Skip to content

Pull requests: google/XNNPACK

New pull request New
179 Open 8,152 Closed
179 Open 8,152 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

Automated Code Change
#8616 opened Jun 21, 2025 by copybara-service bot Loading…
Roll back commit d37e958
#8615 opened Jun 21, 2025 by copybara-service bot Loading…
Add xnn_hardware_config::primary_uarch, to replace cpuinfo_get_core(0)->uarch.
#8614 opened Jun 20, 2025 by copybara-service bot Loading…
Remove xnn_internal_set_arm_sme2. We have decided not to enable SME2 by default unless we can test it normally via QEMU, so there is no need for this feature.
#8613 opened Jun 20, 2025 by copybara-service bot Loading…
Hook up the new xnn_qd8_f32_qc4w_gemm_minmax_ukernel_4x8__asm_aarch32_neonmlal_ld64_2 kernel for A53 and A55.
#8611 opened Jun 20, 2025 by copybara-service bot Loading…
Add a hand-written assembly microkernel for qd8_f32_qc4w_gemm_minmax on aarch32 and neonmlal based on the qd8_f32_qc8w_gemm_minmax kernel with mr=4 and nr=8.
#8610 opened Jun 20, 2025 by copybara-service bot Loading…
Added const_weights version of batch matrix multiply for QS8 on operator level.
#8609 opened Jun 20, 2025 by copybara-service bot Loading…
Fix clang-tidy warnings regarding marking single-argument constructors as explicit.
#8607 opened Jun 20, 2025 by copybara-service bot Loading…
Added Int8xInt8 path with Int32 accumulation to XNNPACK to cover the use case from tflite to get rid of other implementations for this path.
#8605 opened Jun 19, 2025 by copybara-service bot Loading…
Update KleidiAI, use new SDOT kernels for QB4W
#8604 opened Jun 19, 2025 by mcr229 Loading…
1
Remove check_config script for testing gemm_config for errors
#8599 opened Jun 17, 2025 by copybara-service bot Loading…
Add convolution-nhwc and deconvolution-nhwc subgraph support for KleidiAI iGEMM kernels for qs8 with packed left-hand operands.
#8597 opened Jun 17, 2025 by copybara-service bot Loading…
Rename xnn_init_*_config to xnn_get_*_config
#8594 opened Jun 16, 2025 by copybara-service bot Loading…
Add convolution-nhwc and deconvolution-nhwc operator support for KleidiAI iGEMM kernels for qs8 with packed left-hand operands.
#8591 opened Jun 16, 2025 by copybara-service bot Loading…
Use SSA rewrite to fix up converted fp16 graphs
#8589 opened Jun 14, 2025 by copybara-service bot Loading…
Avoid "ignoring duplicate libraries" warning in macOS cmake builds.
#8585 opened Jun 13, 2025 by copybara-service bot Loading…
Hexagon TAP presubmit and blueprint
#8580 opened Jun 12, 2025 by copybara-service bot Loading…
Update dependency versions for XNNPACK and KleidiAI.
#8579 opened Jun 12, 2025 by copybara-service bot Loading…
Change from 3x4 to 4x4 for igemm scalar fallback microkernel
#8565 opened Jun 10, 2025 by copybara-service bot Loading…
Add QNX to recognized systems
#8562 opened Jun 10, 2025 by eleir-contrib Loading…
1
Fix batch size of blockwise quantization scales for validation
#8560 opened Jun 10, 2025 by copybara-service bot Loading…
PACKW GOI microkernels add n_stride
#8551 opened Jun 6, 2025 by copybara-service bot Loading…
Add subgraph benchmarks for variants of the fully-connected op.
#8540 opened Jun 4, 2025 by copybara-service bot Loading…
Explicit type arg in std::max for Hexagon build
#8533 opened Jun 3, 2025 by ejparkqc Loading…
Add AVX512 sqrt microkernel
#8530 opened Jun 3, 2025 by copybara-service bot Loading…
ProTip! Type g p on any issue or pull request to go back to the pull request listing page.