COMPMID-935 - Implementing Convolution with Winograd on OpenCL (part 4)
Implemented Winograd Output Transform (2x2,3x3) on OpenCL
Implemented CLWinogradConvolutionLayer on OpenCL
Change-Id: I6a113fc5f052ca07f878d2b800d2ab003f84af65
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/125148
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Tested-by: Jenkins <bsgcomp@arm.com>
diff --git a/arm_compute/runtime/CL/CLFunctions.h b/arm_compute/runtime/CL/CLFunctions.h
index 7c2377a..adf240e 100644
--- a/arm_compute/runtime/CL/CLFunctions.h
+++ b/arm_compute/runtime/CL/CLFunctions.h
@@ -107,6 +107,7 @@
#include "arm_compute/runtime/CL/functions/CLTranspose.h"
#include "arm_compute/runtime/CL/functions/CLWarpAffine.h"
#include "arm_compute/runtime/CL/functions/CLWarpPerspective.h"
+#include "arm_compute/runtime/CL/functions/CLWinogradConvolutionLayer.h"
#include "arm_compute/runtime/CL/functions/CLWinogradInputTransform.h"
#endif /* __ARM_COMPUTE_CLFUNCTIONS_H__ */
diff --git a/arm_compute/runtime/CL/functions/CLWinogradConvolutionLayer.h b/arm_compute/runtime/CL/functions/CLWinogradConvolutionLayer.h
new file mode 100644
index 0000000..14de169
--- /dev/null
+++ b/arm_compute/runtime/CL/functions/CLWinogradConvolutionLayer.h
@@ -0,0 +1,97 @@
+/*
+ * Copyright (c) 2018 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_CLWINOGRADCONVOLUTIONLAYER_H__
+#define __ARM_COMPUTE_CLWINOGRADCONVOLUTIONLAYER_H__
+
+#include "arm_compute/core/CL/kernels/CLWinogradFilterTransformKernel.h"
+#include "arm_compute/core/CL/kernels/CLWinogradOutputTransformKernel.h"
+#include "arm_compute/core/Types.h"
+#include "arm_compute/runtime/CL/functions/CLGEMM.h"
+#include "arm_compute/runtime/CL/functions/CLWinogradInputTransform.h"
+#include "arm_compute/runtime/IFunction.h"
+
+namespace arm_compute
+{
+class ICLTensor;
+
+/** Basic function to execute Winograd-based convolution on OpenCL. This function calls the following OpenCL functions/kernels:
+ *
+ * -# @ref CLWinogradInputTransform
+ * -# @ref CLWinogradFilterTransformKernel (only once)
+ * -# @ref CLGEMM
+ * -# @ref CLWinogradOutputTransformKernel
+ *
+ */
+class CLWinogradConvolutionLayer : public IFunction
+{
+public:
+ /** Default constructor */
+ CLWinogradConvolutionLayer(std::shared_ptr<IMemoryManager> memory_manager = nullptr);
+ /** Set the input and output tensors.
+ *
+ * @note: This function only works with 3x3 kernels and unit strides
+ *
+ * @param[in] input Source tensor. 3 lower dimensions represent a single input [width, height, IFM],
+ * while every optional dimension from 4 and above represent a batch of inputs.
+ * Data types supported: F32.
+ * @param[in] weights Weights tensor. Weights are 4D tensor with dimensions [kernel_x, kernel_y, IFM, OFM]. Data type supported:Same as @p input.
+ * @param[in] biases Biases tensor. Shared biases supported. Biases are 1D tensor with dimensions [OFM].Data type supported: Same as @p input
+ * @param[out] output Destination tensor. 3 lower dimensions represent a single output [width, height, OFM], while the rest represent batch of outputs.
+ * Data types supported: Same as @p input.
+ * @param[in] conv_info Contains padding and stride information described in @ref PadStrideInfo.
+ */
+ void configure(ICLTensor *input, const ICLTensor *weights, const ICLTensor *biases, ICLTensor *output, const PadStrideInfo &conv_info);
+ /** Static function to check if given info will lead to a valid configuration of @ref CLWinogradConvolutionLayer
+ *
+ * @note: This function only works with 3x3 kernels and unit strides
+ *
+ * @param[in] input Source tensor. 3 lower dimensions represent a single input [width, height, IFM],
+ * while every optional dimension from 4 and above represent a batch of inputs.
+ * Data types supported: F32.
+ * @param[in] weights Weights tensor. Weights are 4D tensor with dimensions [kernel_x, kernel_y, IFM, OFM]. Data type supported:Same as @p input.
+ * @param[in] biases Biases tensor. Shared biases supported. Biases are 1D tensor with dimensions [OFM].Data type supported: Same as @p input
+ * @param[out] output Destination tensor. 3 lower dimensions represent a single output [width, height, OFM], while the rest represent batch of outputs.
+ * Data types supported: Same as @p input.
+ * @param[in] conv_info Contains padding and stride information described in @ref PadStrideInfo.
+ *
+ * @return a status
+ */
+ static Status validate(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *biases, const ITensorInfo *output, const PadStrideInfo &conv_info);
+
+ // Inherited methods overridden:
+ void run() override;
+
+private:
+ CLMemoryGroup _memory_group;
+ CLGEMM _batched_mm;
+ CLWinogradInputTransform _input_transform;
+ CLWinogradFilterTransformKernel _filter_transform;
+ CLWinogradOutputTransformKernel _output_transform;
+ CLTensor _input0;
+ CLTensor _input1;
+ CLTensor _batched_mm_output;
+ bool _is_first_run;
+};
+}
+#endif /* __ARM_COMPUTE_CLWINOGRADCONVOLUTIONLAYER_H__ */