From c42f28d45e9b990276d54880d2cee9c9ee675a41 Mon Sep 17 00:00:00 2001 From: Giorgio Arena Date: Thu, 26 Apr 2018 11:33:05 +0100 Subject: COMPMID-1048 Add NHWC data format support to Winograd input transform 4x4_3x3 https://confluence.arm.com/display/MLENG/Winograd+Input+Transform%3A+NCHW+vs+NHWC+on+OpenCL Change-Id: Iac35a54389266701b7d8f5434a7a37df85b7b187 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/133315 Reviewed-by: Gian Marco Iodice Tested-by: Jenkins --- .../CL/kernels/CLWinogradInputTransformKernel.cpp | 70 ++++++++++++++++------ 1 file changed, 53 insertions(+), 17 deletions(-) (limited to 'src/core/CL/kernels/CLWinogradInputTransformKernel.cpp') diff --git a/src/core/CL/kernels/CLWinogradInputTransformKernel.cpp b/src/core/CL/kernels/CLWinogradInputTransformKernel.cpp index febd22b04e..e73ac7df76 100644 --- a/src/core/CL/kernels/CLWinogradInputTransformKernel.cpp +++ b/src/core/CL/kernels/CLWinogradInputTransformKernel.cpp @@ -40,13 +40,13 @@ namespace Status validate_arguments(const ITensorInfo *input, const ITensorInfo *output, const WinogradInfo &winograd_info) { ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(input, 1, DataType::F32); - ARM_COMPUTE_RETURN_ERROR_ON(input->data_layout() != DataLayout::NCHW); const PadStrideInfo conv_info = winograd_info.convolution_info; const Size2D output_tile_size = winograd_info.output_tile_size; const Size2D kernel_size = winograd_info.kernel_size; ARM_COMPUTE_RETURN_ERROR_ON_MSG(conv_info.stride().first != 1 || conv_info.stride().second != 1, "Winograd input transform only supports unit strides"); ARM_COMPUTE_RETURN_ERROR_ON_MSG(kernel_size != Size2D(3U, 3U) && kernel_size != Size2D(5U, 5U), "Winograd input transform only supports 3x3 and 5x5 kernels"); + ARM_COMPUTE_RETURN_ERROR_ON(input->data_layout() == DataLayout::NHWC && (output_tile_size != Size2D(4U, 4U) || kernel_size != Size2D(3U, 3U))); ARM_COMPUTE_RETURN_ERROR_ON_MSG(kernel_size == Size2D(3U, 3U) && output_tile_size != Size2D(2U, 2U) && output_tile_size != Size2D(4U, 4U), "Winograd input transform only supports 2x2 or 4x4 output tile for 3x3 kernels"); @@ -75,12 +75,28 @@ std::pair validate_and_configure_window(ITensorInfo *input, ITen const Size2D output_tile_size = winograd_info.output_tile_size; const Size2D kernel_size = winograd_info.kernel_size; - const unsigned int num_elems_read_per_iteration_x = output_tile_size.width + kernel_size.width - 1; - const unsigned int num_elems_read_per_iteration_y = output_tile_size.height + kernel_size.height - 1; + unsigned int num_elems_read_per_iteration_x = 0; + unsigned int num_elems_read_per_iteration_y = 0; + unsigned int pad_left = 0; + unsigned int pad_top = 0; + + if(input->data_layout() == DataLayout::NCHW) + { + num_elems_read_per_iteration_x = output_tile_size.width + kernel_size.width - 1; + num_elems_read_per_iteration_y = output_tile_size.height + kernel_size.height - 1; + pad_left = conv_info.pad_left(); + pad_top = conv_info.pad_top(); + } + else + { + num_elems_read_per_iteration_x = 1; + num_elems_read_per_iteration_y = output_tile_size.width + kernel_size.width - 1; + pad_top = 1; + } Window win = calculate_max_window(*input, Steps(1, 1)); - AccessWindowRectangle input_access(input, -conv_info.pad_left(), -conv_info.pad_top(), num_elems_read_per_iteration_x, num_elems_read_per_iteration_y); + AccessWindowRectangle input_access(input, -pad_left, -pad_top, num_elems_read_per_iteration_x, num_elems_read_per_iteration_y); bool window_changed = update_window_and_padding(win, input_access); @@ -108,17 +124,27 @@ void CLWinogradInputTransformKernel::configure(const ICLTensor *input, ICLTensor const Size2D output_tile_size = winograd_info.output_tile_size; const Size2D kernel_size = winograd_info.kernel_size; + const size_t idx_w = get_data_layout_dimension_index(input->info()->data_layout(), DataLayoutDimension::WIDTH); + const size_t idx_h = get_data_layout_dimension_index(input->info()->data_layout(), DataLayoutDimension::HEIGHT); + // Compute number of elements to process in the X and Y direction - const int num_elements_x = input->info()->dimension(0) - (kernel_size.width - 1) + conv_info.pad_left() + conv_info.pad_right(); - const int num_elements_y = input->info()->dimension(1) - (kernel_size.height - 1) + conv_info.pad_top() + conv_info.pad_bottom(); + const int num_elements_x = input->info()->dimension(idx_w) - (kernel_size.width - 1) + conv_info.pad_left() + conv_info.pad_right(); + const int num_elements_y = input->info()->dimension(idx_h) - (kernel_size.height - 1) + conv_info.pad_top() + conv_info.pad_bottom(); - // Check if we need to extend the right or bottom border - const unsigned int extra_border_right = ((num_elements_x % output_tile_size.width) == 0) ? 0u : static_cast(output_tile_size.width - 1); - const unsigned int extra_border_bottom = ((num_elements_y % output_tile_size.height) == 0) ? 0u : static_cast(output_tile_size.height - 1); + _input = input; + _output = output; + if(input->info()->data_layout() == DataLayout::NCHW) + { + // Check if we need to extend the right or bottom border + const unsigned int extra_border_right = ((num_elements_x % output_tile_size.width) == 0) ? 0u : static_cast(output_tile_size.width - 1); + const unsigned int extra_border_bottom = ((num_elements_y % output_tile_size.height) == 0) ? 0u : static_cast(output_tile_size.height - 1); - _input = input; - _output = output; - _border_size = BorderSize(conv_info.pad_top(), conv_info.pad_right() + extra_border_right, conv_info.pad_bottom() + extra_border_bottom, conv_info.pad_left()); + _border_size = BorderSize(conv_info.pad_top(), conv_info.pad_right() + extra_border_right, conv_info.pad_bottom() + extra_border_bottom, conv_info.pad_left()); + } + else + { + _border_size = BorderSize(1U, 0U, 1U, 0); + } _num_tiles_x = std::ceil(num_elements_x / static_cast(output_tile_size.width)); _num_tiles_y = std::ceil(num_elements_y / static_cast(output_tile_size.height)); @@ -134,6 +160,12 @@ void CLWinogradInputTransformKernel::configure(const ICLTensor *input, ICLTensor build_opts.add_option("-DPAD_LEFT=" + support::cpp11::to_string(conv_info.pad_left())); build_opts.add_option("-DPAD_TOP=" + support::cpp11::to_string(conv_info.pad_top())); + if(input->info()->data_layout() == DataLayout::NHWC) + { + build_opts.add_option("-DSRC_DIM_1=" + support::cpp11::to_string(_input->info()->dimension(1))); + build_opts.add_option("-DSRC_DIM_2=" + support::cpp11::to_string(_input->info()->dimension(2))); + } + // Create kernel std::string kernel_name = "winograd_input_transform_" + output_tile_size.to_string() + "_" + kernel_size.to_string(); @@ -148,7 +180,7 @@ void CLWinogradInputTransformKernel::configure(const ICLTensor *input, ICLTensor // Append stepz and data layout kernel_name += "_stepz"; kernel_name += support::cpp11::to_string(_step_z); - kernel_name += "_nchw"; + kernel_name += "_" + lower_string(string_from_data_layout(input->info()->data_layout())); _kernel = static_cast(CLKernelLibrary::get().create_kernel(kernel_name, build_opts.options())); @@ -183,12 +215,16 @@ void CLWinogradInputTransformKernel::run(const Window &window, cl::CommandQueue ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(this); ARM_COMPUTE_ERROR_ON_INVALID_SUBWINDOW(IKernel::window(), window); + const size_t idx_w = get_data_layout_dimension_index(_input->info()->data_layout(), DataLayoutDimension::WIDTH); + const size_t idx_h = get_data_layout_dimension_index(_input->info()->data_layout(), DataLayoutDimension::HEIGHT); + const size_t idx_c = get_data_layout_dimension_index(_input->info()->data_layout(), DataLayoutDimension::CHANNEL); + Window slice = window.first_slice_window_3D(); - slice.set(Window::DimX, Window::Dimension(0, _num_tiles_x, 1)); - slice.set(Window::DimY, Window::Dimension(0, _num_tiles_y, 1)); + slice.set(idx_w, Window::Dimension(0, _num_tiles_x, 1)); + slice.set(idx_h, Window::Dimension(0, _num_tiles_y, 1)); - ARM_COMPUTE_ERROR_ON(((slice.z().end() - slice.z().start()) % _step_z) != 0); - slice.set(Window::DimZ, Window::Dimension(slice.z().start(), slice.z().end(), _step_z)); + ARM_COMPUTE_ERROR_ON(((slice[idx_c].end() - slice[idx_c].start()) % _step_z) != 0); + slice.set(idx_c, Window::Dimension(slice[idx_c].start(), slice[idx_c].end(), _step_z)); do { -- cgit v1.2.1