-tosa): lower pad and tile to TOSA
#1049 opened Sep 17, 2026 by nvunnam57128 Contributor Loading…
feat(hip-to-tosa): lower hip.pool and hip.global_pool to TOSA
#1048 opened Sep 17, 2026 by tuhinp-amd Contributor Draft
1 of 2 tasks
kernels: int8/dp4a MatMulNBits paths, GQA autotune key bucketing, elementwise flat path large-pr Soft size threshold (>500 LOC or >10 files). Reviewer signal, not a block.
#1043 opened Sep 17, 2026 by penglu12-quark Contributor Loading…
feat(hip-to-tosa): lower reduce_max, reduce_min, reduce_prod and reduce_l2
#1042 opened Sep 17, 2026 by nvunnam57128 Contributor Loading…
4 tasks done
ProTip! What’s not been updated in a month: updated:<2026-08-20.