diff options
| author | Milos Puzovic <milos.puzovic@arm.com> | 2024-03-26 14:34:30 +0000 |
|---|---|---|
| committer | Milos Puzovic <milos.puzovic@arm.com> | 2024-03-27 10:27:49 +0000 |
| commit | 905786ea0c1abb2b8df36c56eae93a97823cace1 (patch) | |
| tree | 7578977256edcffcd08bf18f839ae647f9691179 /src/core/NEON/kernels/arm_gemm/kernels/sve_ffhybrid_fp32bf16fp32_mmla_4x6VL.hpp | |
| parent | 37d84451eb1f8b4811faa4359ce154c832602782 (diff) | |
| download | ComputeLibrary-905786ea0c1abb2b8df36c56eae93a97823cace1.tar.gz | |
Added new NEON fixed format fast math mode hybrid kernel with maximum height of 6 for accumulation and updated heuristics
Change-Id: Ib52ea6825e164f4a8b8422eab7991b50af0b0d7c
Signed-off-by: Milos Puzovic <milos.puzovic@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/11354
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Jakub Sujak <jakub.sujak@arm.com>
Benchmark: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'src/core/NEON/kernels/arm_gemm/kernels/sve_ffhybrid_fp32bf16fp32_mmla_4x6VL.hpp')
| -rw-r--r-- | src/core/NEON/kernels/arm_gemm/kernels/sve_ffhybrid_fp32bf16fp32_mmla_4x6VL.hpp | 6 |
1 files changed, 4 insertions, 2 deletions
diff --git a/src/core/NEON/kernels/arm_gemm/kernels/sve_ffhybrid_fp32bf16fp32_mmla_4x6VL.hpp b/src/core/NEON/kernels/arm_gemm/kernels/sve_ffhybrid_fp32bf16fp32_mmla_4x6VL.hpp index 887d78e1de..23f686a902 100644 --- a/src/core/NEON/kernels/arm_gemm/kernels/sve_ffhybrid_fp32bf16fp32_mmla_4x6VL.hpp +++ b/src/core/NEON/kernels/arm_gemm/kernels/sve_ffhybrid_fp32bf16fp32_mmla_4x6VL.hpp @@ -1,5 +1,5 @@ /* - * Copyright (c) 2022-2023 Arm Limited. + * Copyright (c) 2022-2024 Arm Limited. * * SPDX-License-Identifier: MIT * @@ -88,8 +88,10 @@ public: { if (std::is_same<T, float>::value) { switch (ci->get_cpu_model()) { + case CPUModel::V1: + return { 28.74 }; default: - return { 32.35 }; + return { 15.27 }; } } |