Skip to content

Pull requests: ROCm/FlyDSL

New pull request New
48 Open 618 Closed
48 Open 618 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

[layernorm] Add backward pass for training (PR 3/3, #769)
#801 opened Jul 3, 2026 by jhinpan Contributor Loading…
2
[rmsnorm] Add fused-add / residual backward (PR 2/3, #769)
#800 opened Jul 3, 2026 by jhinpan Contributor Loading…
2
[Enh] Support chained comparison
#799 opened Jul 3, 2026 by sjfeng1999 Collaborator Loading…
1 task
4
4wave fp4 gemm opt
#798 opened Jul 3, 2026 by benenzhu Contributor Loading…
3 tasks done
2
[Fix] Consistent promotion rules for Vector and Numeric
#797 opened Jul 3, 2026 by sjfeng1999 Collaborator Loading…
1 task
4
[rmsnorm] Add backward pass + forward store_rstd for training (PR 1/3, #769)
#795 opened Jul 3, 2026 by jhinpan Contributor Loading…
2
[Kernel] conv3d: 8-wave double-buffered FP8 implicit-GEMM kernel(gfx950)
#794 opened Jul 1, 2026 by jiacao-amd Loading…
2 tasks
8
[Perf] preshuffle_gemm: vectorize scale_a epilogue load (4x32b -> 1x128b)
#791 opened Jul 1, 2026 by coderfeli Collaborator Loading…
[Enh] Add finer FastMathFlag control
#789 opened Jul 1, 2026 by sjfeng1999 Collaborator Loading…
1 task
5
[4/5] autotune: CI guard + committed-config regression check (#770)
#788 opened Jul 1, 2026 by jhinpan Contributor Loading…
5
[3/5] autotune: offline config emit + runtime lookup (#770)
#786 opened Jul 1, 2026 by jhinpan Contributor Loading…
5
[2/5] autotune: two-track config + first real adopter (rmsnorm) (#770)
#785 opened Jul 1, 2026 by jhinpan Contributor Loading…
5
[1/5] autotune: harden cache key + add restore_value (#770)
#783 opened Jul 1, 2026 by jhinpan Contributor Loading…
3
[Kernel] Add compile_mxfp6_gemm: MXFP6×MXFP4 preshuffle GEMM (gfx950)
#780 opened Jun 30, 2026 by amd-satre Contributor Loading…
1 task done
5
[Feat]: Add dispatch & combine tuning configuration multi-gpu
#772 opened Jun 30, 2026 by yanboshao Contributor Loading…
1 task
flash_attn_generic: replace raw arith.* FP ops with FlyDSL-typed fast…
#764 opened Jun 29, 2026 by xudoyuan Collaborator Loading…
1 task
1
3
[repro] flash_attn_generic: concise causal-mask loop trips dynamic-if…
#763 opened Jun 29, 2026 by xudoyuan Collaborator • Draft
1 task
FlyDSL gemm_decode: small-M dense GEMM kernels (BF16/FP8/blockscale)
#757 opened Jun 26, 2026 by vedenev-amd • Draft
1 task
@vedenev-amd
feat(moe): layout-API MXFP4 (a4w4/a8w4) MoE gemm
#753 opened Jun 26, 2026 by coderfeli Collaborator Loading…
runtime: hard-fail mgpuModuleLoadJIT on HIP
#748 opened Jun 25, 2026 by fallintoplace Loading…
fix tile syntax in divide
#746 opened Jun 25, 2026 by tingqli Contributor Loading…
1 task
3
[Fix] Set identity block scales for CDNA4 MFMA_Scale in fp8 row-scale…
#744 opened Jun 25, 2026 by amd-songpiao Loading…
1
MXFP4 MoE tuning harness: legality filter, measurement, ledger + validated baseline (#708)
#736 opened Jun 25, 2026 by jhinpan Contributor • Draft
1
[Kernel] conv2d: 8-wave double-buffered implicit-GEMM BF16 kernel(gfx950)
#733 opened Jun 24, 2026 by jiacao-amd Loading…
1 task
7
[Kernel] feat: Add MXFP6-E2M3 activation support to mixed_moe_gemm_2stage
#709 opened Jun 19, 2026 by amd-satre Contributor Loading…
1 task done
1
ProTip! Adding no:label will show everything without a label.