Skip to content

Pull requests: google/XNNPACK

New pull request New
188 Open 8,410 Closed
188 Open 8,410 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

Remove spurious transposes of non-static GEMM weights.
#8891 opened Sep 16, 2025 by copybara-service bot Loading…
Enable F16 for cpus with F16C instead of AVX2
#8889 opened Sep 15, 2025 by copybara-service bot Loading…
Add the configuration identifier to the weight cache look up key for deconvolution-nhwc.
#8888 opened Sep 15, 2025 by copybara-service bot Loading…
Add the configuration identifier to the weight cache look up key for convolution-nhcw.
#8887 opened Sep 15, 2025 by copybara-service bot Loading…
Add the configuration identifier to the weight cache look up key for convolution-nchw.
#8885 opened Sep 15, 2025 by copybara-service bot Loading…
Add the configuration identifier to the weight cache look up key for batch-matrix-multiply-nc.
#8886 opened Sep 15, 2025 by copybara-service bot Loading…
Remove windows.c and use C11's concurrency support library instead.
#8884 opened Sep 15, 2025 by copybara-service bot Loading…
Remove GCD-based version of pthreadpool as these are handled by C11 atomics and the condvar implementation.
#8883 opened Sep 15, 2025 by copybara-service bot Loading…
Add an interface to initialize XNNPACK with a xnn_scheduler_v2.
#8882 opened Sep 15, 2025 by copybara-service bot Loading…
Clean up the XNN_SLINKY_ENABLED and XNN_SLINKY_AVAILABLE macros to match other similar macros.
#8881 opened Sep 15, 2025 by copybara-service bot Loading…
Added SME1 support for int8 GEMM and IGEMM operations
#8880 opened Sep 15, 2025 by qti-vaiskv Loading…
Automated Code Change
#8877 opened Sep 14, 2025 by copybara-service bot Loading…
Automated Code Change
#8876 opened Sep 14, 2025 by copybara-service bot Loading…
Automated Code Change
#8875 opened Sep 14, 2025 by copybara-service bot Loading…
Use the new fingerprinting method for XNNPack delegate cache files.
#8866 opened Sep 11, 2025 by copybara-service bot Loading…
Use non-temporal stores where possible in weight packing kernels.
#8865 opened Sep 10, 2025 by copybara-service bot Loading…
Add int8xint4 FC XNNPACK kernel for AVXVNNIINT8
#8857 opened Sep 8, 2025 by copybara-service bot Loading…
Add a new fingerprinting method for XNNPack delegate cache files.
#8841 opened Sep 3, 2025 by copybara-service bot Loading…
Add rms-norm op for fp32 and fp16.
#8822 opened Aug 26, 2025 by copybara-service bot Loading…
Fix incorrect declarations
#8812 opened Aug 24, 2025 by cyyever Loading…
4
Added PF32 IGEMM Support with SME1
#8811 opened Aug 24, 2025 by qti-vaiskv Loading…
F32-MAXPOOL for AVX
#8804 opened Aug 19, 2025 by copybara-service bot Loading…
vector implementation of vexp and vtanh functions for RISC-V-based processors
#8740 opened Jul 28, 2025 by kozinove Loading…
4
Clean up 4-bit datatypes:
#8726 opened Jul 17, 2025 by copybara-service bot Loading…
Build bazel-linux-aarch64-sme2-qemu tests when called with run-tests=false, and make more noise when tests actually break.
#8725 opened Jul 17, 2025 by copybara-service bot Loading…
ProTip! Add no:assignee to see everything that’s not assigned.