diff options
author | Georgios Pinitas <georgios.pinitas@arm.com> | 2018-06-04 19:27:13 +0100 |
---|---|---|
committer | Anthony Barbier <anthony.barbier@arm.com> | 2018-11-02 16:52:54 +0000 |
commit | 17812ba9f7cf2c8f5121c11760ac45fbbdb7aeaf (patch) | |
tree | 28c7bb65a8306e82de91a644fdcc1c0947c6f6d7 /src/runtime/CL/functions/CLGEMMConvolutionLayer.cpp | |
parent | f8d8f3aff04faf731f20411ecb91027eab4365c5 (diff) | |
download | ComputeLibrary-17812ba9f7cf2c8f5121c11760ac45fbbdb7aeaf.tar.gz |
COMPMID-817: Tuner: Port kernels to new design.
Change-Id: Iaabb1153c2abe0400ec79d51a21347debe92d642
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/134062
Tested-by: Jenkins <bsgcomp@arm.com>
Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
Diffstat (limited to 'src/runtime/CL/functions/CLGEMMConvolutionLayer.cpp')
-rw-r--r-- | src/runtime/CL/functions/CLGEMMConvolutionLayer.cpp | 8 |
1 files changed, 5 insertions, 3 deletions
diff --git a/src/runtime/CL/functions/CLGEMMConvolutionLayer.cpp b/src/runtime/CL/functions/CLGEMMConvolutionLayer.cpp index 4f87043373..27bed44098 100644 --- a/src/runtime/CL/functions/CLGEMMConvolutionLayer.cpp +++ b/src/runtime/CL/functions/CLGEMMConvolutionLayer.cpp @@ -230,10 +230,11 @@ void CLGEMMConvolutionLayer::configure(const ICLTensor *input, const ICLTensor * _gemm_output.allocator()->init(info_gemm); _memory_group.manage(&_gemm_output); - // Configure im2col + // Configure and tune im2col _im2col_kernel.configure(input, &_im2col_output, Size2D(kernel_width, kernel_height), conv_info, append_bias, dilation); + CLScheduler::get().tune_kernel_static(_im2col_kernel); - // Configure GEMM + // Configure and tune GEMM configure_mm(&_im2col_output, weights, &_gemm_output); _im2col_output.allocator()->allocate(); @@ -250,8 +251,9 @@ void CLGEMMConvolutionLayer::configure(const ICLTensor *input, const ICLTensor * _gemmlowp_output_stage.configure(&_gemm_output, biases, &_tmp_output, output_multiplier, output_shift, output_quant_info.offset); } - // Configure Col2Im + // Configure and tune Col2Im _col2im_kernel.configure(_is_quantized ? &_tmp_output : &_gemm_output, output, std::make_pair(conv_w, conv_h)); + CLScheduler::get().tune_kernel_static(_col2im_kernel); if(_is_quantized) { _tmp_output.allocator()->allocate(); |