src/core/NEON/kernels/NEDepthwiseConvolutionLayerNativeKernel.cpp

/*
 * 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.
 */
#include "arm_compute/core/NEON/kernels/NEDepthwiseConvolutionLayerNativeKernel.h"

#include "arm_compute/core/AccessWindowStatic.h"
#include "arm_compute/core/NEON/wrapper/traits.h"
#include "arm_compute/core/NEON/wrapper/wrapper.h"
#include "arm_compute/core/utils/misc/ShapeCalculator.h"

namespace arm_compute
{
namespace
{
template <typename T, int S, bool has_biases>
void depthwise_loop_multiplier1(const ITensor *input, const ITensor *weights, const ITensor *biases, ITensor *output, const PadStrideInfo &conv_info,
                                const Size2D &dilation, const Window &window)
{
    using VectorType = typename wrapper::traits::neon_vector<T, S>::type;
    using TagType    = typename wrapper::traits::neon_vector<T, S>::tag_type;

    const size_t input_stride_y   = input->info()->strides_in_bytes().y();
    const size_t input_stride_z   = input->info()->strides_in_bytes().z();
    const size_t input_max_offset = input->info()->strides_in_bytes().z() * input->info()->dimension(2) - (input->info()->padding().bottom + input->info()->padding().top) *
                                    input->info()->strides_in_bytes().y();
    const size_t weights_width    = weights->info()->dimension(1);
    const size_t weights_height   = weights->info()->dimension(2);
    const size_t weights_stride_y = weights->info()->strides_in_bytes().y();
    const size_t weights_stride_z = weights->info()->strides_in_bytes().z();
    const size_t conv_stride_x    = conv_info.stride().first;
    const size_t conv_stride_y    = conv_info.stride().second;
    const size_t conv_pad_left    = conv_info.pad_left();
    const size_t conv_pad_top     = conv_info.pad_top();

    Window win_input = window;
    win_input.set(Window::DimY, Window::Dimension(0, 0, 0));
    win_input.set(Window::DimZ, Window::Dimension(0, 0, 0));

    Window win_weights = win_input;
    win_weights.set(3, Window::Dimension(0, 0, 0));

    Iterator input_it(input, win_input);
    Iterator weights_it(weights, win_weights);
    Iterator output_it(output, window);
    Iterator biases_it{};

    if(has_biases)
    {
        biases_it = Iterator(biases, win_weights);
    }

    execute_window_loop(window, [&](const Coordinates & id)
    {
        VectorType acc = wrapper::vdup_n(static_cast<T>(0), TagType{});

        const int input_y      = id.y() * conv_stride_x - conv_pad_left;
        const int input_z      = id.z() * conv_stride_y - conv_pad_top;
        int       input_offset = input_y * input_stride_y + input_z * input_stride_z;

        auto weights_ptr = weights_it.ptr();
        for(size_t h = 0; h < weights_height; ++h)
        {
            int offs = input_offset;
            for(size_t w = 0; w < weights_width; ++w)
            {
                const auto input_vals   = wrapper::vload(reinterpret_cast<T *>(input_it.ptr() + std::min(static_cast<size_t>(offs), input_max_offset)));
                const auto weights_vals = wrapper::vload(reinterpret_cast<T *>(weights_ptr + w * weights_stride_y));

                acc = wrapper::vmla(acc, weights_vals, input_vals);
                offs += dilation.x() * input_stride_y;
            }

            weights_ptr += weights_stride_z;
            input_offset += dilation.y() * input_stride_z;
        }

        if(has_biases)
        {
            const auto biases_vals = wrapper::vload(reinterpret_cast<T *>(biases_it.ptr()));
            acc                    = wrapper::vadd(acc, biases_vals);
        }

        wrapper::vstore(reinterpret_cast<T *>(output_it.ptr()), acc);
    },
    input_it, weights_it, biases_it, output_it);
}

template <typename T, bool has_biases>
void depthwise_loop_generic(const ITensor *input, const ITensor *weights, const ITensor *biases, ITensor *output, const PadStrideInfo &conv_info,
                            const Size2D &dilation, unsigned int depth_multiplier, const Window &window)
{
    const size_t input_stride_y   = input->info()->strides_in_bytes().y();
    const size_t input_stride_z   = input->info()->strides_in_bytes().z();
    const size_t input_max_offset = input->info()->strides_in_bytes().z() * input->info()->dimension(2) - (input->info()->padding().bottom + input->info()->padding().top) *
                                    input->info()->strides_in_bytes().y();
    const size_t weights_width    = weights->info()->dimension(1);
    const size_t weights_height   = weights->info()->dimension(2);
    const size_t weights_stride_y = weights->info()->strides_in_bytes().y();
    const size_t weights_stride_z = weights->info()->strides_in_bytes().z();
    const size_t conv_stride_x    = conv_info.stride().first;
    const size_t conv_stride_y    = conv_info.stride().second;
    const size_t conv_pad_left    = conv_info.pad_left();
    const size_t conv_pad_top     = conv_info.pad_top();

    Window win_input = window;
    win_input.set(Window::DimY, Window::Dimension(0, 0, 0));
    win_input.set(Window::DimZ, Window::Dimension(0, 0, 0));

    Window win_weights = win_input;
    win_weights.set(3, Window::Dimension(0, 0, 0));

    win_input.set_dimension_step(Window::DimX, 1);

    Iterator input_it(input, win_input);
    Iterator weights_it(weights, win_weights);
    Iterator output_it(output, window);
    Iterator biases_it{};

    if(has_biases)
    {
        biases_it = Iterator(biases, win_weights);
    }

    execute_window_loop(window, [&](const Coordinates & id)
    {
        std::vector<T> acc(depth_multiplier, static_cast<T>(0));

        const int input_y      = id.y() * conv_stride_x - conv_pad_left;
        const int input_z      = id.z() * conv_stride_y - conv_pad_top;
        int       input_offset = input_y * input_stride_y + input_z * input_stride_z;

        auto weights_ptr = weights_it.ptr();
        for(size_t h = 0; h < weights_height; ++h)
        {
            int offs = input_offset;
            for(size_t w = 0; w < weights_width; ++w)
            {
                const auto input_val = *(reinterpret_cast<T *>(input_it.ptr() + std::min(static_cast<size_t>(offs), input_max_offset)));

                for(size_t m = 0; m < depth_multiplier; ++m)
                {
                    const auto weights_val = *(reinterpret_cast<T *>(weights_ptr + m * sizeof(T) + w * weights_stride_y));
                    acc.at(m)              = std::fma(weights_val, input_val, acc.at(m));
                }

                offs += dilation.x() * input_stride_y;
            }

            weights_ptr += weights_stride_z;
            input_offset += dilation.y() * input_stride_z;
        }

        if(has_biases)
        {
            for(size_t m = 0; m < depth_multiplier; ++m)
            {
                const auto biases_val                                     = *(reinterpret_cast<T *>(biases_it.ptr() + m * sizeof(T)));
                *(reinterpret_cast<T *>(output_it.ptr() + m * sizeof(T))) = acc.at(m) + biases_val;
            }
        }
        else
        {
            for(size_t m = 0; m < depth_multiplier; ++m)
            {
                *(reinterpret_cast<T *>(output_it.ptr() + m * sizeof(T))) = acc.at(m);
            }
        }
    },
    input_it, weights_it, biases_it, output_it);
}

Status validate_arguments(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *biases, const ITensorInfo *output, const PadStrideInfo &conv_info, unsigned int depth_multiplier,
                          const Size2D &dilation)
{
    ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(input, weights, output);
    ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(input, 1, DataType::F32);
    ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES(input, weights, output);
    ARM_COMPUTE_RETURN_ERROR_ON(depth_multiplier == 0);
    ARM_COMPUTE_RETURN_ERROR_ON((input->dimension(0) * depth_multiplier) != weights->dimension(0));
    ARM_COMPUTE_RETURN_ERROR_ON((dilation.x() < 1) || (dilation.y() < 1));
    ARM_COMPUTE_RETURN_ERROR_ON((conv_info.stride().first < 1) || (conv_info.stride().second < 1));

    if(biases != nullptr)
    {
        ARM_COMPUTE_RETURN_ERROR_ON(biases->num_dimensions() > 1);
        ARM_COMPUTE_RETURN_ERROR_ON(biases->dimension(0) != weights->dimension(0));
    }

    if(output->total_size() != 0)
    {
        const TensorShape output_shape = misc::shape_calculator::compute_depthwise_convolution_shape(*input, *weights, conv_info, depth_multiplier, dilation);
        ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DIMENSIONS(output->tensor_shape(), output_shape);
    }

    return Status{};
}
} // namespace

std::pair<Status, Window> validate_and_configure_window(ITensorInfo *input, ITensorInfo *weights, ITensorInfo *biases,
                                                        ITensorInfo *output, const PadStrideInfo &conv_info,
                                                        unsigned int depth_multiplier, const Size2D &dilation)
{
    // Get convolved dimensions
    const TensorShape output_shape = misc::shape_calculator::compute_depthwise_convolution_shape(*input, *weights, conv_info, depth_multiplier, dilation);

    // Output auto inizialitation if not yet initialized
    auto_init_if_empty(*output, input->clone()->set_is_resizable(true).reset_padding().set_tensor_shape(output_shape));

    // Configure kernel window (generic)
    const unsigned int num_elems_read_per_iteration    = (depth_multiplier == 1) ? 8 / element_size_from_data_type(input->data_type()) : 1;
    const unsigned int num_elems_written_per_iteration = num_elems_read_per_iteration * depth_multiplier;

    // Configure kernel window
    Window win = calculate_max_window(*output, Steps(num_elems_written_per_iteration));

    AccessWindowStatic input_access(input, 0, -conv_info.pad_left(), ceil_to_multiple(num_elems_read_per_iteration, input->dimension(0)),
                                    input->dimension(1) + std::max(std::max(conv_info.pad_right(), conv_info.pad_bottom()), conv_info.pad_top()));
    AccessWindowHorizontal weights_access(weights, 0, num_elems_written_per_iteration);
    AccessWindowHorizontal output_access(output, 0, num_elems_written_per_iteration);

    bool window_changed = update_window_and_padding(win, input_access, weights_access, output_access);

    if(biases != nullptr)
    {
        AccessWindowHorizontal biases_access(biases, 0, num_elems_written_per_iteration);
        window_changed |= update_window_and_padding(win, biases_access);
    }

    output_access.set_valid_region(win, ValidRegion(Coordinates(), output->tensor_shape()));

    Status err = (window_changed) ? ARM_COMPUTE_CREATE_ERROR(ErrorCode::RUNTIME_ERROR, "Insufficient Padding!") : Status{};
    return std::make_pair(err, win);
}

NEDepthwiseConvolutionLayerNativeKernel::NEDepthwiseConvolutionLayerNativeKernel()
    : _func(), _border_size(0), _input(), _weights(), _biases(), _output(), _conv_info(), _depth_multiplier(1), _dilation()
{
}

BorderSize NEDepthwiseConvolutionLayerNativeKernel::border_size() const
{
    return _border_size;
}

void NEDepthwiseConvolutionLayerNativeKernel::configure(const ITensor *input, const ITensor *weights, const ITensor *biases, ITensor *output,
                                                        const PadStrideInfo &conv_info, unsigned int depth_multiplier, const Size2D &dilation)
{
    ARM_COMPUTE_ERROR_ON_NULLPTR(input, weights, output);
    ARM_COMPUTE_ERROR_THROW_ON(validate_arguments(input->info(), weights->info(), (biases != nullptr) ? biases->info() : nullptr, output->info(), conv_info, depth_multiplier, dilation));

    _input            = input;
    _weights          = weights;
    _biases           = biases;
    _output           = output;
    _conv_info        = conv_info;
    _depth_multiplier = depth_multiplier;
    _border_size      = BorderSize(_conv_info.pad_left(), 0, std::max(std::max(conv_info.pad_right(), conv_info.pad_bottom()), conv_info.pad_top()), 0);
    _dilation         = dilation;

    switch(_input->info()->data_type())
    {
        case DataType::F32:
            _func = (biases != nullptr) ? &NEDepthwiseConvolutionLayerNativeKernel::run_depthwise<float, 2, true> : &NEDepthwiseConvolutionLayerNativeKernel::run_depthwise<float, 2, false>;
            break;
        default:
            ARM_COMPUTE_ERROR("Data type not supported");
            break;
    }

    auto win_config = validate_and_configure_window(_input->info(), _weights->info(), (biases != nullptr) ? biases->info() : nullptr, _output->info(), _conv_info, _depth_multiplier, dilation);
    ARM_COMPUTE_ERROR_THROW_ON(win_config.first);
    INEKernel::configure(win_config.second);
}

Status NEDepthwiseConvolutionLayerNativeKernel::validate(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *biases, const ITensorInfo *output, const PadStrideInfo &conv_info,
                                                         unsigned int  depth_multiplier,
                                                         const Size2D &dilation)
{
    ARM_COMPUTE_RETURN_ON_ERROR(validate_arguments(input, weights, biases, output, conv_info, depth_multiplier, dilation));
    ARM_COMPUTE_RETURN_ON_ERROR(validate_and_configure_window(input->clone().get(), weights->clone().get(), (biases != nullptr) ? biases->clone().get() : nullptr, output->clone().get(), conv_info,
                                                              depth_multiplier, dilation)
                                .first);
    return Status{};
}

void NEDepthwiseConvolutionLayerNativeKernel::run(const Window &window, const ThreadInfo &info)
{
    ARM_COMPUTE_UNUSED(info);
    ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(this);
    ARM_COMPUTE_ERROR_ON_INVALID_SUBWINDOW(INEKernel::window(), window);

    (this->*_func)(window);
}

template <typename T, int S, bool has_biases>
void NEDepthwiseConvolutionLayerNativeKernel::run_depthwise(const Window &window)
{
    ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(this);
    ARM_COMPUTE_ERROR_ON_INVALID_SUBWINDOW(INEKernel::window(), window);

    if(_depth_multiplier == 1)
    {
        depthwise_loop_multiplier1<T, S, has_biases>(_input, _weights, _biases, _output, _conv_info, _dilation, window);
    }
    else
    {
        depthwise_loop_generic<T, has_biases>(_input, _weights, _biases, _output, _conv_info, _dilation, _depth_multiplier, window);
    }
}
} // namespace arm_compute