Skip to content

Pull requests: NVIDIA/cudnn-frontend

New pull request New
51 Open 434 Closed
51 Open 434 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

Fix bias GMEM pointer alignment in the Rubin grouped GEMM GLU kernel cat-bugfix mod-cutedsl CuTeDSL kernels, generated kernels, examples, or related integration work. orig-nv-eng Reported or requested by NVIDIA engineering.
#698 opened Aug 21, 2026 by Anerudhan Collaborator Loading…
1
Fix sm_107 SMEM overflow in Rubin grouped GEMM stage sizing cat-bugfix mod-cutedsl CuTeDSL kernels, generated kernels, examples, or related integration work. orig-nv-eng Reported or requested by NVIDIA engineering.
#696 opened Aug 21, 2026 by Anerudhan Collaborator Loading… Frontend 1.28.0
@Anerudhan
1
Add public GELU-MLP and a ModelOpt-anchored Qwen-Image NVFP4 proxy cat-feature Requests for new functionality, APIs, examples, or behavior improvements. mod-frontend cuDNN frontend APIs, operation graph construction, plans, and user-facing wrappers. mod-infra Infrastructure, CI/CD, build systems, packaging, releases, or repo maintenance. orig-nv-eng Reported or requested by NVIDIA engineering.
#695 opened Aug 21, 2026 by YangXu1990uiuc Collaborator Loading…
3 tasks done
4
GGEMM+GLU+RHT+quant kernel now outputs column-wise RHT in ragged tensor layout
#693 opened Aug 21, 2026 by timmoon10 Member Loading…
1 of 3 tasks
3
[SDPA] fwd heuristics: complete knob-set recommendations (propose/place) + graph-reachable split-KV cat-feature Requests for new functionality, APIs, examples, or behavior improvements. mod-cutedsl CuTeDSL kernels, generated kernels, examples, or related integration work. mod-frost orig-nv-eng Reported or requested by NVIDIA engineering.
#692 opened Aug 21, 2026 by vedaanta Collaborator Loading…
17
Improve performance of SM100 DSA backward kernel
#684 opened Aug 20, 2026 by Butterfingrz Loading…
2 of 3 tasks
2
benchmark: add block_sparse_attention (BSA fwd+bwd vs FA4)
#674 opened Aug 19, 2026 by vedaanta Collaborator Loading… Frontend 1.29.0
4
Add SM100 H16 DSA backward specialization cat-enhancements mod-cutedsl CuTeDSL kernels, generated kernels, examples, or related integration work. orig-external Reported or requested by an external user, customer, or community contributor.
#664 opened Aug 19, 2026 by icavan Loading…
2 of 3 tasks
Frontend 1.28.0
@icavan
1
frost(sdpa): port #585 LPT schedulers to the SM107 fp8 sibling; defaults must obey declared knob domains cat-bug Reports of incorrect behavior, crashes, regressions, or unexpected results. mod-cutedsl CuTeDSL kernels, generated kernels, examples, or related integration work. mod-frost orig-nv-eng Reported or requested by NVIDIA engineering.
#656 opened Aug 19, 2026 by vedaanta Collaborator Loading…
1
1
sdpa fp8: softmax_precision knob — f16x2 exponent path on the SM107 sibling cat-feature Requests for new functionality, APIs, examples, or behavior improvements. mod-cutedsl CuTeDSL kernels, generated kernels, examples, or related integration work. mod-frost orig-nv-eng Reported or requested by NVIDIA engineering.
#651 opened Aug 18, 2026 by vedaanta Collaborator Loading…
4
Add Python API for cuDNN GNN simple aggregation cat-feature Requests for new functionality, APIs, examples, or behavior improvements. mod-backend cuDNN backend API, graph execution, descriptors, engines, or backend integration. orig-nv-eng Reported or requested by NVIDIA engineering.
#647 opened Aug 18, 2026 by tingyu66 Member Loading…
3 tasks done
Frontend 1.28.0
29
test(mhas_v2): NaN-poison ragged capacity tails so capacity-vs-live-token bugs fail deterministically (#624)
#646 opened Aug 18, 2026 by vedaanta Collaborator Loading… Frontend 1.28.0
6
dsa(indexer_backward): opt-in SM100 sparse backward v2 — 1.16-1.92x faster (fp32-accurate d_index_k at no extra cost)
#640 opened Aug 18, 2026 by zkyue Contributor Loading…
9
test(python): stop segfaults and CUDA IMAs from poisoning the rest of the run cat-ci CI failures, test flakiness, workflow breakage, or automation issues. mod-infra Infrastructure, CI/CD, build systems, packaging, releases, or repo maintenance. orig-nv-eng Reported or requested by NVIDIA engineering.
#639 opened Aug 17, 2026 by hwanseoc Member • Draft
4
sdpa: throw cudnnGraphNotSupportedException when no implementation supports the attributes
#633 opened Aug 17, 2026 by vedaanta Collaborator Loading…
1
test: directed SDPA bwd test for rows with very negative LSE
#628 opened Aug 17, 2026 by vedaanta Collaborator Loading…
3
reduce cuteDSL grouped gemm CPU latency cat-enhancements mod-cutedsl CuTeDSL kernels, generated kernels, examples, or related integration work. mod-frontend cuDNN frontend APIs, operation graph construction, plans, and user-facing wrappers. orig-nv-eng Reported or requested by NVIDIA engineering.
#627 opened Aug 17, 2026 by hwanseoc Member Loading…
5 tasks done
Frontend 1.28.0
15
frost(sdpa): order every execute-path tensor prep on the launch stream (Rule 5) cat-bugfix mod-cutedsl CuTeDSL kernels, generated kernels, examples, or related integration work. mod-frost orig-nv-eng Reported or requested by NVIDIA engineering.
#614 opened Aug 16, 2026 by vedaanta Collaborator Loading…
3 tasks done
6
NSA compression: suppress aliased O1 store for static short queries
#599 opened Aug 14, 2026 by mgoldfarb-nvidia • Draft
2 of 3 tasks
1
Memoize the CuTeDSL hot-path dtype/stride conversions (~20% off per-call host)
#589 opened Aug 13, 2026 by YangXu1990uiuc Collaborator Loading…
2 of 3 tasks
1
sdpa fp8: d<=128 envelope for per-tensor FP8 cat-feature Requests for new functionality, APIs, examples, or behavior improvements. mod-cutedsl CuTeDSL kernels, generated kernels, examples, or related integration work. mod-frost orig-nv-eng Reported or requested by NVIDIA engineering.
#587 opened Aug 13, 2026 by vedaanta Collaborator Loading…
22
sdpa fp8 sm100: fused LDTM row-max on cc10.3 (per-tensor FP8) cat-enhancements mod-cutedsl CuTeDSL kernels, generated kernels, examples, or related integration work. mod-frost orig-nv-eng Reported or requested by NVIDIA engineering.
#586 opened Aug 13, 2026 by vedaanta Collaborator Loading…
4
Improve large tensor fuzzer plan fallback and failure output
#583 opened Aug 13, 2026 by msalasooNV Collaborator Loading…
2 of 3 tasks
@Anerudhan @chmayer-gpu
3
Emit the FROST gemm parameter tables, and a demo that launches from them
#582 opened Aug 13, 2026 by YangXu1990uiuc Collaborator • Draft
2 of 3 tasks
8
frost(sdpa): dense-padded cu_seq_len support (kernel CU read mode) cat-feature Requests for new functionality, APIs, examples, or behavior improvements. mod-cutedsl CuTeDSL kernels, generated kernels, examples, or related integration work. mod-frost orig-nv-eng Reported or requested by NVIDIA engineering.
#578 opened Aug 13, 2026 by vedaanta Collaborator Loading…
3 tasks done
1
3
ProTip! Mix and match filters to narrow down what you’re looking for.