Skip to content

Pull requests: uxlfoundation/oneDNN

New pull request New
122 Open 3,397 Closed
122 Open 3,397 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

[backport] all: support for optional upper bound arg for grouped matmul backport component:api Codeowner: @oneapi-src/onednn-arch component:common component:tests Codeowner: @oneapi-src/onednn-arch documentation A request to change/fix/improve the documentation. Codeowner: @oneapi-src/onednn-doc platform:gpu-intel Codeowner: @oneapi-src/onednn-gpu-intel
#4965 opened Apr 7, 2026 by mzhukova • Draft
1
ci, build, docs: aarch64: bump ComputeLibrary version to 52.7.0 component:build devops Github automation documentation A request to change/fix/improve the documentation. Codeowner: @oneapi-src/onednn-doc platform:cpu-aarch64 Codeowner: @oneapi-src/onednn-cpu-aarch64
#4963 opened Apr 7, 2026 by Sqvid Loading…
2 tasks done
benchdnn: inputs: graph: add cases for gated mlp with gelu activation component:tests Codeowner: @oneapi-src/onednn-arch
#4962 opened Apr 7, 2026 by TaoLv Loading…
1
graph: sdpa: support dropout seed/offset/prob in fused sdpa component:graph-api Codeowner: @oneapi-src/onednn-graph
#4961 opened Apr 7, 2026 by TaoLv Loading…
5
[GPU][NVL-P] Use upconversion for unsupported scales on NVL-P platform:gpu-intel Codeowner: @oneapi-src/onednn-gpu-intel
#4960 opened Apr 6, 2026 by kealan-barbieri Loading…
3 of 4 tasks
2
ze api: add support for persistent cache component:common component:tests Codeowner: @oneapi-src/onednn-arch platform:gpu-intel Codeowner: @oneapi-src/onednn-gpu-intel
#4959 opened Apr 6, 2026 by dzarukin Loading…
[GPU] fixup matmul ref implementation component:tests Codeowner: @oneapi-src/onednn-arch platform:gpu-intel Codeowner: @oneapi-src/onednn-gpu-intel
#4958 opened Apr 6, 2026 by dyoussif Loading…
4
benchdnn: cold cache improvement: attempt 2 component:graph-api Codeowner: @oneapi-src/onednn-graph component:tests Codeowner: @oneapi-src/onednn-arch
#4957 opened Apr 6, 2026 by dzarukin Loading…
1
[WIP][Do not review] common: memory: remove UB for tensor dimension checks component:common component:tests Codeowner: @oneapi-src/onednn-arch
#4956 opened Apr 6, 2026 by avmanerikar • Draft
cpu: rv64: rebalance weekly CI test partitions devops Github automation
#4955 opened Apr 6, 2026 by xiazhuozhao Loading…
Copy of Upconvert fp8 weights to xf16 in Matmul in case of xf16 activations component:common component:tests Codeowner: @oneapi-src/onednn-arch platform:cpu-x64 Intel64/AMD64 processors. Codeowner: @oneapi-src/onednn-cpu-x64
#4954 opened Apr 5, 2026 by asimonov1 • Draft
cpu: rv64: enable f16 graph coverage in benchdnn component:common component:graph-api Codeowner: @oneapi-src/onednn-graph component:tests Codeowner: @oneapi-src/onednn-arch
#4953 opened Apr 4, 2026 by xiazhuozhao • Draft
benchdnn: add GatedMLP driver component:tests Codeowner: @oneapi-src/onednn-arch documentation A request to change/fix/improve the documentation. Codeowner: @oneapi-src/onednn-doc
#4951 opened Apr 3, 2026 by kwieloch-intel • Draft
4 of 6 tasks
benchdnn: prohibit --mode=CP component:tests Codeowner: @oneapi-src/onednn-arch
#4950 opened Apr 3, 2026 by kwieloch-intel • Draft
3
xe: ggemm: Add support for Gen12 GPUs platform:gpu-intel Codeowner: @oneapi-src/onednn-gpu-intel
#4949 opened Apr 3, 2026 by umar456 Loading…
4
benchdnn: allow gpu_smoke files for TEST_SET=SMOKE component:tests Codeowner: @oneapi-src/onednn-arch documentation A request to change/fix/improve the documentation. Codeowner: @oneapi-src/onednn-doc
#4948 opened Apr 2, 2026 by dzarukin Loading…
3
all: support for optional upper bound arg for grouped matmul component:api Codeowner: @oneapi-src/onednn-arch component:common component:tests Codeowner: @oneapi-src/onednn-arch documentation A request to change/fix/improve the documentation. Codeowner: @oneapi-src/onednn-doc platform:gpu-intel Codeowner: @oneapi-src/onednn-gpu-intel
#4946 opened Apr 2, 2026 by mzhukova Loading…
27
cpu: aarch64: enable BWD_D for jit:sve_128 and jit_1x1:sve_128 conv component:common platform:cpu-aarch64 Codeowner: @oneapi-src/onednn-cpu-aarch64
#4944 opened Apr 2, 2026 by Anndrey24 Loading…
4
cpu: x64: matmul: enable brg for fp32 on avx2 for all non-gemv cases backport platform:cpu-x64 Intel64/AMD64 processors. Codeowner: @oneapi-src/onednn-cpu-x64
#4943 opened Apr 2, 2026 by azhai219 Loading…
10 tasks
1
common: TLS freelist for memory descriptors to speed up PyTorch CPU inference component:common documentation A request to change/fix/improve the documentation. Codeowner: @oneapi-src/onednn-doc
#4939 opened Apr 1, 2026 by lslusarczyk Loading…
5
tests: benchdnn: fix binary post-op strides printing component:tests Codeowner: @oneapi-src/onednn-arch
#4934 opened Apr 1, 2026 by asztukow Loading…
rfcs: graph: add deterministic mode support documentation A request to change/fix/improve the documentation. Codeowner: @oneapi-src/onednn-doc RFC A design document
#4927 opened Mar 31, 2026 by TaoLv Loading…
[wip, do not review] component:api Codeowner: @oneapi-src/onednn-arch component:common component:tests Codeowner: @oneapi-src/onednn-arch platform:gpu-intel Codeowner: @oneapi-src/onednn-gpu-intel
#4925 opened Mar 31, 2026 by mzhukova • Draft
1
cpu: x64: matmul: add MXFP4 weights support component:common component:tests Codeowner: @oneapi-src/onednn-arch platform:cpu-x64 Intel64/AMD64 processors. Codeowner: @oneapi-src/onednn-cpu-x64
#4922 opened Mar 30, 2026 by amakarev Loading…
35
cpu: x64: pool: enable different dt on src and dst in jit platform:cpu-x64 Intel64/AMD64 processors. Codeowner: @oneapi-src/onednn-cpu-x64
#4914 opened Mar 27, 2026 by asztukow Loading…
ProTip! no:milestone will show everything without a milestone.