Skip to content

Pull requests: ROCm/FlyDSL

New pull request New
68 Open 753 Closed
68 Open 753 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

[WIP][Fix] Fix lld invocation failed when ROCm is not at the baked-in path
#987 opened Aug 8, 2026 by jli-melchior Collaborator Loading…
1 task
[Kernel] Fix single-accumulator RDNA3 GEMM tiles
#986 opened Aug 8, 2026 by skyguan92 Contributor Loading…
fix(pa): select the arch-native fp8 dtype in the metadata grid tuner
#984 opened Aug 7, 2026 by zjin-lcf Loading…
docs: Update docs style to fit style guide
#982 opened Aug 7, 2026 by peterjunpark • 2/2 Loading…
1 task
4
docs: Update configs for documentation to host on Read the Docs
#981 opened Aug 7, 2026 by peterjunpark • 1/2 Loading…
1 task
8
[Perf] Choose the RDNA3 GEMM tile from the shape
#980 opened Aug 7, 2026 by vlluvia Contributor Loading…
1
[Feat] Add an experimental cuda nvvm backend
#976 opened Aug 6, 2026 by sjfeng1999 Collaborator Loading…
1 task
6
Replace legacy HGEMM kernels with gfx950 A16W16 universe GEMM
#975 opened Aug 6, 2026 by xytpai Contributor • Draft
add A4W4 and FP8 P2P transport support to MegaMoE v2 multi-gpu
#972 opened Aug 6, 2026 by Yaowu-Xiong Contributor Loading…
1 task
[gfx1250] Add 256x256 compute-bound GEMM (WIP)
#971 opened Aug 6, 2026 by aoli26 Contributor • Draft
1 task done
[Kernel][MI350] Add bias, alibi bias and sink to flash attention
#960 opened Aug 3, 2026 by amd-nprotaso Contributor Loading…
Add SVD quant for Quark
#957 opened Aug 2, 2026 by amd-xiaoyu12 • Draft
1 task
[MoE] Port moe_gemm_2stage (stage1+stage2) to the new fx.* pipeline — fp8 (gfx942 + gfx950)
#947 opened Aug 1, 2026 by coderfeli Collaborator Loading…
6
[LLVM][FlyDSL] Bump up LLVM and Adapt FlyDSL to upstream LLVM/MLIR API changes
#945 opened Jul 31, 2026 by jli-melchior Collaborator Loading…
1 task
5
[Kernel] Add submanifold sparse 3D convolution (bf16 implicit GEMM, gfx950)
#942 opened Jul 31, 2026 by jiacao-amd Contributor • Draft
Add flex_attention (score_mod / mask_mod) on the generic flash-attention kernel
#931 opened Jul 30, 2026 by RichardChamberlain1 • Draft
3 of 4 tasks
1
DSL-ify raw arith float ops in kernels (flash/mla/pa/moe), fastmath v… multi-gpu
#930 opened Jul 30, 2026 by xudoyuan Collaborator Loading…
1 task
1
3
[Tool] Add per-kernel ISA resource delta table
#928 opened Jul 30, 2026 by Phil-amd Member Loading…
3 of 4 tasks
@Phil-amd
Unify benchmark timing contracts and add calibrated CI gates
#924 opened Jul 30, 2026 by jhinpan Collaborator Loading…
6 tasks done
3
[DSL] Preserve logical signedness of unsigned integer dtypes
#920 opened Jul 28, 2026 by Arist12 Contributor Loading…
6 tasks done
1
@sjfeng1999
[Bugfix][Dialect] Reject vector operands in atomic copy atoms
#918 opened Jul 28, 2026 by AiyyappanMR Loading…
1 task done
[Dialect][Perf] Don't merge mixed static/runtime offsets on LDS pointers
#914 opened Jul 27, 2026 by Arist12 Contributor Loading…
6 tasks done
1
1
Fix hierarchical reduced predicates in copy layout lowering
#912 opened Jul 27, 2026 by HydraQYH Loading…
1 task
@sjfeng1999
1
[Kernel][Perf] Optimize paged-attention metadata decode
#910 opened Jul 27, 2026 by fsx950223 Contributor Loading…
5 tasks done
3
Add fast_divmod magic-number division helper
#906 opened Jul 26, 2026 by kashif Contributor Loading…
@sjfeng1999
4
ProTip! Updated in the last three days: updated:>2026-08-05.