Skip to content

Pull requests: google/XNNPACK

New pull request New
254 Open 10,061 Closed
254 Open 10,061 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

Clean up reduce kernels
#10632 opened Jul 1, 2026 by copybara-service Bot Loading…
qd8-qb4w-bf16 i8mm
#10631 opened Jul 1, 2026 by GregoryComer Contributor • Draft
Add fp32->bf16 rewrite pass
#10629 opened Jul 1, 2026 by GregoryComer Contributor Loading…
Refactor LiteRT backend numerical tests into a unified suite.
#10628 opened Jul 1, 2026 by copybara-service Bot Loading…
Copybara import of the project:
#10620 opened Jun 29, 2026 by copybara-service Bot Loading…
Include arithmetic.h in dot_base.py
#10605 opened Jun 26, 2026 by copybara-service Bot Loading…
Add support for neon dot qd8_bf16_qb4w
#10602 opened Jun 26, 2026 by JakeStevens Loading…
2
Add fp32 dot kernels for Hexagon
#10598 opened Jun 26, 2026 by copybara-service Bot Loading…
Skip extent=1 loops in consumer nest during fusion.
#10596 opened Jun 25, 2026 by copybara-service Bot Loading…
guard partial-block bias and ksum writes in x64c4 avx256vnni packw
#10589 opened Jun 25, 2026 by aizu-m Contributor Loading…
Pass quantized weights in MLD and fix Gemma 3 inference on WebGPU.
#10577 opened Jun 24, 2026 by copybara-service Bot Loading…
Add ARM f8mm8 dot kernels
#10574 opened Jun 24, 2026 by copybara-service Bot Loading…
Add NEONv8 microkernels for BF16/F16/F32 to QS8/QU8 vcvt.
#10573 opened Jun 24, 2026 by copybara-service Bot Loading…
Cap maximum tensor sizes in YNNPACK reduce subgraph tests
#10562 opened Jun 23, 2026 by copybara-service Bot Loading…
Implement 2-bit packw templates for AVX2, SSSE3, NEONDOT, and Scalar
#10557 opened Jun 23, 2026 by copybara-service Bot Loading…
Fix required dot tiling
#10551 opened Jun 22, 2026 by copybara-service Bot Loading…
Fix qd8_f16_qb4w GEMM config getter always returning NULL on Arm
#10537 opened Jun 19, 2026 by outum1500 Loading…
1
ci: add cmake-windows-arm64-native job using windows-11-arm runner
#10528 opened Jun 18, 2026 by chinazhangchao Loading…
3
Propagate source regions through broadcasts too.
#10503 opened Jun 16, 2026 by copybara-service Bot Loading…
Update polynomial constants for f16-vapproxgelu to improve accuracy.
#10493 opened Jun 16, 2026 by copybara-service Bot Loading…
Fix XNNPACK build configuration target name for Windows on ARM64
#10483 opened Jun 16, 2026 by copybara-service Bot Loading…
Modify subgraph benchmarks to not rely on XNN_FLAG_FORCE_FP16_INFERENCE
#10478 opened Jun 15, 2026 by copybara-service Bot Loading…
bound split axis to input rank in split_dims reshape
#10473 opened Jun 15, 2026 by aizu-m Contributor Loading…
Bounds-check data-derived axes and quantization params in tensor builder
#10464 opened Jun 11, 2026 by sin99xx Contributor Loading…
Fix out-of-bounds read when packing transposed qb4w weights
#10463 opened Jun 11, 2026 by sin99xx Contributor Loading…
ProTip! Updated in the last three days: updated:>2026-06-27.