commit 23ac91b6ba235e67847802d4b49e494fa5bedbb6
author ASIAPAC\steli01 <stephen.li@arm.com> Tue Nov 07 16:14:44 2017 +0800
committer Anthony Barbier <anthony.barbier@arm.com> Fri Nov 02 16:42:17 2018 +0000
tree 49824ba9d389ee37a7656c10b960aaac5aba3567
parent 923241eb998ad031f4cce7b12d8c24a0b6c80be8

APPBROWSER-290: DC5x5 optimization for FP16

Change-Id: I0833437d8353515a29aad69c15c99c7b0fd65156
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/111631
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>

src/core/GLES_COMPUTE/kernels/GCDirectConvolutionLayerKernel.cpp

diff --git a/src/core/GLES_COMPUTE/kernels/GCDirectConvolutionLayerKernel.cpp b/src/core/GLES_COMPUTE/kernels/GCDirectConvolutionLayerKernel.cpp
index 962f044..b032bc5 100644
--- a/src/core/GLES_COMPUTE/kernels/GCDirectConvolutionLayerKernel.cpp
+++ b/src/core/GLES_COMPUTE/kernels/GCDirectConvolutionLayerKernel.cpp
@@ -211,6 +211,7 @@
         switch(input->info()->data_type())
         {
             case DataType::F16:
+                options.emplace("#define PROCESS_4X_1Y_1Z");
                 num_elems_read_per_iteration_x    = 8;
                 num_elems_written_per_iteration_x = 4;