external/XNNPACK.git - [no description]

Age	Commit message (Expand)	Author
2 days	Softmax kernels for AVX/AVX512 generate smaller unrolled variantsupstream-master	Frank Barchard
4 days	AVX512skx RSUM F16F32ACC microkernels accumulate into output	Frank Barchard
4 days	Don't use mmap/munmap/mprotect for XNN_PLATFORM_QURT: the functions aren't av...	XNNPACK Team
7 days	Fix GEMM config for xnn_qd8_f16_qc8w_igemm_minmax_ukernel_4x16c4__neondotfp16...	Frank Barchard
7 days	Rdsum microkernels are accumlating	Alan Kelly
7 days	Add SSE rdsum microkernels	Alan Kelly
7 days	Fix GEMM config for xnn_qd8_f16_qc8w_gemm_minmax_ukernel_2x8c2s4__neonfp16arith	Frank Barchard
7 days	Fix GEMM config for xnn_qd8_f16_qc8w_igemm_minmax_ukernel_4x16c8__neoni8mm	Frank Barchard
7 days	Fix GEMM cnofig for xnn_qd8_f16_qc8w_igemm_minmax_ukernel_4x16c4__neondotfp16...	Frank Barchard
8 days	Enable F16-F32ACC-RSUM AVX512SKX microkernel for sum of FP16 values	Frank Barchard
8 days	Fix GEMM cnofig for xnn_qd8_f32_qc8w_gemm_minmax_ukernel_2x8c2s4__neon_mlal	Frank Barchard
8 days	Introduce TransposeConv with dynamic range quantization Subgraph API	Artsiom Ablavatski
8 days	Introduce QD8 TranposeConv operator	Artsiom Ablavatski
8 days	RDSum microkernels are no longer minmax and have their own tester.	Alan Kelly
8 days	Clean-up rdsum benches	Alan Kelly
8 days	Internal config change	Alan Kelly
9 days	Merge pull request #6326 from ejparkqc:master	XNNPACK Team
9 days	Merge pull request #6267 from mcr229:fc_qcint32_bias	XNNPACK Team
9 days	F16-F32ACC-RSUM AVX512SKX microkernel for sum of FP16 values	Frank Barchard
9 days	Improve GEMM unittest performance	Frank Barchard
9 days	Change one of the flags for hexagon-sim to use v68	EJ Park
9 days	F16-RMAX scalar optimized keep max sign complement	Frank Barchard
10 days	Removing build script using Hexagon v66	EJ Park
10 days	Change Hexagon minimum supported version to v68 (from v66)	EJ Park
10 days	Add f32 rsum discontig benchmarks	Alan Kelly
10 days	Add f32 rsum discontig neon microkernels	Alan Kelly
10 days	Add rsum discontiguous ukernels.	Alan Kelly
10 days	F16-RMAX - enable rmax F16C microkernel for F16C instead of AVX2	Frank Barchard
10 days	Fix math.h spelling - compliment changed to complement	Frank Barchard
10 days	F16-RMAX - enable rmax scalar for all platforms	Frank Barchard
10 days	F16-RMINMAX - move math_min_f16 and math_max_f16 to math.h	Frank Barchard
10 days	F16-RMAX benchmark include f16c_u32 for AVX	Frank Barchard
11 days	Mean op can handle arbitrary reduction axis in the contiguous axes.	Alan Kelly
11 days	Fix missing `#include`s in the `XNNPACK/src` subdirectory.	Pedro Gonnet
11 days	Remove printf that were used during debugging reduce and resize	Frank Barchard
12 days	GEMM unittest step thru NC by NextPrime	Frank Barchard
13 days	Step thru k-block test range using prime numbers	Frank Barchard
14 days	Add iterative `vsqrt` microkernels for `x86_64`, which computes `x*rsqrt(x)`,...	Pedro Gonnet
14 days	Pass VNNI and AMX flags to hardware-config.c	Frank Barchard
14 days	Add support for broadcasting of scalar weights to Prelu	Alan Kelly
2024-04-19	Automated Code Change	XNNPACK Team
2024-04-18	Rsum ukernels accumulate into output.	Alan Kelly
2024-04-17	X8-PACKW use unaligned_store_s32	Frank Barchard
2024-04-17	Use the FARF() macros for debug logging on Hexagon, rather than qurt_printf()...	XNNPACK Team
2024-04-17	Fix order of external values in slinky	Dillon Sharlet
2024-04-17	Re-generate tests after template update	Alan Kelly
2024-04-17	Use the `ReplicableRandomDevice` instead of `std::random_device`/`std::mt1993...	Pedro Gonnet
2024-04-17	Enable AVX2 F16-F32ACC GEMM for improved performance	Frank Barchard
2024-04-17	Fix missing `#include`s in `XNNPACK/test` subdirectory.	Pedro Gonnet
2024-04-16	Rollback of new `f32-vsqrt` microkernels.	Pedro Gonnet