diff options
| author | Michael Tyler <michael.tyler@arm.com> | 2023-01-17 11:04:14 +0000 |
|---|---|---|
| committer | Gian Marco Iodice <gianmarco.iodice@arm.com> | 2023-01-18 09:43:38 +0000 |
| commit | be13cead34e566bdd561ad3ffc3f645b460e482e (patch) | |
| tree | cdc086de205d5a07fdd816afa6333d0b2f38d4e9 /src/core/NEON/kernels/arm_gemm/gemm_uint8.cpp | |
| parent | 13bab71a76096985752a9e12711507021e25858d (diff) | |
| download | ComputeLibrary-be13cead34e566bdd561ad3ffc3f645b460e482e.tar.gz | |
Revert "Update CPU kernels to remove x19"
This reverts commit 3c59f01c209d2732a15d97d65565ead964787a8b.
Resolves: COMPMID-5817
Change-Id: Ie2443a21854a95db1e3d0cafa2121c0187a5e237
Signed-off-by: Michael Tyler <michael.tyler@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8974
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Benchmark: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'src/core/NEON/kernels/arm_gemm/gemm_uint8.cpp')
| -rw-r--r-- | src/core/NEON/kernels/arm_gemm/gemm_uint8.cpp | 10 |
1 files changed, 9 insertions, 1 deletions
diff --git a/src/core/NEON/kernels/arm_gemm/gemm_uint8.cpp b/src/core/NEON/kernels/arm_gemm/gemm_uint8.cpp index fcc95eb503..03e9cd6c1f 100644 --- a/src/core/NEON/kernels/arm_gemm/gemm_uint8.cpp +++ b/src/core/NEON/kernels/arm_gemm/gemm_uint8.cpp @@ -1,5 +1,5 @@ /* - * Copyright (c) 2017-2020, 2022-2023 Arm Limited. + * Copyright (c) 2017-2020, 2022 Arm Limited. * * SPDX-License-Identifier: MIT * @@ -43,6 +43,7 @@ #include "kernels/sve_hybrid_u8u32_mmla_6x4VL.hpp" #include "kernels/sve_interleaved_u8u32_dot_8x3VL.hpp" #include "kernels/sve_interleaved_u8u32_mmla_8x3VL.hpp" +#include "kernels/sve_smallK_hybrid_u8u32_dot_8x1VL.hpp" namespace arm_gemm { @@ -62,6 +63,13 @@ GemmImplementation<uint8_t, uint32_t>::with_estimate( [](const GemmArgs &args) { return GemmInterleaved<cls_sve_interleaved_u8u32_mmla_8x3VL, uint8_t, uint32_t>::estimate_cycles<uint32_t>(args); }, [](const GemmArgs &args) { return new GemmInterleaved<cls_sve_interleaved_u8u32_mmla_8x3VL, uint8_t, uint32_t>(args); } ), +{ + GemmMethod::GEMM_HYBRID, + "sve_smallK_hybrid_u8u32_dot_8x1VL", + [](const GemmArgs &args) { return args._ci->has_sve() && args._Ksize<=64 && !args._indirect_input; }, + [](const GemmArgs &args) { return !(args._ci->has_svei8mm() || args._ci->has_i8mm()); }, + [](const GemmArgs &args) { return new GemmHybrid<cls_sve_smallK_hybrid_u8u32_dot_8x1VL, uint8_t, uint32_t>(args); } +}, GemmImplementation<uint8_t, uint32_t>::with_estimate( GemmMethod::GEMM_HYBRID, "sve_hybrid_u8u32_dot_6x4VL", |