aboutsummaryrefslogtreecommitdiff
path: root/bench/qs8-gemm.cc
AgeCommit message (Collapse)Author
2020-10-13Replace QS8 4x8 with 2x8 neon microkernel.Frank Barchard
Improves performance for aarch32. PiperOrigin-RevId: 336945809
2020-10-12Cortex A55r1 QS8 GEMM microkernelFrank Barchard
PiperOrigin-RevId: 336803747
2020-10-12Add RUY benchmark to qs8_gemm_benchFrank Barchard
PiperOrigin-RevId: 336711804
2020-10-10Rename QS8 assembly GEMM kernels to ld64Frank Barchard
PiperOrigin-RevId: 336494103
2020-10-061x16 QS8 GEMM AARCH64 assembly microkernel using dot product.Frank Barchard
PiperOrigin-RevId: 335801231
2020-10-064x16 QS8 GEMM AARCH64 assembly microkernel using dot product.Frank Barchard
PiperOrigin-RevId: 335584437
2020-09-244x8, 6x8 and 8x16 Neon dot product GEMM microkernelsFrank Barchard
PiperOrigin-RevId: 333462985
2020-09-226x16 QS8 GEMM for Neon dot productFrank Barchard
PiperOrigin-RevId: 333099606
2020-08-13Add xnn_qs8_gemm_minmax_ukernel_${MR}x${NR}c4__neondot (ARMv8.2+dotprod).Benoit Jacob
PiperOrigin-RevId: 326503942
2020-08-10AVX512 variants of QS8 GEMM and IGEMM microkernelsMarat Dukhan
PiperOrigin-RevId: 325850791
2020-08-05Benchmark ARM NEON versions of QS8 GEMM microkernelsMarat Dukhan
PiperOrigin-RevId: 325127096
2020-08-05WAsm SIMD versions of QS8 GEMM and IGEMM microkernelsMarat Dukhan
PiperOrigin-RevId: 325112592
2020-08-03Rename s8rng/s32rng -> i8rng/i32rngMarat Dukhan
PiperOrigin-RevId: 324746732
2020-08-03XW (eXtended Weights) optimization for QS8 GEMM microkernelMarat Dukhan
PiperOrigin-RevId: 324734460
2020-08-03Add 3x4c8 variants of SSE2/SSSE3/SSE4.1/XOP GEMM/IGEMM microkernelsMarat Dukhan
PiperOrigin-RevId: 324710790
2020-08-02AVX2 version of QS8 GEMM and IGEMM microkernelsMarat Dukhan
PiperOrigin-RevId: 324543693
2020-08-02XOP versions of QS8 GEMM/IGEMM microkernelsMarat Dukhan
PiperOrigin-RevId: 324541139
2020-08-02Bind RNG by reference in microbenchmarksMarat Dukhan
PiperOrigin-RevId: 324540550
2020-07-31LD128 versions of QS8 GEMM SSE2/SSSE3/SSE4.1 microkernelsMarat Dukhan
PiperOrigin-RevId: 324316646
2020-07-31Add LD64 suffix in QS8 GEMM/IGEMM microkernelsMarat Dukhan
LD64 denotes that weights are loaded 64 bits at a time and sign-extended to 128 bits PiperOrigin-RevId: 324305250
2020-07-31QS8 GEMM MRx4c8 SSE2/SSSE3/SSE4.1 microkernelsMarat Dukhan
PiperOrigin-RevId: 324300862
2020-07-31QS8 GEMM microkernels and infrastructureMarat Dukhan
- QS8 GEMM microkernels for SSE2/SSSE3/SSE4.1 - Updated unit test generator to support SSSE3 ISA - Updated GEMM tester to support QS8 GEMM - Updated weights packing functions to support QS8 GEMM - Microbenchmark for QS8 GEMM microkernels PiperOrigin-RevId: 324231357