From 8be9148814b88e5b0cabd5a4d2b1f4ff470a8c1c Mon Sep 17 00:00:00 2001 From: Georgios Pinitas Date: Tue, 26 Mar 2019 17:23:28 +0000 Subject: COMPMID-1959: Implements 2D FFT on OpenCL Change-Id: I73cf3984a5463acc854c8a59dc2bd9a5234cd99c Signed-off-by: Georgios Pinitas Reviewed-on: https://review.mlplatform.org/c/936 Comments-Addressed: Arm Jenkins Tested-by: Arm Jenkins Reviewed-by: Gian Marco Iodice --- arm_compute/core/CL/CLKernels.h | 1 + .../core/CL/kernels/CLFFTDigitReverseKernel.h | 10 +-- .../core/CL/kernels/CLFFTRadixStageKernel.h | 4 +- arm_compute/core/CL/kernels/CLFFTScaleKernel.h | 78 ++++++++++++++++++++++ .../CL/kernels/CLPixelWiseMultiplicationKernel.h | 48 +++++++++++-- 5 files changed, 131 insertions(+), 10 deletions(-) create mode 100644 arm_compute/core/CL/kernels/CLFFTScaleKernel.h (limited to 'arm_compute/core/CL') diff --git a/arm_compute/core/CL/CLKernels.h b/arm_compute/core/CL/CLKernels.h index e3ffcd0704..57498715c8 100644 --- a/arm_compute/core/CL/CLKernels.h +++ b/arm_compute/core/CL/CLKernels.h @@ -66,6 +66,7 @@ #include "arm_compute/core/CL/kernels/CLErodeKernel.h" #include "arm_compute/core/CL/kernels/CLFFTDigitReverseKernel.h" #include "arm_compute/core/CL/kernels/CLFFTRadixStageKernel.h" +#include "arm_compute/core/CL/kernels/CLFFTScaleKernel.h" #include "arm_compute/core/CL/kernels/CLFastCornersKernel.h" #include "arm_compute/core/CL/kernels/CLFillBorderKernel.h" #include "arm_compute/core/CL/kernels/CLFlattenLayerKernel.h" diff --git a/arm_compute/core/CL/kernels/CLFFTDigitReverseKernel.h b/arm_compute/core/CL/kernels/CLFFTDigitReverseKernel.h index 10652cdb4d..3082cb186f 100644 --- a/arm_compute/core/CL/kernels/CLFFTDigitReverseKernel.h +++ b/arm_compute/core/CL/kernels/CLFFTDigitReverseKernel.h @@ -26,6 +26,8 @@ #include "arm_compute/core/CL/ICLKernel.h" +#include "arm_compute/core/KernelDescriptors.h" + namespace arm_compute { // Forward declarations @@ -52,19 +54,19 @@ public: * @param[in] input Source tensor. Data types supported: F32. * @param[out] output Destination tensor. Data type supported: same as @p input * @param[in] idx Digit reverse index tensor. Data type supported: U32 - * @param[in] axis Axis to perform digit reverse on. + * @param[in] config Kernel configuration. */ - void configure(const ICLTensor *input, ICLTensor *output, const ICLTensor *idx, unsigned int axis); + void configure(const ICLTensor *input, ICLTensor *output, const ICLTensor *idx, const FFTDigitReverseKernelInfo &config); /** Static function to check if given info will lead to a valid configuration of @ref CLFFTDigitReverseKernel * * @param[in] input Source tensor info. Data types supported: F32. * @param[in] output Destination tensor info. Data type supported: same as @p input * @param[in] idx Digit reverse index tensor info. Data type supported: U32 - * @param[in] axis Axis to perform digit reverse on. + * @param[in] config Kernel configuration. * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *output, const ITensorInfo *idx, unsigned int axis); + static Status validate(const ITensorInfo *input, const ITensorInfo *output, const ITensorInfo *idx, const FFTDigitReverseKernelInfo &config); // Inherited methods overridden: void run(const Window &window, cl::CommandQueue &queue) override; diff --git a/arm_compute/core/CL/kernels/CLFFTRadixStageKernel.h b/arm_compute/core/CL/kernels/CLFFTRadixStageKernel.h index 9de775eafa..16fa390e5d 100644 --- a/arm_compute/core/CL/kernels/CLFFTRadixStageKernel.h +++ b/arm_compute/core/CL/kernels/CLFFTRadixStageKernel.h @@ -59,7 +59,7 @@ public: * @param[out] output Destination tensor. Can be nullptr. Data type supported: same as @p input * @param[in] config FFT descriptor metadata. */ - void configure(ICLTensor *input, ICLTensor *output, const FFTRadixStageKernelDescriptor &config); + void configure(ICLTensor *input, ICLTensor *output, const FFTRadixStageKernelInfo &config); /** Static function to check if given info will lead to a valid configuration of @ref CLFFTRadixStageKernel * * @param[in] input Source tensor info. Data types supported: F32. @@ -68,7 +68,7 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *output, const FFTRadixStageKernelDescriptor &config); + static Status validate(const ITensorInfo *input, const ITensorInfo *output, const FFTRadixStageKernelInfo &config); /** Returns the radix that are support by the FFT kernel * * @return A set of supported radix diff --git a/arm_compute/core/CL/kernels/CLFFTScaleKernel.h b/arm_compute/core/CL/kernels/CLFFTScaleKernel.h new file mode 100644 index 0000000000..39ecac42af --- /dev/null +++ b/arm_compute/core/CL/kernels/CLFFTScaleKernel.h @@ -0,0 +1,78 @@ +/* + * Copyright (c) 2019 ARM Limited. + * + * SPDX-License-Identifier: MIT + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software and associated documentation files (the "Software"), to + * deal in the Software without restriction, including without limitation the + * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or + * sell copies of the Software, and to permit persons to whom the Software is + * furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in all + * copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ +#ifndef __ARM_COMPUTE_CLFFTSCALEKERNEL_H__ +#define __ARM_COMPUTE_CLFFTSCALEKERNEL_H__ + +#include "arm_compute/core/CL/ICLKernel.h" + +#include "arm_compute/core/KernelDescriptors.h" + +namespace arm_compute +{ +// Forward declarations +class ICLTensor; + +/** Interface for the inverse fft scale kernel. */ +class CLFFTScaleKernel : public ICLKernel +{ +public: + /** Constructor */ + CLFFTScaleKernel(); + /** Prevent instances of this class from being copied (As this class contains pointers) */ + CLFFTScaleKernel(const CLFFTScaleKernel &) = delete; + /** Prevent instances of this class from being copied (As this class contains pointers) */ + CLFFTScaleKernel &operator=(const CLFFTScaleKernel &) = delete; + /** Default Move Constructor. */ + CLFFTScaleKernel(CLFFTScaleKernel &&) = default; + /** Default move assignment operator */ + CLFFTScaleKernel &operator=(CLFFTScaleKernel &&) = default; + /** Default destructor */ + ~CLFFTScaleKernel() = default; + /** Set the input and output tensors. + * + * @param[in,out] input Source tensor. Data types supported: F32. + * @param[out] output Destination tensor. Data type supported: same as @p input + * @param[in] config Kernel configuration + */ + void configure(ICLTensor *input, ICLTensor *output, const FFTScaleKernelInfo &config); + /** Static function to check if given info will lead to a valid configuration of @ref CLFFTScaleKernel + * + * @param[in] input Source tensor info. Data types supported: F32. + * @param[in] output Destination tensor info. Data type supported: same as @p input + * @param[in] config Kernel configuration + * + * @return a status + */ + static Status validate(const ITensorInfo *input, const ITensorInfo *output, const FFTScaleKernelInfo &config); + + // Inherited methods overridden: + void run(const Window &window, cl::CommandQueue &queue) override; + +private: + ICLTensor *_input; + ICLTensor *_output; + bool _run_in_place; +}; +} // namespace arm_compute +#endif /*__ARM_COMPUTE_CLFFTSCALEKERNEL_H__ */ diff --git a/arm_compute/core/CL/kernels/CLPixelWiseMultiplicationKernel.h b/arm_compute/core/CL/kernels/CLPixelWiseMultiplicationKernel.h index b835aa701b..804182b187 100644 --- a/arm_compute/core/CL/kernels/CLPixelWiseMultiplicationKernel.h +++ b/arm_compute/core/CL/kernels/CLPixelWiseMultiplicationKernel.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2016-2018 ARM Limited. + * Copyright (c) 2016-2019 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -29,11 +29,10 @@ namespace arm_compute { +// Forward declarations class ICLTensor; -/** Interface for the pixelwise multiplication kernel. - * - */ +/** Interface for the pixelwise multiplication kernel. */ class CLPixelWiseMultiplicationKernel : public ICLKernel { public: @@ -78,6 +77,47 @@ public: void run(const Window &window, cl::CommandQueue &queue) override; BorderSize border_size() const override; +private: + const ICLTensor *_input1; + const ICLTensor *_input2; + ICLTensor *_output; +}; + +/** Interface for the complex pixelwise multiplication kernel. */ +class CLComplexPixelWiseMultiplicationKernel : public ICLKernel +{ +public: + /** Default constructor.*/ + CLComplexPixelWiseMultiplicationKernel(); + /** Prevent instances of this class from being copied (As this class contains pointers) */ + CLComplexPixelWiseMultiplicationKernel(const CLComplexPixelWiseMultiplicationKernel &) = delete; + /** Prevent instances of this class from being copied (As this class contains pointers) */ + CLComplexPixelWiseMultiplicationKernel &operator=(const CLComplexPixelWiseMultiplicationKernel &) = delete; + /** Allow instances of this class to be moved */ + CLComplexPixelWiseMultiplicationKernel(CLComplexPixelWiseMultiplicationKernel &&) = default; + /** Allow instances of this class to be moved */ + CLComplexPixelWiseMultiplicationKernel &operator=(CLComplexPixelWiseMultiplicationKernel &&) = default; + /** Initialise the kernel's input, output and border mode. + * + * @param[in] input1 An input tensor. Data types supported: F32. Number of channels supported: 2. + * @param[in] input2 An input tensor. Data types supported: same as @p input1. Number of channels supported: same as @p input1. + * @param[out] output The output tensor, Data types supported: same as @p input1. Number of channels supported: same as @p input1. + */ + void configure(const ICLTensor *input1, const ICLTensor *input2, ICLTensor *output); + /** Static function to check if given info will lead to a valid configuration of @ref CLComplexPixelWiseMultiplicationKernel + * + * @param[in] input1 An input tensor info. Data types supported: F32. Number of channels supported: 2. + * @param[in] input2 An input tensor info. Data types supported: same as @p input1. Number of channels supported: same as @p input1. + * @param[in] output The output tensor info. Data types supported: same as @p input1. Number of channels supported: same as @p input1. + * + * @return a status + */ + static Status validate(const ITensorInfo *input1, const ITensorInfo *input2, const ITensorInfo *output); + + // Inherited methods overridden: + void run(const Window &window, cl::CommandQueue &queue) override; + BorderSize border_size() const override; + private: const ICLTensor *_input1; const ICLTensor *_input2; -- cgit v1.2.1