diff options
| author | Omar Al Khatib <omar.alkhatib@arm.com> | 2024-01-02 14:45:07 +0000 |
|---|---|---|
| committer | Omar Al Khatib <omar.alkhatib@arm.com> | 2024-03-12 15:45:42 +0000 |
| commit | 93e743fbe7d52f4c41fcd90762fc38b95be802f7 (patch) | |
| tree | d0ded85f3cf08f3aabcac68caee4842f3e94da4a /src/cpu/kernels/softmax/generic/neon/fp32.cpp | |
| parent | d0611c10a08a4e4f78885e76856155a1f05e6720 (diff) | |
| download | ComputeLibrary-93e743fbe7d52f4c41fcd90762fc38b95be802f7.tar.gz | |
Optimize CpuSoftmaxKernel for axis != 0 and neon kernels
Resolves: COMPMID-6501
Signed-off-by: Omar Al Khatib <omar.alkhatib@arm.com>
Change-Id: I0abd3cbb5f861301f407c443988fb7efaa205b5d
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/11056
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Gunes Bayir <gunes.bayir@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Benchmark: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'src/cpu/kernels/softmax/generic/neon/fp32.cpp')
| -rw-r--r-- | src/cpu/kernels/softmax/generic/neon/fp32.cpp | 22 |
1 files changed, 15 insertions, 7 deletions
diff --git a/src/cpu/kernels/softmax/generic/neon/fp32.cpp b/src/cpu/kernels/softmax/generic/neon/fp32.cpp index c281d1bf31..0701620636 100644 --- a/src/cpu/kernels/softmax/generic/neon/fp32.cpp +++ b/src/cpu/kernels/softmax/generic/neon/fp32.cpp @@ -1,5 +1,5 @@ /* - * Copyright (c) 2021-2023 Arm Limited. + * Copyright (c) 2021-2024 Arm Limited. * * SPDX-License-Identifier: MIT * @@ -31,15 +31,23 @@ namespace cpu { template <bool IS_LOG> -void neon_fp32_softmax(const ITensor *in, void *const tmp, ITensor *out, const float beta, const Window &window) +void neon_fp32_softmax( + const ITensor *in, void *const tmp, ITensor *out, const float beta, int axis, const Window &window) { - return neon_softmax_float<float, IS_LOG>(in, tmp, out, beta, window); + if (axis == 0) + { + return neon_softmax_x_float<float, IS_LOG>(in, tmp, out, beta, axis, window); + } + else + { + return neon_softmax_non_x_float<float, IS_LOG>(in, tmp, out, beta, axis, window); + } } -template void -neon_fp32_softmax<true>(const ITensor *in, void *const tmp, ITensor *out, const float beta, const Window &window); -template void -neon_fp32_softmax<false>(const ITensor *in, void *const tmp, ITensor *out, const float beta, const Window &window); +template void neon_fp32_softmax<true>( + const ITensor *in, void *const tmp, ITensor *out, const float beta, int axis, const Window &window); +template void neon_fp32_softmax<false>( + const ITensor *in, void *const tmp, ITensor *out, const float beta, int axis, const Window &window); } // namespace cpu } // namespace arm_compute |