external/XNNPACK.git - [no description]

Age	Commit message (Expand)	Author
2022-02-03	Enable QU8 AAarch microkernels based on uarch	Frank Barchard
2022-02-03	Graph rewriting for FP16 inference	Marat Dukhan
2022-02-03	QS8 AArch32 GEMM benchmark build fix	Frank Barchard
2022-02-02	Add AArch32 GEMM benchmarks for Cortex A53 and Cortex A7	Frank Barchard
2022-02-02	QS8 GEMM benchmark for JIT add ISA check	Frank Barchard
2022-02-02	Enable QS8/QC8 4x8 lane GEMM AArch32 microkernel for Cortex A7	Frank Barchard
2022-02-02	Include JIT_SRCS in XNNPACK build	Marat Dukhan
2022-02-02	Support vld1r_32 with 1 or 2 register(s) in list	Zhi An Ng
2022-02-02	Fix incorrect k argument to QC8/QS8 GEMM microkernel test	Zhi An Ng
2022-02-02	Fix passing of kc JIT generator in F32 GEMM benchmarks	Zhi An Ng
2022-02-02	Include missing <limits> header in 4x8 F32 GEMM codegen for A53	Marat Dukhan
2022-02-02	Make void* params argument of JIT generators const	Zhi An Ng
2022-02-02	QS8 4x8 lane GEMM AArch32 microkernel for Cortex A7	Frank Barchard
2022-02-02	Enable QS8 4x8 lane GEMM AArch32 microkernel for Cortex A5r0 and A7	Frank Barchard
2022-02-02	Fix tfjs build by adding dependency on jit	Zhi An Ng
2022-02-02	Specialize 6x8-aarch64-neonfma-cortex-a75 on min/max params	Zhi An Ng
2022-02-02	QC8 4x8 lane GEMM AArch32 microkernel for Cortex A7	Frank Barchard
2022-02-02	Enable QU8 4x8 lane GEMM AArch32 microkernel for Cortex A53	Frank Barchard
2022-02-02	Enable QC8 4x8 lane GEMM AArch32 microkernel for Cortex A53	Frank Barchard
2022-02-02	Make SSE2 microkernels consistent with neon zip microkernels.	Alan Kelly
2022-02-02	Enable QS8 4x8 lane GEMM AArch32 microkernel for Cortex A53	Frank Barchard
2022-02-02	QS8 4x8 lane GEMM AArch32 microkernel for Cortex A53	Frank Barchard
2022-02-02	Add neon zip microkernel generator	Alan Kelly
2022-02-01	Set F32 GEMM generator function for A75 if XNN_ENABLE_JIT is set (defaults to...	Zhi An Ng
2022-02-01	Store rows in direct order in F16 GEMM microkernels	Marat Dukhan
2022-02-01	Explicitly disable -ffast-math for scalar & WAsm microkernels	Marat Dukhan
2022-02-01	Guard JIT-related structs and functionality behind XNN_PLATFORM_JIT	Zhi An Ng
2022-02-01	Integrate JIT generated GEMM microkernels into create_convolution2d_nhwc	Zhi An Ng
2022-02-01	Reoptimize QC8/QS8/QU8 GEMM/IGEMM WAsm SIMD microkernel selection	Marat Dukhan
2022-02-01	QU8 GEMM/IGEMM WAsm SIMD microkernels with SR=4	Marat Dukhan
2022-01-31	Reoptimize NEON QC8/QS8 GEMM/IGEMM microkernels with SR > 1	Marat Dukhan
2022-01-31	Re-generate amalgamated FMA3 microkernels	Marat Dukhan
2022-01-31	Reoptimize QS8/QC8 GEMM/IGEMM WAsm SIMD microkernels with swizzle	Marat Dukhan
2022-01-31	Pad K to a multiple of SR in GEMM/IGEMM microkernels	Marat Dukhan
2022-01-31	Fix excessive memory allocation for packed weights in Deconvolution	Marat Dukhan
2022-01-31	Improve test coverage for quantized Depthwise Convolutions in TFLite weight l...	Marat Dukhan
2022-01-31	Link LibM to indirection target in CMake build	Marat Dukhan
2022-01-31	Make SSE2 microkernels consistent with neon zip microkernels.	Alan Kelly
2022-01-31	Make SSE2 microkernels consistent with neon zip microkernels.	Alan Kelly
2022-01-31	Make SSE2 microkernels consistent with neon zip microkernels.	Alan Kelly
2022-01-31	Integrate JIT generated GEMM microkernels into create_convolution2d_nhwc	XNNPACK Team
2022-01-31	Guard JIT-related structs and functionality behind XNN_PLATFORM_JIT	XNNPACK Team
2022-01-28	Guard JIT-related structs and functionality behind XNN_PLATFORM_JIT	Zhi An Ng
2022-01-28	Integrate JIT generated GEMM microkernels into create_convolution2d_nhwc	Zhi An Ng
2022-01-28	Check code_buffer capacity before attempting to release it	Zhi An Ng
2022-01-27	Remove wb from JIT aarch32 instructions, use mem operand and ++ instead	Zhi An Ng
2022-01-27	Add F32 GEMM 6x8 aarch64 neonfma cortex a75 JIT microkernel to benchmark	Zhi An Ng
2022-01-27	Fix encoding of prfm	Zhi An Ng
2022-01-26	QS8/QC8 4x8 dot product IGEMM AArch32 microkernel for Cortex A55	Frank Barchard
2022-01-26	Add default cases for switch, GCC warns that control reaches the end of non-v...	Zhi An Ng