diff options
author | ramelg01 <ramy.elgammal@arm.com> | 2022-07-07 15:12:20 +0100 |
---|---|---|
committer | Pablo Marquez Tello <pablo.tello@arm.com> | 2022-07-18 05:29:41 +0000 |
commit | 1f86449ef4be37c2e0e8e6c90d4bcb6307dca1fb (patch) | |
tree | ea63e45bcfe2b556580b9a515fe16762c4b14722 /src/core/NEON/kernels/arm_gemm/kernels/sve_interleaved_bf16fp32_mmla_8x3VL.hpp | |
parent | a1f7851e2f776610019db8725c2963c36b0c85eb (diff) | |
download | ComputeLibrary-1f86449ef4be37c2e0e8e6c90d4bcb6307dca1fb.tar.gz |
Fix Neoverse V1 heuristics for FP32 fast mode
Resolves: COMPMID-5401
Signed-off-by: Ramy Elgammal <ramy.elgammal@arm.com>
Change-Id: I432b304483236efd392dfc47d541e6759c135104
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/7934
Benchmark: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Pablo Marquez Tello <pablo.tello@arm.com>
Diffstat (limited to 'src/core/NEON/kernels/arm_gemm/kernels/sve_interleaved_bf16fp32_mmla_8x3VL.hpp')
-rw-r--r-- | src/core/NEON/kernels/arm_gemm/kernels/sve_interleaved_bf16fp32_mmla_8x3VL.hpp | 6 |
1 files changed, 3 insertions, 3 deletions
diff --git a/src/core/NEON/kernels/arm_gemm/kernels/sve_interleaved_bf16fp32_mmla_8x3VL.hpp b/src/core/NEON/kernels/arm_gemm/kernels/sve_interleaved_bf16fp32_mmla_8x3VL.hpp index fc91dd71ad..1de8c68494 100644 --- a/src/core/NEON/kernels/arm_gemm/kernels/sve_interleaved_bf16fp32_mmla_8x3VL.hpp +++ b/src/core/NEON/kernels/arm_gemm/kernels/sve_interleaved_bf16fp32_mmla_8x3VL.hpp @@ -1,5 +1,5 @@ /* - * Copyright (c) 2019-2021 Arm Limited. + * Copyright (c) 2019-2022 Arm Limited. * * SPDX-License-Identifier: MIT * @@ -80,7 +80,7 @@ public: case CPUModel::A510: return { 7.78, 4.01, 2.43 }; case CPUModel::V1: - return { 47.63, 5.11, 6.80 }; + return { 62.50, 5.09, 11.32 }; } } @@ -92,7 +92,7 @@ public: case CPUModel::A510: return { 7.75, 2.47, 2.39 }; case CPUModel::V1: - return { 60.83, 2.69, 8.66 }; + return { 47.63, 5.11, 6.80 }; } } |