From 484e7b3724c0e77751b5bed05180271fd5376e5d Mon Sep 17 00:00:00 2001 From: Moritz Pflanzer Date: Wed, 9 Aug 2017 11:43:18 +0100 Subject: COMPMID-417: Cleanup NEON FullyConnectedLayer Change-Id: Ie02a0a1a28ca2771e29a5e6552242caf0f6db1cf Reviewed-on: http://mpd-gerrit.cambridge.arm.com/83555 Tested-by: Kaizen Reviewed-by: Anthony Barbier --- .../NEON/functions/NEFullyConnectedLayer.cpp | 282 ++++++++------------- 1 file changed, 105 insertions(+), 177 deletions(-) (limited to 'src/runtime/NEON/functions/NEFullyConnectedLayer.cpp') diff --git a/src/runtime/NEON/functions/NEFullyConnectedLayer.cpp b/src/runtime/NEON/functions/NEFullyConnectedLayer.cpp index 4d9ee85f9b..39983bf643 100644 --- a/src/runtime/NEON/functions/NEFullyConnectedLayer.cpp +++ b/src/runtime/NEON/functions/NEFullyConnectedLayer.cpp @@ -30,8 +30,8 @@ #include #include -using namespace arm_compute; - +namespace arm_compute +{ NEFullyConnectedLayerReshapeWeights::NEFullyConnectedLayerReshapeWeights() : _transpose_kernel(), _transpose1xW_kernel(), _transpose_output(), _transpose_weights(false), _is_batched_fc_layer(false) { @@ -40,11 +40,11 @@ NEFullyConnectedLayerReshapeWeights::NEFullyConnectedLayerReshapeWeights() void NEFullyConnectedLayerReshapeWeights::configure(const ITensor *input, ITensor *output, bool transpose_weights, bool is_batched_fc_layer) { ARM_COMPUTE_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(input, 1, DataType::QS8, DataType::QS16, DataType::F16, DataType::F32); + ARM_COMPUTE_ERROR_ON(input->info()->num_dimensions() > 2); ARM_COMPUTE_ERROR_ON(output == nullptr); - ARM_COMPUTE_ERROR_ON(input->info()->num_dimensions() != 2); - ARM_COMPUTE_ERROR_ON((transpose_weights == false) && (is_batched_fc_layer == false)); + ARM_COMPUTE_ERROR_ON(!transpose_weights && !is_batched_fc_layer); - const DataType dt = input->info()->data_type(); + const DataType data_type = input->info()->data_type(); const int fixed_point_position = input->info()->fixed_point_position(); _transpose_weights = transpose_weights; @@ -57,7 +57,7 @@ void NEFullyConnectedLayerReshapeWeights::configure(const ITensor *input, ITenso { // Initialize the output tensor for transpose TensorShape shape_transposed(input->info()->dimension(1), input->info()->dimension(0)); - _transpose_output.allocator()->init(TensorInfo(shape_transposed, 1, dt, fixed_point_position)); + _transpose_output.allocator()->init(TensorInfo(shape_transposed, 1, data_type, fixed_point_position)); _transpose_kernel.configure(input, &_transpose_output); // Configure transpose 1xW kernel @@ -91,6 +91,7 @@ void NEFullyConnectedLayerReshapeWeights::run() { NEScheduler::get().schedule(&_transpose_kernel, Window::DimY); } + if(_is_batched_fc_layer) { NEScheduler::get().schedule(&_transpose1xW_kernel, Window::DimY); @@ -99,216 +100,142 @@ void NEFullyConnectedLayerReshapeWeights::run() NEFullyConnectedLayer::NEFullyConnectedLayer() : _im2col_kernel(), _reshape_weights_kernel(), _interleave4x4_kernel(), _mm_kernel(), _accumulate_biases_kernel(), _im2col_output(), _interleave4x4_output(), _reshape_weights_output(), - _are_weights_reshaped(false), _is_fc_after_conv(false), _is_batched_fc_layer(false), _accumulate_biases(false) + _are_weights_reshaped(false), _is_batched_fc_layer(false), _linearize_input(false), _accumulate_biases(false) { } -void NEFullyConnectedLayer::configure_conv_fc_wb(const ITensor *input, const ITensor *weights, ITensor *output) +void NEFullyConnectedLayer::configure(const ITensor *input, const ITensor *weights, const ITensor *biases, ITensor *output, bool transpose_weights, bool are_weights_reshaped) { - ARM_COMPUTE_ERROR_ON(weights->info()->dimension(0) != (input->info()->dimension(0) * input->info()->dimension(1) * input->info()->dimension(2) * (16 / weights->info()->element_size()))); - - const DataType dt = input->info()->data_type(); - const int fixed_point_position = input->info()->fixed_point_position(); + // With the Fully Connected layer we can have 4 different cases: + // 1) Convolution layer -> Fully Connected layer without batches + // 2) Fully Connected layer -> Fully Connected layer without batches + // 3) Convolution layer -> Fully Connected layer with batches + // 4) Fully Connected layer -> Fully Connected layer with batches - // If the fully connected layer is called after a convolution layer, the input tensor must be linearized + // Expected shape before transpose and reshaping + // Input: In x B (In and B can be multi-dimensional) + // Weights: flat(In) x Out + // Biases: Out + // Output: Out x B (B can be multi-dimensional) - // Initialize output tensor for im2col - TensorShape shape_im2col; - shape_im2col.set(0, input->info()->dimension(0) * input->info()->dimension(1) * input->info()->dimension(2)); - shape_im2col.set(1, input->info()->dimension(3)); - shape_im2col.set(2, input->info()->dimension(4)); - shape_im2col.set(3, input->info()->dimension(5)); - _im2col_output.allocator()->init(TensorInfo(shape_im2col, 1, dt, fixed_point_position)); + ARM_COMPUTE_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(input, 1, DataType::QS8, DataType::QS16, DataType::F16, DataType::F32); + ARM_COMPUTE_ERROR_ON_MISMATCHING_DATA_TYPES(input, weights, output); + ARM_COMPUTE_ERROR_ON_MISMATCHING_FIXED_POINT_POSITION(input, weights, output); - // Initialize output tensor for interleave 4x4 - TensorShape shape_interleaved = _im2col_output.info()->tensor_shape(); - shape_interleaved.set(0, shape_interleaved.x() * 4); - shape_interleaved.set(1, std::ceil(static_cast(shape_interleaved.y()) / 4)); - _interleave4x4_output.allocator()->init(TensorInfo(shape_interleaved, 1, dt, fixed_point_position)); + const DataType data_type = input->info()->data_type(); + const int fixed_point_position = input->info()->fixed_point_position(); + const int num_batch_dimensions = std::max(0, static_cast(output->info()->tensor_shape().num_dimensions()) - 1); + const int num_input_dimensions = input->info()->tensor_shape().num_dimensions() - num_batch_dimensions; + const size_t linear_input_size = input->info()->tensor_shape().total_size_lower(num_input_dimensions); - // Configure im2col kernel - _im2col_kernel.configure(input, &_im2col_output, Size2D(1, 1), PadStrideInfo(1, 1, 0, 0), false); + _linearize_input = input->info()->tensor_shape().x() != linear_input_size; + _are_weights_reshaped = are_weights_reshaped; + _accumulate_biases = biases != nullptr; + _is_batched_fc_layer = num_batch_dimensions > 0; - // Configure interleave4x4 kernel - _interleave4x4_kernel.configure(&_im2col_output, &_interleave4x4_output); + // Check if number of batches match + ARM_COMPUTE_ERROR_ON(input->info()->tensor_shape().total_size_upper(num_input_dimensions) != output->info()->tensor_shape().total_size_upper(1)); + ARM_COMPUTE_ERROR_ON(weights->info()->num_dimensions() > 2); - // Configure matrix multiply kernel - _mm_kernel.configure(&_interleave4x4_output, weights, output, 1.0f); + const size_t interleave_width = 16 / input->info()->element_size(); + const ITensor *weights_to_use = weights; - // Allocate the tensors once all the configure methods have been called - _im2col_output.allocator()->allocate(); - _interleave4x4_output.allocator()->allocate(); -} + if(!are_weights_reshaped && (transpose_weights || _is_batched_fc_layer)) + { + weights_to_use = &_reshape_weights_output; -void NEFullyConnectedLayer::configure_fc_fc_wb(const ITensor *input, const ITensor *weights, ITensor *output) -{ - const DataType dt = input->info()->data_type(); - const int fixed_point_position = input->info()->fixed_point_position(); + TensorShape reshaped_weights_shape(weights->info()->tensor_shape()); - // Initialize output tensor for interleave 4x4 - TensorShape shape_interleaved = input->info()->tensor_shape(); - shape_interleaved.set(0, shape_interleaved.x() * 4); - shape_interleaved.set(1, std::ceil(static_cast(shape_interleaved.y()) / 4)); - _interleave4x4_output.allocator()->init(TensorInfo(shape_interleaved, 1, dt, fixed_point_position)); + // Transpose weights if the user hasn't done it + if(transpose_weights) + { + const size_t shape_x = reshaped_weights_shape.x(); + reshaped_weights_shape.set(0, reshaped_weights_shape.y()); + reshaped_weights_shape.set(1, shape_x); + } - // Configure interleave4x4 kernel - _interleave4x4_kernel.configure(input, &_interleave4x4_output); + // If the we run multiple batches we need 1xW transpose, too. + if(_is_batched_fc_layer) + { + const float shape_x = reshaped_weights_shape.x(); + reshaped_weights_shape.set(0, reshaped_weights_shape.y() * interleave_width); + reshaped_weights_shape.set(1, static_cast(std::ceil(shape_x / interleave_width))); + } - // Configure matrix multiply kernel - _mm_kernel.configure(&_interleave4x4_output, weights, output, 1.0f); + _reshape_weights_output.allocator()->init(TensorInfo(reshaped_weights_shape, 1, data_type, fixed_point_position)); - // Allocate the tensors once all the configure methods have been called - _interleave4x4_output.allocator()->allocate(); -} + // Reshape the weights + _reshape_weights_kernel.configure(weights, &_reshape_weights_output, transpose_weights, _is_batched_fc_layer); + } -void NEFullyConnectedLayer::configure_conv_fc_nb(const ITensor *input, const ITensor *weights, ITensor *output) -{ - ARM_COMPUTE_ERROR_ON((weights->info()->dimension(1) != (input->info()->dimension(0) * input->info()->dimension(1) * input->info()->dimension(2)))); + // Check correct shape of weights + if(_is_batched_fc_layer) + { + // Transpose + Transpose1xW + ARM_COMPUTE_ERROR_ON(weights_to_use->info()->tensor_shape().x() != linear_input_size * interleave_width); + ARM_COMPUTE_ERROR_ON(weights_to_use->info()->tensor_shape().y() != static_cast(std::ceil(static_cast(output->info()->tensor_shape().x()) / interleave_width))); + } + else + { + // Transpose + ARM_COMPUTE_ERROR_ON(weights_to_use->info()->tensor_shape().x() != output->info()->tensor_shape().x()); + ARM_COMPUTE_ERROR_ON(weights_to_use->info()->tensor_shape().y() != linear_input_size); + } - const DataType dt = input->info()->data_type(); - const int fixed_point_position = input->info()->fixed_point_position(); + const ITensor *multiply_input = input; - // If the fully connected layer is called after a convolution layer, the input tensor must be linearized + if(_linearize_input) + { + TensorShape shape_im2col(input->info()->tensor_shape()); + shape_im2col.collapse(num_input_dimensions); + _im2col_output.allocator()->init(TensorInfo(shape_im2col, 1, data_type, fixed_point_position)); - // Initialize output tensor for im2col - TensorShape shape_im2col; - shape_im2col.set(0, input->info()->dimension(0) * input->info()->dimension(1) * input->info()->dimension(2)); - shape_im2col.set(1, 1); - _im2col_output.allocator()->init(TensorInfo(shape_im2col, 1, dt, fixed_point_position)); + // Configure im2col kernel + _im2col_kernel.configure(input, &_im2col_output, Size2D(1, 1), PadStrideInfo(1, 1, 0, 0), false); - // Configure im2col kernel - _im2col_kernel.configure(input, &_im2col_output, Size2D(1, 1), PadStrideInfo(1, 1, 0, 0), false); + multiply_input = &_im2col_output; + } - // Configure matrix multiply kernel - _mm_kernel.configure(&_im2col_output, weights, output, 1.0f); + if(_is_batched_fc_layer) + { + TensorShape shape_interleaved(multiply_input->info()->tensor_shape()); + shape_interleaved.set(0, shape_interleaved.x() * 4); + shape_interleaved.set(1, std::ceil(shape_interleaved.y() / 4.f)); + _interleave4x4_output.allocator()->init(TensorInfo(shape_interleaved, 1, data_type, fixed_point_position)); - // Allocate the output tensor for im2col once all the configure methods have been called - _im2col_output.allocator()->allocate(); -} + // Configure interleave4x4 kernel + _interleave4x4_kernel.configure(multiply_input, &_interleave4x4_output); -void NEFullyConnectedLayer::configure_fc_fc_nb(const ITensor *input, const ITensor *weights, ITensor *output) -{ - ARM_COMPUTE_ERROR_ON(input->info()->dimension(0) != weights->info()->dimension(1)); + multiply_input = &_interleave4x4_output; + } // Configure matrix multiply kernel - _mm_kernel.configure(input, weights, output, 1.0f); -} - -void NEFullyConnectedLayer::configure(const ITensor *input, const ITensor *weights, const ITensor *biases, ITensor *output, bool transpose_weights, bool are_weights_reshaped) -{ - ARM_COMPUTE_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(input, 1, DataType::QS8, DataType::QS16, DataType::F16, DataType::F32); - ARM_COMPUTE_ERROR_ON_MISMATCHING_DATA_TYPES(input, weights, output); - ARM_COMPUTE_ERROR_ON_MISMATCHING_FIXED_POINT_POSITION(input, weights, output); - ARM_COMPUTE_ERROR_ON(weights->info()->num_dimensions() != 2); - - const DataType dt = input->info()->data_type(); - const int fixed_point_position = input->info()->fixed_point_position(); - - _are_weights_reshaped = are_weights_reshaped; - _is_fc_after_conv = true; - _is_batched_fc_layer = false; - _accumulate_biases = false; + _mm_kernel.configure(multiply_input, weights_to_use, output, 1.0f); - if(biases != nullptr) + if(_accumulate_biases) { ARM_COMPUTE_ERROR_ON_MISMATCHING_DATA_TYPES(input, biases); - - _accumulate_biases = true; + ARM_COMPUTE_ERROR_ON(biases->info()->tensor_shape().x() != output->info()->tensor_shape().x()); // Configure accumulate biases kernel _accumulate_biases_kernel.configure(output, biases); } - // With the Fully Connected layer we can have 4 different cases: - // 1) Convolution layer -> Fully Connected layer without batches - // 2) Fully Connected layer -> Fully Connected layer without batches - // 3) Convolution layer -> Fully Connected layer with batches - // 4) Fully Connected layer -> Fully Connected layer with batches - - // Check if we have a fully connected layer with batches - _is_batched_fc_layer = (output->info()->dimension(1) > 1); - - const ITensor *weights_to_use = weights; - - if(!are_weights_reshaped) + // Allocate the transpose tensor if the are_weights_reshaped flag is false and once all the configure methods have been called + if(!are_weights_reshaped && (transpose_weights || _is_batched_fc_layer)) { - if((transpose_weights || _is_batched_fc_layer)) - { - weights_to_use = &_reshape_weights_output; - - if(transpose_weights) - { - if(_is_batched_fc_layer) - { - const float transpose_width = 16.0f / input->info()->element_size(); - TensorShape shape_wt(weights->info()->dimension(0) * static_cast(transpose_width), static_cast(std::ceil(weights->info()->dimension(1) / transpose_width))); - TensorInfo info_wt(shape_wt, 1, dt, fixed_point_position); - _reshape_weights_output.allocator()->init(info_wt); - } - else - { - TensorShape shape_wt(weights->info()->dimension(1), weights->info()->dimension(0)); - TensorInfo info_wt(shape_wt, 1, dt, fixed_point_position); - _reshape_weights_output.allocator()->init(info_wt); - } - } - else - { - ARM_COMPUTE_ERROR_ON(!_is_batched_fc_layer); - - const float transpose_width = 16.0f / input->info()->element_size(); - TensorShape shape_wt(weights->info()->dimension(1) * static_cast(transpose_width), static_cast(std::ceil(weights->info()->dimension(0) / transpose_width))); - TensorInfo info_wt(shape_wt, 1, dt, fixed_point_position); - _reshape_weights_output.allocator()->init(info_wt); - } - - // Reshape the weights - _reshape_weights_kernel.configure(weights, &_reshape_weights_output, transpose_weights, _is_batched_fc_layer); - } + // Allocate the tensor for the weights reshaped + _reshape_weights_output.allocator()->allocate(); } - if(_is_batched_fc_layer) - { - _is_fc_after_conv = (TensorShape::num_max_dimensions >= 4) && (std::equal(input->info()->tensor_shape().cbegin() + 3, - input->info()->tensor_shape().cend(), - output->info()->tensor_shape().cbegin() + 1)); - - if(_is_fc_after_conv) - { - // Fully Connected layer after a Convolution Layer with batches - configure_conv_fc_wb(input, weights_to_use, output); - } - else - { - // Fully Connected layer after a Fully Connected Layer with batches - configure_fc_fc_wb(input, weights_to_use, output); - } - } - else + if(_linearize_input) { - // In case of not batched fully connected layer, the weights will not be reshaped using transposed1xW - _is_fc_after_conv = ((weights_to_use->info()->dimension(1)) == (input->info()->dimension(0) * input->info()->dimension(1) * input->info()->dimension(2))); - - if(_is_fc_after_conv) - { - // Fully Connected layer after a Convolution Layer without batches - configure_conv_fc_nb(input, weights_to_use, output); - } - else - { - // Fully Connected layer after a Fully Connected Layer without batches - configure_fc_fc_nb(input, weights_to_use, output); - } + _im2col_output.allocator()->allocate(); } - // Allocate the transpose tensor if the are_weights_reshaped flag is false and once all the configure methods have been called - if(!are_weights_reshaped) + if(_is_batched_fc_layer) { - if(transpose_weights || _is_batched_fc_layer) - { - // Allocate the tensor for the weights reshaped - _reshape_weights_output.allocator()->allocate(); - } + _interleave4x4_output.allocator()->allocate(); } } @@ -321,8 +248,8 @@ void NEFullyConnectedLayer::run() _reshape_weights_kernel.run(); } - // Linearize input if comes from a convolutional layer - if(_is_fc_after_conv) + // Linearize input if it comes from a convolutional layer + if(_linearize_input) { NEScheduler::get().schedule(&_im2col_kernel, Window::DimY); } @@ -342,3 +269,4 @@ void NEFullyConnectedLayer::run() NEScheduler::get().schedule(&_accumulate_biases_kernel, Window::DimY); } } +} // namespace arm_compute -- cgit v1.2.1