diff options
author | Michele Di Giorgio <michele.digiorgio@arm.com> | 2021-07-28 14:10:47 +0100 |
---|---|---|
committer | Georgios Pinitas <georgios.pinitas@arm.com> | 2021-07-28 16:31:53 +0000 |
commit | a86433a64ab25c2ea8e274bd2f357a9709636f5b (patch) | |
tree | 48978522014f8d0423522ec37f2d491bee110944 /src/core/cpu/kernels/CpuConvertFullyConnectedWeightsKernel.h | |
parent | 120c87b1622fa2472d27e43e5b7d861fb9735d9c (diff) | |
download | ComputeLibrary-a86433a64ab25c2ea8e274bd2f357a9709636f5b.tar.gz |
Reduce binary footprint of CpuConvertFullyConnectedWeightsKernel
Binary size reduction for this kernel is almost 50%.
Also remove unused NEConvertFullyConnectedWeightsManaged.
Change-Id: Ia46a1342a0737397b4aac2578d963c2ebb7446e3
Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6011
Reviewed-by: Giorgio Arena <giorgio.arena@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'src/core/cpu/kernels/CpuConvertFullyConnectedWeightsKernel.h')
-rw-r--r-- | src/core/cpu/kernels/CpuConvertFullyConnectedWeightsKernel.h | 9 |
1 files changed, 0 insertions, 9 deletions
diff --git a/src/core/cpu/kernels/CpuConvertFullyConnectedWeightsKernel.h b/src/core/cpu/kernels/CpuConvertFullyConnectedWeightsKernel.h index 3ba3162c34..7baaf13417 100644 --- a/src/core/cpu/kernels/CpuConvertFullyConnectedWeightsKernel.h +++ b/src/core/cpu/kernels/CpuConvertFullyConnectedWeightsKernel.h @@ -69,15 +69,6 @@ public: private: unsigned int _factor1{ 0 }; /* equals to the number of elements per original src plane if @p data_layout == NCHW; its number of channels otherwise */ unsigned int _factor2{ 0 }; /* equals to the number of elements per original src plane if @p data_layout == NHWC; its number of channels otherwise */ - - /** Template function to run the permute - * - * @param[in] in Source weights tensor info to convert. Must be 2 dimensional. Data types supported: All. - * @param[in] out The converted weights tensor info. Shape and Data Type: Same as @p in. - * @param[in] window Region on which to execute the kernel. (Must be a valid region of the window returned by window()). - */ - template <typename T> - void run_convert_fc_weights(const ITensor *in, ITensor *out, const Window &window); }; } // namespace kernels } // namespace cpu |