/

/

Обзор ДокументацияВойти

/

/

Аналитика

mirror_xnnpack

Зеркало из https://github.com/google/xnnpack.git

Описание

High-efficiency floating-point neural network inference operators for mobile, server, and Web

convolutional-neural-network

convolutional-neural-networks

inference-optimization

matrix-multiplication

mobile-inference

neural-networks

223.28 MiB

Лицензия

Правила участия

В избранном0

Языки

C78,1%
C++14,1%
Assembly4,6%
Shell1,2%
Starlark0,7%
CMake0,6%
Остальные0,7%

Сообщить о нарушении

Коммиты:

Dillon Sharlet

Rewrite x86 uint8 x int2 kernels to use tile_k = 16

25 июл 2026, 23:09

25 июл 2026, 23:096032039

Use cpuinfo support for ARM fp8

9 дней назад

Add 32768 (2^15) to the kDims array in the fully-connected subgraph benchmark.

3 дня назад

Fix Bazel Windows ARM64 MSVC config

месяц назад

build_overrides

[gn] Add extra GN config for MacOS

5 месяцев назад

f16-vsin/vcos use degree 3 polynomial - fix catastrophic cancellation in cosine and vectorize sine - cosine using modulo-pi/2 range reduction

2 дня назад

Rename all non self-contained header files from `.h` to `.inc` to avoid confusing our tooling into trying to parse them in isolation.

год назад

Add patch to allow enable FPMR register access in our qemu build

месяц назад

Add taller fma microkernels

год назад

f16-vsin/vcos use degree 3 polynomial - fix catastrophic cancellation in cosine and vectorize sine - cosine using modulo-pi/2 range reduction

2 дня назад

Expose native FP16 support query in XNNPACK.

16 дней назад

Merge pull request #10772 from destro4evr-rgb:fix/litert-tile-multiples-rank-check

2 дня назад

Remove duplicate generator x4-pack.sh

2 дня назад

Merge pull request #10842 from prasanna8585:fix-batch-matmul-packed-size-overflow

2 дня назад

f16-vsin/vcos use degree 3 polynomial - fix catastrophic cancellation in cosine and vectorize sine - cosine using modulo-pi/2 range reduction

2 дня назад

[gn] Add experimental integration with KleidiAI

5 месяцев назад

[RVV] add rvv f32 kernel for ibilinear

18 дней назад

Rewrite x86 uint8 x int2 kernels to use tile_k = 16

13 часов назад

Remove WORKSPACE

6 месяцев назад

Reformat the whole repo

год назад

[KleidiAI] new SDOT kernels for qb4

год назад

[gn] DEPS update for June, 2026

месяц назад

Add fp32->bf16 rewrite pass

24 дня назад

xx-pad AVX512SKX microkernel with p64_u128

12 дней назад

[gn] Add experimental integration with KleidiAI

5 месяцев назад

f16-vsin/vcos use degree 3 polynomial - fix catastrophic cancellation in cosine and vectorize sine - cosine using modulo-pi/2 range reduction

2 дня назад

CONTRIBUTING.md

Initial open-source release

7 лет назад

Add a target in `BUILD.gn` that allows using the XNNPack export of the Tensor API.

25 дней назад

Initial open-source release

7 лет назад

Update cpuinfo in XNNPACK

9 дней назад

Update cpuinfo, gtest, KleidiAI dependencies

6 месяцев назад

Fix Bazel Windows ARM64 MSVC config

месяц назад

build_params.bzl

Add bf16 i8mm qd8_qb4w GEMM microkernels

24 дня назад

Add bf16-qu8 convert operator support

месяц назад

cpuinfo_windows_arm64_msvc.patch

Fix Bazel Windows ARM64 MSVC config

месяц назад

Fixed wasm_relaxed_simd tests.

2 года назад

generated_file.bzl

Add YNNPACK to XNNPACK

9 месяцев назад

preamble.js.lds

wasmsimd preamble fix for to support timeOrigin

год назад

register_extension_info.bzl

Add YNNPACK to XNNPACK

9 месяцев назад

README.md