diff options
| author | zhenglin <zhenglin.li@arm.com> | 2017-12-04 14:38:09 +0800 |
|---|---|---|
| committer | Anthony Barbier <anthony.barbier@arm.com> | 2018-11-02 16:41:58 +0000 |
| commit | 666635c68ebbb182d1db4a85f33ed5325d472a65 (patch) | |
| tree | bfd45697f2a2615681adc5c3510948f8350b4710 /src/core/GLES_COMPUTE/kernels | |
| parent | 83f826537a195378fb3cc5d266fa06faf301a894 (diff) | |
| download | ComputeLibrary-666635c68ebbb182d1db4a85f33ed5325d472a65.tar.gz | |
APPBROWSER-324 FP16 Weights optimization in DirectConvolution1x1
Change-Id: I94f09caa3d9852b7ecd1d3716ba7853de9dc1c43
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/111636
Tested-by: BSG Visual Compute Jenkins server to access repositories on http://mpd-gerrit.cambridge.arm.com <bsgcomp@arm.com>
Reviewed-by: Joel Liang <joel.liang@arm.com>
Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
Diffstat (limited to 'src/core/GLES_COMPUTE/kernels')
| -rw-r--r-- | src/core/GLES_COMPUTE/kernels/GCDirectConvolutionLayerKernel.cpp | 9 |
1 files changed, 8 insertions, 1 deletions
diff --git a/src/core/GLES_COMPUTE/kernels/GCDirectConvolutionLayerKernel.cpp b/src/core/GLES_COMPUTE/kernels/GCDirectConvolutionLayerKernel.cpp index 5c7320aa8d..962f044214 100644 --- a/src/core/GLES_COMPUTE/kernels/GCDirectConvolutionLayerKernel.cpp +++ b/src/core/GLES_COMPUTE/kernels/GCDirectConvolutionLayerKernel.cpp @@ -191,6 +191,10 @@ void GCDirectConvolutionLayerKernel<kernel_size>::configure(const IGCTensor *inp case DataType::F16: num_elems_read_per_iteration_x = 8; num_elems_written_per_iteration_x = 8; + if(weights->info()->dimension(2) % 2 == 0) + { + options.emplace("#define WEIGHTS_OPTIMIZATION"); + } break; case DataType::F32: @@ -255,7 +259,10 @@ void GCDirectConvolutionLayerKernel<kernel_size>::configure(const IGCTensor *inp switch(weights->info()->data_type()) { case DataType::F16: - weights_access = AccessWindowStatic(weights->info(), 0, 0, kernel_size + 1, kernel_size); + if((weights->info()->dimension(2) % 2 != 0) || (kernel_size != 1)) + { + weights_access = AccessWindowStatic(weights->info(), 0, 0, kernel_size + 1, kernel_size); + } if(_bias != nullptr) { bias_access = AccessWindowStatic(_bias->info(), 0, 0, _bias->info()->dimension(0) + 1, 1); |