From c799ed85bcf0b18269e65a64f34382073a68bd93 Mon Sep 17 00:00:00 2001 From: Gian Marco Date: Thu, 1 Feb 2018 16:57:48 +0000 Subject: 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 Tested-by: Jenkins --- src/runtime/CL/CLTuner.cpp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'src/runtime') diff --git a/src/runtime/CL/CLTuner.cpp b/src/runtime/CL/CLTuner.cpp index 917d56756a..cf5b5bce2d 100644 --- a/src/runtime/CL/CLTuner.cpp +++ b/src/runtime/CL/CLTuner.cpp @@ -204,4 +204,4 @@ cl_int Interceptor::operator()(cl_command_queue command_queue, cl_kernel kernel, _tuner.set_cl_kernel_event(tmp); return retval; -} \ No newline at end of file +} -- cgit v1.2.1