commit c799ed85bcf0b18269e65a64f34382073a68bd93
author Gian Marco <gianmarco.iodice@arm.com> Thu Feb 01 16:57:48 2018 +0000
committer Anthony Barbier <anthony.barbier@arm.com> Fri Nov 02 16:45:42 2018 +0000
tree bda7b56321638968a24032ff4e685a2a1d58acd7
parent 604afc059c546b3c060a814f15798019cfb7a3de

COMPMID-895 - Optimizing CLDepthwiseConvolution3x3Kernel

This patch brings the MACs utilisation up to 25 % when both stride_x and stride_y are equal to 1
Performance reported in the following confluence page:
https://confluence.arm.com/display/MLENG/Depthwise+convolution+3x3+FP32+performance%3A+ACL+18.02

Change-Id: Ida1b64be9a88805902a3d90194559b58eb1224a3
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/119068
Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com>
Tested-by: Jenkins <bsgcomp@arm.com>

src/core/CL/CLKernelLibrary.cpp

diff --git a/src/core/CL/CLKernelLibrary.cpp b/src/core/CL/CLKernelLibrary.cpp
index d7ae6e3..c26d8d8 100644
--- a/src/core/CL/CLKernelLibrary.cpp
+++ b/src/core/CL/CLKernelLibrary.cpp
@@ -189,6 +189,8 @@
     { "deconvolution_upsample", "deconvolution_layer.cl" },
     { "depthwise_convolution_3x3", "depthwise_convolution.cl" },
     { "depthwise_convolution_3x3_quantized", "depthwise_convolution_quantized.cl" },
+    { "depthwise_convolution_3x3_stridex1_stridey1_bifrost", "depthwise_convolution.cl" },
+    { "depthwise_convolution_3x3_stridex2_stridey2_bifrost", "depthwise_convolution.cl" },
     { "depthwise_im2col", "depthwise_convolution.cl" },
     { "depthwise_vector_to_tensor", "depthwise_convolution.cl" },
     { "depthwise_weights_reshape", "depthwise_convolution.cl" },