- 329da64 Refactor MaxPool and ArgMaxPool micro-kernels by Marat Dukhan · 4 years, 11 months ago
- 7ccaab6 IGEMM kernels add asserts for a, c, and w pointers. by Frank Barchard · 4 years, 11 months ago
- 4ad5115 Process 7 input rows / 3 output rows at a time in 5x5p2 chw dwconv. ~10% speedup. by Erich Elsen · 4 years, 11 months ago
- 04f03be Support overriding memory allocation functions by Marat Dukhan · 4 years, 11 months ago
- 7bee751 SSE2 Sigmoid micro-kernels by Marat Dukhan · 4 years, 11 months ago
- 581c1ac CMake targets for f32-sigmoid-test and f32-sigmoid-bench by Marat Dukhan · 4 years, 11 months ago
- 7c6c3b2 Fix CMake build of sigmoid test and benchmark by Marat Dukhan · 4 years, 11 months ago
- 14bec50 Benchmark for F32 Sigmoid micro-kernels by Marat Dukhan · 4 years, 11 months ago
- 80b537a 6x8 IGEMM for Cortex A53 pipelined. by Frank Barchard · 4 years, 11 months ago
- 8fd7b5f Add slightly less accurate sigmoid approximation that is approximately 50% by Erich Elsen · 4 years, 11 months ago
- 80bafd2 SSE2 Sigmoid implementation and evaluation stub by Marat Dukhan · 4 years, 11 months ago
- efc47b8 Unify operator implementations filenames and tests by Marat Dukhan · 4 years, 11 months ago
- 1b09229 Fix issues in PReLU benchmark by Marat Dukhan · 4 years, 11 months ago
- 44e06a6 Fix Sigmoid microkernel incompatibilities with ARM gcc by Marat Dukhan · 4 years, 11 months ago
- c3b9e86 Benchmark Sigmoid operator in TFLite implementation by Marat Dukhan · 4 years, 11 months ago
- 7c8e0c7 4x8 IGEMM for Cortex-A53 pipelined by Frank Barchard · 4 years, 11 months ago
- cb10f26 Build accuracy evaluation benchmarks with CMake by Marat Dukhan · 4 years, 11 months ago
- 684bbb0 CMP 2 instructions earlier in A/C clamping. by Frank Barchard · 4 years, 11 months ago
- 9d501d5 Add new micro-kernels, operators, and unit tests to CMake build by Marat Dukhan · 4 years, 11 months ago
- ca2733c ND elementwise Multiply operator with broadcasting support by Marat Dukhan · 5 years ago
- de9b0fb Process 6 rows at a time in 5x5s2 spchw dwconv kernel. by Erich Elsen · 5 years ago
- 95bebc9 Benchmarks rename sgemm and sppmm to f32_gemm and f32_ppmm by Frank Barchard · 5 years ago
- 179ac85 Add tail to depthwise 5x5s2p2_chw kernels. by Erich Elsen · 5 years ago
- 9efaed7 A53 GEMM and IGEMM pipelined kernels prefetch C in epilogue by Frank Barchard · 5 years ago
- ffd6840 SSE2 exp and expminus evaluation stubs by Marat Dukhan · 5 years ago
- 19418b5 GEMM 4x8 and 4x12 kernels use forward stores for C. by Frank Barchard · 5 years ago
- 82cfe18 4x8 a53 epilogue NOPs in group 5 by Frank Barchard · 5 years ago
- a99918a Support 5x5 Depthwise Convolution in SpNCHW layout by Marat Dukhan · 5 years ago
- 430efbc Merge pull request #153 from Maratyszcza:readme-fix by XNNPACK Team · 5 years ago
- 346a9e5 Sigmoid evaluation stubs, micro-kernels, and operator by Marat Dukhan · 5 years ago
- 21a57c9 Minor grammatical fix to READ by Andrew Tulloch · 5 years ago
- 797a8fe NEONFMA versions of vector expf function by Marat Dukhan · 5 years ago
- c76065f Fix CMake build of elementwise microkernels by Marat Dukhan · 5 years ago
- c07cb7f Binary elementwise microkernels by Marat Dukhan · 5 years ago
- 0ecc2ab 4x8 GEMM for Cortex A53 by Frank Barchard · 5 years ago
- 5abe43c ST1 post increment for Cortex A53 GEMM/IGEMM microkernels by Frank Barchard · 5 years ago
- e097028 Python3 compatibility for xngen by Marat Dukhan · 5 years ago
- 72b3250 Provide compiler hints to generate branch-free code for math helper functions by Marat Dukhan · 5 years ago
- e67b783 ST1 post increment for ld64/ld128 GEMM/IGEMM microkernels by Frank Barchard · 5 years ago
- e334cc6 Revert "Exclude unused parts of xnn_params" by Marat Dukhan · 5 years ago
- 6972249 Resize Bilinear operator by Marat Dukhan · 5 years ago
- 5093cbc Exclude unused parts of xnn_params by Marat Dukhan · 5 years ago
- e64f91a Pipelined 6x8 GEMM for Cortex A53 by Frank Barchard · 5 years ago
- 9fab3f9 Support input offset in BILINEAR micro-kernels by Marat Dukhan · 5 years ago
- 38709a6 Add scalar chw 5x5p2 and 5x5s2p2 kernels by Erich Elsen · 5 years ago
- 2a64a1a Fix incompatibility with ARM gcc by Marat Dukhan · 5 years ago
- 0f06b5c Fix gcc incompatibility in SSE PReLU microkernels by Marat Dukhan · 5 years ago
- c465fc2 Add missed f32-bilinear-test to CMake build by Marat Dukhan · 5 years ago
- 35dacfb BILINEAR micro-kernels by Marat Dukhan · 5 years ago
- 5098c3e Refactor DWCONV micro-kernels by Marat Dukhan · 5 years ago
- 49e6ee9 Refactor VMulCAddC micro-kernel by Marat Dukhan · 5 years ago
- 69c3f2c Refactor PReLU microkernels by Marat Dukhan · 5 years ago
- d5208d6 Remove a_sum buffer by Marat Dukhan · 5 years ago
- 1898b91 Move adjustment_* arguments of Deconvolution into setup by Marat Dukhan · 5 years ago
- a41533d Reduce image sizes in Deconvolution unit tests by Marat Dukhan · 5 years ago
- bad48fe Vary number of threads in the End-to-End benchmark by Marat Dukhan · 5 years ago
- 70ad409 Make ARM microkernels compatible with gcc by Marat Dukhan · 5 years ago
- fb60914 Make F32 CLAMP NEON micro-kernel compatible with gcc on AArch32 by Marat Dukhan · 5 years ago
- bd41971 A57 branch a version of A53 kernel by Frank Barchard · 5 years ago
- 8e6e997 Fix ARM64 build with CMake by Marat Dukhan · 5 years ago
- c9d2f3f Fix CMake build of End-to-End DWCONV & GEMM benchmarks by Marat Dukhan · 5 years ago
- c712fa4 Add Freq to end2end benchmark. by Frank Barchard · 5 years ago
- ef4416e End-to-end benchmarks for DWCONV microkernels by Marat Dukhan · 5 years ago
- e72e287 Add Freq to E2E benchmark by Frank Barchard · 5 years ago
- 0a5a53f Fix CMake build of GEMM E2E benchmark by Marat Dukhan · 5 years ago
- e2142e7 Merge pull request #103 from Maratyszcza:master by XNNPACK Team · 5 years ago
- 63ba2ed Fix typos in AVX2 ExtExp micro-kernels by Marat Dukhan · 5 years ago
- 5f18d26 End-to-end benchmarks for GEMM/IGEMM microkernels by Marat Dukhan · 5 years ago
- 4fa0fbe Add missing <chrono> header in f32-softargmax benchmark by Marat Dukhan · 5 years ago
- 791d01d Minor revision in README by Marat Dukhan · 5 years ago
- a5977df Fix CMake build by Marat Dukhan · 5 years ago
- 64a5bfe A53 6x8 IGEMM kernel prefetch by Frank Barchard · 5 years ago
- 5a9c48d Merge pull request #82 from AshkanAliabadi:master by XNNPACK Team · 5 years ago
- 95b2243 PReLU benchmark on characteristic arguments of ImageNet classifiers by Marat Dukhan · 5 years ago
- bd1d5d9 6x8 A53 GEMM use prefetch. by Frank Barchard · 5 years ago
- b864235 Refactor BUILD.bazel file by Marat Dukhan · 5 years ago
- 8c19e3c Internal change by Marat Dukhan · 5 years ago
- ab7424a Fix typos in Deconvolution operator tests by Marat Dukhan · 5 years ago
- f568f08 Support Convolution, Deconvolution, and Fully Connected operators without bias by Marat Dukhan · 5 years ago
- d70028a Update pthreadpool revision by Marat Dukhan · 5 years ago
- 263bb09 Cortex A76 use 6x8 micro kernel by Frank Barchard · 5 years ago
- feb4923 AVX512F exp implementation based on PERM2 by Marat Dukhan · 5 years ago
- ba7c3bb Merge generate-f32-gemminc.sh script into generate-f32-gemm.sh by Marat Dukhan · 5 years ago
- 918a4a6 Extract common parts of test generators into a separate file by Marat Dukhan · 5 years ago
- 00bf68e A53 6x8 GEMM unrolled by Frank Barchard · 5 years ago
- c452eb1 Re-generate SpMM micro-kernels by Marat Dukhan · 5 years ago
- ae777b4 4x8 a53 eliminate pushes to stack by Frank Barchard · 5 years ago
- e0601b5 Sort include order for params-init.h and log.h by Frank Barchard · 5 years ago
- 4a2bbc6 Benchmark for Two-Pass Softmax algorithm by Marat Dukhan · 5 years ago
- eeaa7bd Refactor initialization of micro-kernel parameters by Marat Dukhan · 5 years ago
- 6f8d4d3 RADDEXTEXP and VSCALEEXTEXP micro-kernels for AVX2 and AVX512F by Marat Dukhan · 5 years ago
- b3c6c6e 6x8 A53 remove pushes for NEON by Frank Barchard · 5 years ago
- 46fb807 4x8 A53 GEMM, and GEMMINC unpipelined microkernels. by Frank Barchard · 5 years ago
- cd945c6 Re-enable swizzle GEMM/IGEMM micro-kernels in WAsm SIMD on ARM by Marat Dukhan · 5 years ago
- f753a7d Rename BUILD to BUILD.bazel by Marat Dukhan · 5 years ago
- c4ae7de Propagate IGEMM SR argument to weights packing in Deconvolution operator by Marat Dukhan · 5 years ago
- c6afd9b Add blocked scalar spmm kernels. by Erich Elsen · 5 years ago
- 7892d97 Fix softargmax-bench CMake target. by Ashkan Aliabadi · 5 years ago
- 8440fde Support TF-style SAME padding via explicit flag by Marat Dukhan · 5 years ago
- bff791e Use 8x1 SpMM micro-kernel on WebAssembly by Marat Dukhan · 5 years ago