commit b9d38ee6378f3035f8dbad442223d3d9e2f3dc4f
author Frank Lei <frank.lei@arm.com> Tue Dec 05 10:43:33 2017 +0800
committer Anthony Barbier <anthony.barbier@arm.com> Fri Nov 02 16:42:17 2018 +0000
tree 89a4b81430100a4a91902d5987ae42edc438012c
parent 397d58aa40b02a26923c34d8cd4ba274eac45963

APPBROWSER-312 Fully connected performance optimization

Change-Id: Ie93fd630ebbad7b6ca8812cb5044b3f1908b45fd
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/111830
Reviewed-by: Stephen Li <stephen.li@arm.com>
Tested-by: BSG Visual Compute Jenkins server to access repositories on http://mpd-gerrit.cambridge.arm.com <bsgcomp@arm.com>
Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>

src/core/GLES_COMPUTE/kernels/GCIm2ColKernel.cpp

diff --git a/src/core/GLES_COMPUTE/kernels/GCIm2ColKernel.cpp b/src/core/GLES_COMPUTE/kernels/GCIm2ColKernel.cpp
index 97c4dc4..e849891 100644
--- a/src/core/GLES_COMPUTE/kernels/GCIm2ColKernel.cpp
+++ b/src/core/GLES_COMPUTE/kernels/GCIm2ColKernel.cpp
@@ -107,7 +107,38 @@
     else
     {
         build_opts.insert("#define IM2COL_REDUCED");
-        _num_elems_processed_per_iteration = 4 / input->info()->element_size();
+
+        if(input->info()->data_type() == DataType::F32)
+        {
+            _num_elems_processed_per_iteration = 4 / input->info()->element_size();
+        }
+        else if(input->info()->data_type() == DataType::F16)
+        {
+            int input_width  = input->info()->dimension(0);
+            int input_height = input->info()->dimension(1);
+
+            build_opts.insert("#define IMAGE_SIZE " + support::cpp11::to_string(input_width * input_height));
+            if(input_width % 8 == 0)
+            {
+                _num_elems_processed_per_iteration = 8;
+                build_opts.insert("#define IM2COL_REDUCED_8X");
+            }
+            else if(input_width % 4 == 0)
+            {
+                _num_elems_processed_per_iteration = 4;
+                build_opts.insert("#define IM2COL_REDUCED_4X");
+            }
+            else if(input_width % 2 == 0)
+            {
+                _num_elems_processed_per_iteration = 2;
+                build_opts.insert("#define IM2COL_REDUCED_2X");
+            }
+            else
+            {
+                _num_elems_processed_per_iteration = 2;
+                build_opts.insert("#define IM2COL_REDUCED_GENERIC");
+            }
+        }
 
         // Create kernel
         _kernel = static_cast<GCKernel>(GCKernelLibrary::get().create_kernel("im2col_reduced", build_opts));