From 4406fd6cc4abded564d3791324e1f48bdfd34273 Mon Sep 17 00:00:00 2001 From: Frank Lei Date: Thu, 1 Feb 2018 14:47:14 +0800 Subject: APPBROWSER-391: Fix GLES COMPUTE alignment issues APPBROWSER-402: Performance optimization for squeezenet/xray model Change-Id: If31b186b99a6d6087164019fe94d3ac9279e3204 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/119526 Tested-by: Jenkins Reviewed-by: Georgios Pinitas --- .../functions/GCDepthwiseConvolutionLayer.h | 18 +++++++++++++++--- 1 file changed, 15 insertions(+), 3 deletions(-) (limited to 'arm_compute/runtime/GLES_COMPUTE/functions/GCDepthwiseConvolutionLayer.h') diff --git a/arm_compute/runtime/GLES_COMPUTE/functions/GCDepthwiseConvolutionLayer.h b/arm_compute/runtime/GLES_COMPUTE/functions/GCDepthwiseConvolutionLayer.h index 7b99ea5645..e523356fca 100644 --- a/arm_compute/runtime/GLES_COMPUTE/functions/GCDepthwiseConvolutionLayer.h +++ b/arm_compute/runtime/GLES_COMPUTE/functions/GCDepthwiseConvolutionLayer.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2017 ARM Limited. + * Copyright (c) 2017-2018 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -25,8 +25,10 @@ #define __ARM_COMPUTE_GCDEPTHWISECONVOLUTION_H__ #include "arm_compute/core/GLES_COMPUTE/kernels/GCDepthwiseConvolutionLayer3x3Kernel.h" +#include "arm_compute/core/GLES_COMPUTE/kernels/GCFillBorderKernel.h" +#include "arm_compute/core/GLES_COMPUTE/kernels/GCTensorShiftKernel.h" #include "arm_compute/core/Types.h" -#include "arm_compute/runtime/GLES_COMPUTE/IGCSimpleFunction.h" +#include "arm_compute/runtime/IFunction.h" namespace arm_compute { @@ -38,9 +40,11 @@ class IGCTensor; * -# @ref GCFillBorderKernel (if pad_x or pad_y > 0) * */ -class GCDepthwiseConvolutionLayer3x3 : public IGCSimpleFunction +class GCDepthwiseConvolutionLayer3x3 : public IFunction { public: + /** Default constructor */ + GCDepthwiseConvolutionLayer3x3(); /** Initialize the function's source, destination, conv and border_size. * * @param[in, out] input Source tensor. Data type supported: F16. (Written to only for border filling). @@ -51,6 +55,14 @@ public: * @param[in] conv_info Padding and stride information to use for the convolution. */ void configure(IGCTensor *input, const IGCTensor *weights, const IGCTensor *biases, IGCTensor *output, const PadStrideInfo &conv_info); + + // Inherited methods overridden: + void run() override final; + +private: + std::unique_ptr _kernel; + GCFillBorderKernel _border_handler; + GCTensorShiftKernel _shift_handler; }; } #endif /*__ARM_COMPUTE_GCDEPTHWISECONVOLUTION_H__ */ -- cgit v1.2.1