diff options
| author | Georgios Pinitas <georgios.pinitas@arm.com> | 2018-02-02 12:52:07 +0000 |
|---|---|---|
| committer | Anthony Barbier <anthony.barbier@arm.com> | 2018-11-02 16:47:18 +0000 |
| commit | de5a1cc7e5c929b19fb1d3ed7d0d8783b9ac6860 (patch) | |
| tree | a1787371cdaf4976d8781bd453551a85d2fc274a /arm_compute/runtime/CL/functions/CLDepthwiseConvolutionLayer.h | |
| parent | f36ac355e050a4714a951d04a72896e02cf5e2a1 (diff) | |
| download | ComputeLibrary-de5a1cc7e5c929b19fb1d3ed7d0d8783b9ac6860.tar.gz | |
COMPMID-856: CL Depthwise Convolution QASYMM8 support
Change-Id: Ic6097e7cf160e8b829fb521b7b99d9a57d9799d3
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/118774
Tested-by: Jenkins <bsgcomp@arm.com>
Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
Diffstat (limited to 'arm_compute/runtime/CL/functions/CLDepthwiseConvolutionLayer.h')
| -rw-r--r-- | arm_compute/runtime/CL/functions/CLDepthwiseConvolutionLayer.h | 28 |
1 files changed, 16 insertions, 12 deletions
diff --git a/arm_compute/runtime/CL/functions/CLDepthwiseConvolutionLayer.h b/arm_compute/runtime/CL/functions/CLDepthwiseConvolutionLayer.h index f7899415d2..b34967d997 100644 --- a/arm_compute/runtime/CL/functions/CLDepthwiseConvolutionLayer.h +++ b/arm_compute/runtime/CL/functions/CLDepthwiseConvolutionLayer.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2017 ARM Limited. + * Copyright (c) 2017-2018 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -28,6 +28,7 @@ #include "arm_compute/core/CL/kernels/CLDepthwiseIm2ColKernel.h" #include "arm_compute/core/CL/kernels/CLDepthwiseVectorToTensorKernel.h" #include "arm_compute/core/CL/kernels/CLDepthwiseWeightsReshapeKernel.h" +#include "arm_compute/core/CL/kernels/CLDirectConvolutionLayerOutputStageKernel.h" #include "arm_compute/core/CL/kernels/CLFillBorderKernel.h" #include "arm_compute/core/CL/kernels/CLGEMMMatrixVectorMultiplyKernel.h" #include "arm_compute/core/Types.h" @@ -83,10 +84,10 @@ public: CLDepthwiseConvolutionLayer(); /** Initialize the function's source, destination, weights and convolution information. * - * @param[in, out] input Source tensor. Data type supported: F32. (Written to only for border filling). + * @param[in, out] input Source tensor. Data type supported: QASYMM8/F32. (Written to only for border filling). * @param[in] weights Weights tensor. These are 3D tensors with shape [kernel_x, kernel_y, IFM]. Data type supported: Same as @p input. * @param[in] biases (Optional) Biases tensor. A 1D tensor with shape [IFM]. Must be nullptr if not needed. - * Data type supported: Same as @p input. + * Data type supported: Same as @p input, S32 when input is QASYMM8. * @param[out] output Destination tensor. Data type supported: same as @p input. * @param[in] conv_info Padding and stride information to use for the convolution. */ @@ -96,15 +97,18 @@ public: void run() override; private: - CLDepthwiseIm2ColKernel _im2col_kernel; - CLDepthwiseWeightsReshapeKernel _weights_reshape_kernel; - CLGEMMMatrixVectorMultiplyKernel _v2mm_kernel; - CLDepthwiseVectorToTensorKernel _vector_to_tensor_kernel; - CLFillBorderKernel _v2mm_input_fill_border; - CLFillBorderKernel _v2mm_weights_fill_border; - CLTensor _input_reshaped; - CLTensor _weights_reshaped; - CLTensor _v2mm_output; + CLDepthwiseIm2ColKernel _im2col_kernel; + CLDepthwiseWeightsReshapeKernel _weights_reshape_kernel; + CLGEMMMatrixVectorMultiplyKernel _v2mm_kernel; + CLDepthwiseVectorToTensorKernel _vector_to_tensor_kernel; + CLDirectConvolutionLayerOutputStageKernel _output_stage_kernel; + CLFillBorderKernel _v2mm_input_fill_border; + CLFillBorderKernel _v2mm_weights_fill_border; + CLTensor _input_reshaped; + CLTensor _weights_reshaped; + CLTensor _v2mm_output; + CLTensor _output_reshaped; + bool _is_quantized; }; } #endif /*__ARM_COMPUTE_CLDEPTHWISECONVOLUTION_H__ */ |