COMPMID-2063: New Winograd implementation
Refactoring of winograd code reducing the size of the binaries
about 8X.
Change-Id: If8845bda324573e1a5cf436f354ac8603e88a92e
Signed-off-by: Pablo Tello <pablo.tello@arm.com>
Reviewed-on: https://review.mlplatform.org/c/959
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Anthony Barbier <Anthony.barbier@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
diff --git a/arm_compute/core/NEON/kernels/convolution/winograd/winograd_layer.hpp b/arm_compute/core/NEON/kernels/convolution/winograd/winograd_layer.hpp
new file mode 100644
index 0000000..9d418be
--- /dev/null
+++ b/arm_compute/core/NEON/kernels/convolution/winograd/winograd_layer.hpp
@@ -0,0 +1,211 @@
+/*
+ * 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.
+ */
+
+#pragma once
+
+#include <utility>
+
+#include "arm_gemm_local.hpp"
+#include "arm_gemm.hpp"
+#include "winograd.hpp"
+
+namespace winograd
+{
+
+
+class IWinogradConvolutionLayer
+{
+ public:
+ virtual ~IWinogradConvolutionLayer() = default;
+
+ virtual unsigned int weight_transform_get_window(void) const = 0;
+ virtual void weight_transform_run(unsigned int start, unsigned int stop) = 0;
+
+ virtual ITransform& input_transform(void) = 0; // Expose the input transform
+ virtual ITransform& output_transform(void) = 0; // Expose the output transform
+ virtual arm_gemm::IGemmCommon *gemm(void) = 0; // Expose the underlying GEMM
+};
+
+/** Example of how to construct an ACL-like interface.
+ *
+ * Use `get_weight_storage_size`, `get_input_storage_size` and
+ * `get_output_storage_size` to allocate memory for the convolution engine.
+ * Then create a `WinogradConvolutionLayer`.
+ *
+ * Initialise the weights using `weights_transform.run(...)`.
+ *
+ * For each inference:
+ * 1. Transform the inputs to the Winograd domain using `input_transform.run(...)`
+ * 2. Perform a number of GEMMs using `gemms.run(...)`
+ * 3. Transform the output to the spatial domain using `output_transform.run(...)`
+ */
+template <int OutputTileRows, int OutputTileCols, int KernelRows, int KernelCols,
+ typename TIn, typename TInGEMM, typename TOutGEMM, typename TOut,
+ WinogradRoots Roots>
+class WinogradConvolutionLayer : public IWinogradConvolutionLayer
+{
+ private:
+ static constexpr int InnerTileRows = OutputTileRows + KernelRows - 1;
+ static constexpr int InnerTileCols = OutputTileCols + KernelCols - 1;
+ static constexpr int N_GEMMS = InnerTileRows * InnerTileCols;
+
+ const KernelShape _kernel_shape;
+ const Tensor4DShape _input_shape;
+ const PaddingType _padding;
+ const Tensor4DShape _output_shape;
+ const int _n_output_rows, _n_output_cols;
+ const int _kernel_matrix_stride, _kernel_matrix_row_stride;
+ const int _input_matrix_stride, _input_matrix_row_stride;
+ const int _output_matrix_stride, _output_matrix_row_stride;
+ const int _tile_rows, _tile_cols;
+ const int _m, _k, _n;
+
+ public:
+ using WinogradBase = winograd::WinogradGEMM<OutputTileRows, OutputTileCols, KernelRows, KernelCols, Roots>;
+ using WeightsTransform = typename WinogradBase::template WeightsTransform<TIn, TInGEMM>;
+ using InputTransform = typename WinogradBase::template InputTransform<TIn, TInGEMM>;
+ using WinogradConv = typename WinogradBase::template Convolution<TOut, TIn, TInGEMM, TOutGEMM>;
+ using OutputTransform = typename WinogradBase::template OutputTransform<TOutGEMM, TOut>;
+
+ /* Public member variables. */
+ WeightsTransform weights_transform; /** Operator to transform weights to Winograd domain. */
+ InputTransform _input_transform; /** Operator to transform input to Winograd domain. */
+ arm_gemm::UniqueGemmCommon<TInGEMM, TOutGEMM> gemms; /** Operator to perform multiple GEMMs. */
+ OutputTransform _output_transform; /** Operator to transform output from Winograd domain. */
+
+ /** Determine how much memory (in units of TIn) to allocate for the
+ * transformed weights.
+ */
+ static unsigned int get_weight_storage_size(
+ const int n_output_channels, /** Number of output feature maps. */
+ const int n_input_channels /** Number of input feature maps. */
+ );
+
+ static unsigned int get_weight_stride(
+ const int n_output_channels, /** Number of output feature maps. */
+ const int n_input_channels /** Number of input feature maps. */
+ );
+
+ static unsigned int get_weight_multi_stride(
+ const int n_output_channels, /** Number of output feature maps. */
+ const int n_input_channels /** Number of input feature maps. */
+ );
+
+ /** Determine how much memory (in units of TIn) to allocate for the
+ * transformed input.
+ */
+ static unsigned int get_input_storage_size(
+ const int n_batches, /** Number of batches in the input tensor. */
+ const int n_channels, /** Number of feature maps in the input tensor. */
+ const int n_rows, /** Number of rows in each feature map. */
+ const int n_cols, /** Number of columns in each feature map. */
+ const bool same_padding /** Use "SAME" padding, otherwise use "VALID". */
+ );
+
+ /** Get the row stride for the A matrix in the Winograd domain. */
+ static unsigned int get_input_stride(
+ const int n_batches, /** Number of batches in the input tensor. */
+ const int n_channels, /** Number of feature maps in the input tensor. */
+ const int n_rows, /** Number of rows in each feature map. */
+ const int n_cols, /** Number of columns in each feature map. */
+ const bool same_padding /** Use "SAME" padding, otherwise use "VALID". */
+ );
+
+ /** Get the stride between A matrices in the Winograd domain. */
+ static unsigned int get_input_multi_stride(
+ const int n_batches, /** Number of batches in the input tensor. */
+ const int n_channels, /** Number of feature maps in the input tensor. */
+ const int n_rows, /** Number of rows in each feature map. */
+ const int n_cols, /** Number of columns in each feature map. */
+ const bool same_padding /** Use "SAME" padding, otherwise use "VALID". */
+ );
+
+ /** Determine how much memory (in units of TOut) to allocate for the
+ * (Winograd domain) output.
+ */
+ static unsigned int get_output_storage_size(
+ const int n_batches, /** Number of batches in the output tensor. */
+ const int n_rows, /** Number of rows in each feature map of the input tensor. */
+ const int n_cols, /** Number of columns in each feature map of the input tensor. */
+ const int n_output_channels, /** Number of feature maps in the output tensor. */
+ const bool same_padding /** Use "SAME" padding, otherwise use "VALID". */
+ );
+
+ static unsigned int get_output_stride(
+ const int n_batches, /** Number of batches in the output tensor. */
+ const int n_rows, /** Number of rows in each feature map of the input tensor. */
+ const int n_cols, /** Number of columns in each feature map of the input tensor. */
+ const int n_output_channels, /** Number of feature maps in the output tensor. */
+ const bool same_padding /** Use "SAME" padding, otherwise use "VALID". */
+ );
+
+ static unsigned int get_output_multi_stride(
+ const int n_batches, /** Number of batches in the output tensor. */
+ const int n_rows, /** Number of rows in each feature map of the input tensor. */
+ const int n_cols, /** Number of columns in each feature map of the input tensor. */
+ const int n_output_channels, /** Number of feature maps in the output tensor. */
+ const bool same_padding /** Use "SAME" padding, otherwise use "VALID". */
+ );
+
+ /** Get the shape (rows, cols) of a feature map of the output tensor. */
+ static std::pair<int, int> get_output_feature_map_shape(
+ const int n_input_rows, /** Number of rows in the input feature map. */
+ const int n_input_cols, /** Number of columns in the input feature map. */
+ const bool same_padding /** Use "SAME" padding, otherwise use "VALID". */
+ );
+
+ /** Create a new Winograd convolution layer.
+ */
+ WinogradConvolutionLayer(
+ const arm_gemm::CPUInfo &cpuinfo, /** Describes CPU properties. */
+ const int n_threads, /** Maximum number of threads used to execute the convolution. */
+ const int n_batches, /** Number of batches in the input and output tensors. */
+ const int n_input_channels, /** Number of feature maps in a batch of the input tensor. */
+ const int n_input_rows, /** Number of rows in a feature map of the input tensor. */
+ const int n_input_cols, /** Number of columns in a feature map of the input tensor. */
+ const int n_output_channels, /** Number of feature maps in the output tensor. */
+ const bool same_padding, /** Use "SAME" padding, otherwise use "VALID". */
+ const TIn* const weights, /** Pointer to weight tensor in spatial domain. Must be ordered as "Height x Rows x Input Feature Maps x Output Feature Maps. */
+ TInGEMM* const weights_storage, /** Pointer to storage for weight tensor in the Winograd domain. Must be at least the size returned by `get_weight_storage_size`. */
+ const TIn* const input, /** Pointer to NHWC ordered input tensor, in the spatial domain. */
+ TInGEMM* const winograd_input, /** Pointer to working space for the input tensor in the Winograd domain. Must be at least the size returned by `get_input_storage_size`. */
+ const TOut* const biases, /** Pointer to biases vector. Pass nullptr if no bias is provided. */
+ TOut* const output, /** Pointer to NHWC ordered output tensor, in the spatial domain. */
+ TOutGEMM* const winograd_output, /** Pointer to working space for the output tensor in the Winograd domain. Must be at least the size returned by `get_output_storage_size`. */
+ const bool pretranspose_B=true, /** Hint that the B matrix can be pretransposed. */
+ arm_gemm::GemmConfig *gemm_cfg=nullptr /** Pointer to GEMM configuration. */
+ );
+
+ /* Utility methods for interacting with the layer. */
+ unsigned int weight_transform_get_window(void) const;
+ void weight_transform_run(const unsigned int start, const unsigned int stop);
+
+ ITransform& input_transform(void);
+ ITransform& output_transform(void);
+
+ /* Get a pointer to the GEMM underlying the Winograd transform. */
+ arm_gemm::IGemmCommon *gemm(void);
+};
+
+}