- 16b52e3 Remove Channel Pad NC operator by Marat Dukhan · 4 years, 3 months ago
- a5adc1f Remove WAsm versions of SQRDIFF microkernels by Marat Dukhan · 4 years, 3 months ago
- 13bafb0 SQRDIFF (Squared Difference) microkernels by Marat Dukhan · 4 years, 3 months ago
- 5a599a6 FP16 DWCONV microkernel by Frank Barchard · 4 years, 3 months ago
- 8fb51b6 Remove Add NC operator by Marat Dukhan · 4 years, 3 months ago
- 4e21b27 Copy NC operator by Marat Dukhan · 4 years, 3 months ago
- c3065f5 Unify implementation of unary elementwise operators (Clamp, HardSwish, Sigmoid) by Marat Dukhan · 4 years, 3 months ago
- 0bb49a7 FP16 GAVGPOOL microkernel by Frank Barchard · 4 years, 3 months ago
- 5a409d8 Generalize Clamp NC setup implementation across datatypes by Marat Dukhan · 4 years, 3 months ago
- 210e35d Fix Release build with CMake by Marat Dukhan · 4 years, 3 months ago
- 3b59de2 Refactor logging messages in operators and subgraph nodes by Marat Dukhan · 4 years, 3 months ago
- febfe59 Cast void * parameters to fp16__ pointers by Frank Barchard · 4 years, 3 months ago
- 95e8b7a Split Subgraph API functions into separate files by Marat Dukhan · 4 years, 3 months ago
- 2a1049c FP16 VMULCADDC microkernel by Frank Barchard · 4 years, 3 months ago
- b2217dd Disable tsan for micro-kernels which read out-of-bounds by Marat Dukhan · 4 years, 3 months ago
- f3d1205 Fuse Constant Pad into [Depthwise] Convolution 2D by Marat Dukhan · 4 years, 3 months ago
- 33fcf78 Add test_mode targets with stricter checks when running unit tests by Marat Dukhan · 4 years, 3 months ago
- 1f19872 Node fusion in Subgraph API by Marat Dukhan · 4 years, 3 months ago
- 467f636 Fused [I]GEMM+RELU micro-kernels by Marat Dukhan · 4 years, 3 months ago
- 63523d4 Refactor X32 PAD micro-kernels by Marat Dukhan · 4 years, 3 months ago
- f5c4625 Remove redundant TF legacy + align corners mode in Resize Bilinear by Marat Dukhan · 4 years, 3 months ago
- 065b11e Rename Pad operator to Constant Pad in Operator API by Marat Dukhan · 4 years, 3 months ago
- 54cf510 Minor refactoring in internal subgraph objects by Marat Dukhan · 4 years, 3 months ago
- ab2946c Support Static Constant Pad operator in Subgraph API by Marat Dukhan · 4 years, 3 months ago
- 6e15532 Fix MSVC build of the Pad operator by Marat Dukhan · 4 years, 3 months ago
- 4662b19 N-dimensional Pad operator by Marat Dukhan · 4 years, 3 months ago
- 3bb3bfc X32 FILL micro-kernels by Marat Dukhan · 4 years, 3 months ago
- 56b10cd CONV-HWC NEONFMA micro-kernels with width tile 1 by Marat Dukhan · 4 years, 4 months ago
- ce7a3f8 Auto-generate NEON CONV micro-kernels by Marat Dukhan · 4 years, 4 months ago
- 32f9381 4x4 LD64 GEMM microkernel in AArch32+VFP assembly by Marat Dukhan · 4 years, 4 months ago
- d536072 DWCONV add input_offset and zero parameters by Frank Barchard · 4 years, 4 months ago
- 3b98f6b 4x4 LD64 GEMM+MINMAX microkernel in AArch32+VFP assembly by Marat Dukhan · 4 years, 4 months ago
- 20550ef Optimize output row variables in CONV-HWC microkernels by Marat Dukhan · 4 years, 4 months ago
- 3217d31 Simplify remainder processing in 3x3s2 CONV NEONFMA microkernels by Marat Dukhan · 4 years, 4 months ago
- ae89be6 Refactor common macros by Marat Dukhan · 4 years, 4 months ago
- b112088 Use MinNum/MaxNum instructions where available by Marat Dukhan · 4 years, 4 months ago
- 1f29b80 Refactor CHW micro-kernels by Marat Dukhan · 4 years, 4 months ago
- 7e4ca40 3x3s2c3 CONV NEONFMA microkernel with 0+1 padding by Marat Dukhan · 4 years, 4 months ago
- b9a9958 Simplify q8 dwconv to remove unrolled aarch64 code by Frank Barchard · 4 years, 4 months ago
- 8bd4cba Fix ARM/gcc build or ROUNDZ evaluation stubs by Marat Dukhan · 4 years, 4 months ago
- ffbf96a Benchmark for ROUND evaluation stubs by Marat Dukhan · 4 years, 4 months ago
- b389f35 Rename pooling_2d.input_padding_* into pooling_2d.padding_* by Marat Dukhan · 4 years, 4 months ago
- 6ddfc60 Implement "Greedy by size planner" memory optimization by Chao Mei · 4 years, 4 months ago
- c9852ba ROUNDU and ROUNDD evaluation stubs by Marat Dukhan · 4 years, 4 months ago
- 737a3a1 IGEMM 4x8 a53/a55 - Replace STP x19, x19 with STR by Frank Barchard · 4 years, 4 months ago
- 942b576 Increase size of explicit zero vector in NCHW Convolution by Marat Dukhan · 4 years, 4 months ago
- b339045 Comment change rename clamp params to params by Frank Barchard · 4 years, 4 months ago
- 1fd91c8 Further optimize NEON ROUND Add+Sub-based evaluation stubs by Marat Dukhan · 4 years, 4 months ago
- 03723f5 Further optimize NEON & SSE2 conversion-based rounding stubs by Marat Dukhan · 4 years, 4 months ago
- 2dbb944 ROUNDZ evaluation stubs by Marat Dukhan · 4 years, 4 months ago
- f1d2da4 Fix typo in xnn_math_f32_roundne__scalar description by Marat Dukhan · 4 years, 4 months ago
- 075088a Refactor function names for ROUNDNE evaluation stubs by Marat Dukhan · 4 years, 4 months ago
- b196659 FP16 hswish, clamp and prelu microkernels by Frank Barchard · 4 years, 4 months ago
- af846e0 FP16 vbinary microkernels use a_ptr, b_ptr, y_ptr void * by Frank Barchard · 4 years, 4 months ago
- 1019dcd Force SNaN->QNaN conversion in PSIMD ROUND-NE evaluation stub by Marat Dukhan · 4 years, 4 months ago
- 3e19790 Force SNaN->QNaN conversion in scalar ROUND-NE evaluation stub by Marat Dukhan · 4 years, 4 months ago
- 8d50380 Force SNaN->QNaN conversion in NEON ROUND-NE evaluation stub by Marat Dukhan · 4 years, 4 months ago
- c3d8fb3 Force SNaN->QNaN conversion in SSE ROUND-NE evaluation stub by Marat Dukhan · 4 years, 4 months ago
- bcdb1c1 Remove xnn_q8_dwconv_minmax_ukernel_up8x9__aarch32_neon by Frank Barchard · 4 years, 4 months ago
- eda9c11 Update CHW DWCONV to pass in input_height and not output_height. by Erich Elsen · 4 years, 4 months ago
- 3b745a4 Initialize micro-kernels for pre-NEON ARM in non-mobile builds by Marat Dukhan · 4 years, 4 months ago
- d793f6c FP16 vbinary ops by Frank Barchard · 4 years, 4 months ago
- 5592e62 Fix math_nonsign_mask_f32 on MSVC by Marat Dukhan · 4 years, 4 months ago
- 8853b82 Round-to-nearest-even evaluation stubs by Marat Dukhan · 4 years, 4 months ago
- d57dd61 Few fixes in CHW DWCONV by Marat Dukhan · 4 years, 4 months ago
- 4e5db3d CHW DWCONV with implicit padding by Erich Elsen · 4 years, 4 months ago
- 0dc8f47 Fix a typo by Chao Mei · 4 years, 4 months ago
- 3f9f99f Nx16 FP16 intrinsic GEMM and IGEMM ukernels by Frank Barchard · 4 years, 4 months ago
- 5d761c4 Fix NR > 8 for FP16 intrinsic GEMM and IGEMM ukernels by Frank Barchard · 4 years, 4 months ago
- 5cb16e7 ArgMax Pooling 2D and UnPooling 2D nodes in Subgraph API by Marat Dukhan · 4 years, 4 months ago
- e4b8e57 Support TF SAME padding in Argmax Pooling operator by Marat Dukhan · 4 years, 4 months ago
- b0e4fae FP16 IGEMM microkernels by Frank Barchard · 4 years, 4 months ago
- 99003a8 Add xnn_init_f16_scaleminmax_params helper by Frank Barchard · 4 years, 4 months ago
- e92f859 Rename xnn_f16_gemm_ukernel_function to xnn_f16_gemm_minmax_ukernel_function by Frank Barchard · 4 years, 4 months ago
- e70dbeb Rename minmax_params to params for variables. by Frank Barchard · 4 years, 4 months ago
- 142268b F16 pack functions by Frank Barchard · 4 years, 4 months ago
- e826543 Move operator implementations in src/operators/ by Marat Dukhan · 4 years, 4 months ago
- 5871703 Support TF-SAME padding in Deconvolution by Marat Dukhan · 4 years, 4 months ago
- 8aaf186 Disable TSan in potentially in-place micro-kernels with OOB reads by Marat Dukhan · 4 years, 4 months ago
- 0184901 Simplify x86 detection in WAsm builds by Marat Dukhan · 4 years, 4 months ago
- f587084 Support Deconvolution in Subgraph API by Marat Dukhan · 4 years, 4 months ago
- 78583c7 Enable Cygwin and MSYS build in CMake by Marat Dukhan · 4 years, 4 months ago
- fc563f5 Replace LDP with 2 LDR for loads. by Frank Barchard · 4 years, 4 months ago
- f5425ea Additional NEON/NEONFMA DWCONV microkernels by Marat Dukhan · 4 years, 4 months ago
- 38c07ec Fully Connected node in Subgraph API by Marat Dukhan · 4 years, 4 months ago
- 9709e59 Aligned memory incorrectly being deallocated by free() by Tijmen Verhulsdonck · 4 years, 4 months ago
- 05b9830 Refactor inputs and outputs representation in a subgraph by Marat Dukhan · 4 years, 4 months ago
- 83ea939 FP16 intrinsics generate minmax names and GEMMINC by Frank Barchard · 4 years, 4 months ago
- 3b8e566 F16 8x8 GEMM ld64 microkernels by Frank Barchard · 4 years, 4 months ago
- 9b885f7 Remove PRFM from F16 GEMM microkernels by Frank Barchard · 4 years, 4 months ago
- f5cc7e7 GEMM aarch64 microkernels use LDP to fetch param and cn_stride by Frank Barchard · 4 years, 4 months ago
- c4668ed Comment fix for mr <= 4 by Frank Barchard · 4 years, 4 months ago
- 875be77 Change xnn_f16_output_params to xnn_f16_scaleminmax_params by Frank Barchard · 4 years, 5 months ago
- fd1c759 Suppress warnings about implicit type narrowing in Runtime by Marat Dukhan · 4 years, 5 months ago
- 0c864be Fix narrowing type casts in Fully Connected operator by Marat Dukhan · 4 years, 5 months ago
- 0b10b9f Fix implicit type cast warnings in NHWC Convolution by Marat Dukhan · 4 years, 5 months ago
- 5c9e889 Comment change for params from static to XNN_MIN_ELEMENTS by Frank Barchard · 4 years, 5 months ago
- 9913f9c Suppress warning about narrowing type conversion in NCHW Convolution by Marat Dukhan · 4 years, 5 months ago
- 1e0410e Remove dead code in NCHW Convolution by Marat Dukhan · 4 years, 5 months ago
- e8085f7 Comment change from clamping_params to params by Frank Barchard · 4 years, 5 months ago