diff options
Diffstat (limited to 'runtime/neurun/backend/acl_cl/ConstantInitializer.cc')
| -rw-r--r-- | runtime/neurun/backend/acl_cl/ConstantInitializer.cc | 266 |
1 files changed, 266 insertions, 0 deletions
diff --git a/runtime/neurun/backend/acl_cl/ConstantInitializer.cc b/runtime/neurun/backend/acl_cl/ConstantInitializer.cc new file mode 100644 index 000000000..165b17cd1 --- /dev/null +++ b/runtime/neurun/backend/acl_cl/ConstantInitializer.cc @@ -0,0 +1,266 @@ +/* + * Copyright (c) 2019 Samsung Electronics Co., Ltd. All Rights Reserved + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include "ConstantInitializer.h" + +namespace neurun +{ +namespace backend +{ +namespace acl_cl +{ + +ConstantInitializer::ConstantInitializer(const ir::Operands &operands, + const std::shared_ptr<TensorBuilder> &tensor_builder) + : _operands{operands}, _tensor_builder{tensor_builder} +{ + // DO NOTHING +} + +void ConstantInitializer::visit(const ir::operation::BatchToSpaceND &node) +{ + const auto &block_size_index = node.getInputs().at(ir::operation::BatchToSpaceND::BLOCK_SIZE); + const auto &block_size_obj = _operands.at(block_size_index); + + if (block_size_obj.isConstant()) + { + _init_map[block_size_index] = [](const ir::Operand &model_obj, backend::operand::ITensor &obj) { + const auto &shape = model_obj.shape(); + const auto base = reinterpret_cast<const int32_t *>(model_obj.data().base()); + assert(model_obj.shape().rank() == 1); + obj.access([&](::neurun::backend::operand::ITensor &tensor) { + for (size_t i = 0; i < shape.num_elements(); ++i) + { + const int32_t value = base[shape.num_elements() - i - 1]; + int32_t *into = reinterpret_cast<int32_t *>(tensor.buffer() + + tensor.calcOffset({static_cast<int32_t>(i)})); + *into = value; + } + }); + }; + } +} + +void ConstantInitializer::visit(const ir::operation::Conv2D &node) +{ + const auto &kernel_index = node.getInputs().at(ir::operation::Conv2D::KERNEL); + const auto &kernel_obj = _operands.at(kernel_index); + registerPermuteInitializer(kernel_index, kernel_obj); + + const auto &bias_index = node.getInputs().at(ir::operation::Conv2D::BIAS); + const auto &bias_obj = _operands.at(bias_index); + registerCopyInitializer(bias_index, bias_obj); +} + +void ConstantInitializer::visit(const ir::operation::DepthwiseConv2D &node) +{ + const auto &kernel_index = node.getInputs().at(ir::operation::DepthwiseConv2D::KERNEL); + const auto &kernel_obj = _operands.at(kernel_index); + registerPermuteInitializer(kernel_index, kernel_obj); + + const auto &bias_index = node.getInputs().at(ir::operation::DepthwiseConv2D::BIAS); + const auto &bias_obj = _operands.at(bias_index); + registerCopyInitializer(bias_index, bias_obj); +} + +void ConstantInitializer::visit(const ir::operation::EmbeddingLookup &node) +{ + const auto &lookups_index = node.getInputs().at(ir::operation::EmbeddingLookup::LOOKUPS); + const auto &lookups_obj = _operands.at(lookups_index); + registerCopyInitializer(lookups_index, lookups_obj); +} + +void ConstantInitializer::visit(const ir::operation::FullyConnected &node) +{ + const auto &weight_index = node.getInputs().at(ir::operation::FullyConnected::WEIGHT); + const auto &weight_obj = _operands.at(weight_index); + registerCopyInitializer(weight_index, weight_obj); + + const auto &bias_index = node.getInputs().at(ir::operation::FullyConnected::BIAS); + const auto &bias_obj = _operands.at(bias_index); + registerCopyInitializer(bias_index, bias_obj); +} + +void ConstantInitializer::visit(const ir::operation::Gather &node) +{ + const auto &indices_index = node.getInputs().at(ir::operation::Gather::INDICES); + const auto &indices_obj = _operands.at(indices_index); + registerCopyInitializer(indices_index, indices_obj); +} + +void ConstantInitializer::visit(const ir::operation::HashtableLookup &node) +{ + const auto &lookups_index = node.getInputs().at(ir::operation::HashtableLookup::LOOKUPS); + const auto &lookups_obj = _operands.at(lookups_index); + registerCopyInitializer(lookups_index, lookups_obj); + + const auto &keys_index = node.getInputs().at(ir::operation::HashtableLookup::KEYS); + const auto &keys_obj = _operands.at(keys_index); + registerCopyInitializer(keys_index, keys_obj); +} + +void ConstantInitializer::visit(const ir::operation::LSTM &node) +{ + const auto &input_to_input_weights_index = + node.getInputs().at(ir::operation::LSTM::INPUT_TO_INPUT_WEIGHTS); + const auto &input_to_input_weights_obj = _operands.at(input_to_input_weights_index); + registerCopyInitializer(input_to_input_weights_index, input_to_input_weights_obj); + + const auto &input_to_forget_weights_index = + node.getInputs().at(ir::operation::LSTM::INPUT_TO_FORGET_WEIGHTS); + const auto &input_to_forget_weights_obj = _operands.at(input_to_forget_weights_index); + registerCopyInitializer(input_to_forget_weights_index, input_to_forget_weights_obj); + + const auto &input_to_cell_weights_index = + node.getInputs().at(ir::operation::LSTM::INPUT_TO_CELL_WEIGHTS); + const auto &input_to_cell_weights_obj = _operands.at(input_to_cell_weights_index); + registerCopyInitializer(input_to_cell_weights_index, input_to_cell_weights_obj); + + const auto &input_to_output_weights_index = + node.getInputs().at(ir::operation::LSTM::INPUT_TO_OUTPUT_WEIGHTS); + const auto &input_to_output_weights_obj = _operands.at(input_to_output_weights_index); + registerCopyInitializer(input_to_output_weights_index, input_to_output_weights_obj); + + const auto &recurrent_to_input_weights_index = + node.getInputs().at(ir::operation::LSTM::RECURRENT_TO_INPUT_WEIGHTS); + const auto &recurrent_to_input_weights_obj = _operands.at(recurrent_to_input_weights_index); + registerCopyInitializer(recurrent_to_input_weights_index, recurrent_to_input_weights_obj); + + const auto &recurrent_to_forget_weights_index = + node.getInputs().at(ir::operation::LSTM::RECURRENT_TO_FORGET_WEIGHTS); + const auto &recurrent_to_forget_weights_obj = _operands.at(recurrent_to_forget_weights_index); + registerCopyInitializer(recurrent_to_forget_weights_index, recurrent_to_forget_weights_obj); + + const auto &recurrent_to_cell_weights_index = + node.getInputs().at(ir::operation::LSTM::RECURRENT_TO_CELL_WEIGHTS); + const auto &recurrent_to_cell_weights_obj = _operands.at(recurrent_to_cell_weights_index); + registerCopyInitializer(recurrent_to_cell_weights_index, recurrent_to_cell_weights_obj); + + const auto &recurrent_to_output_weights_index = + node.getInputs().at(ir::operation::LSTM::RECURRENT_TO_OUTPUT_WEIGHTS); + const auto &recurrent_to_output_weights_obj = _operands.at(recurrent_to_output_weights_index); + registerCopyInitializer(recurrent_to_output_weights_index, recurrent_to_output_weights_obj); + + const auto &cell_to_input_weights_index = + node.getInputs().at(ir::operation::LSTM::CELL_TO_INPUT_WEIGHTS); + const auto &cell_to_input_weights_obj = _operands.at(cell_to_input_weights_index); + registerCopyInitializer(cell_to_input_weights_index, cell_to_input_weights_obj); + + const auto &cell_to_forget_weights_index = + node.getInputs().at(ir::operation::LSTM::CELL_TO_FORGET_WEIGHTS); + const auto &cell_to_forget_weights_obj = _operands.at(cell_to_forget_weights_index); + registerCopyInitializer(cell_to_forget_weights_index, cell_to_forget_weights_obj); + + const auto &cell_to_output_weights_index = + node.getInputs().at(ir::operation::LSTM::CELL_TO_OUTPUT_WEIGHTS); + const auto &cell_to_output_weights_obj = _operands.at(cell_to_output_weights_index); + registerCopyInitializer(cell_to_output_weights_index, cell_to_output_weights_obj); + + const auto &input_gate_bias_index = node.getInputs().at(ir::operation::LSTM::INPUT_GATE_BIAS); + const auto &input_gate_bias_obj = _operands.at(input_gate_bias_index); + registerCopyInitializer(input_gate_bias_index, input_gate_bias_obj); + + const auto &forget_gate_bias_index = node.getInputs().at(ir::operation::LSTM::FORGET_GATE_BIAS); + const auto &forget_gate_bias_obj = _operands.at(forget_gate_bias_index); + registerCopyInitializer(forget_gate_bias_index, forget_gate_bias_obj); + + const auto &output_gate_bias_index = node.getInputs().at(ir::operation::LSTM::OUTPUT_GATE_BIAS); + const auto &output_gate_bias_obj = _operands.at(output_gate_bias_index); + registerCopyInitializer(output_gate_bias_index, output_gate_bias_obj); + + const auto &projection_weights_index = + node.getInputs().at(ir::operation::LSTM::PROJECTION_WEIGHTS); + const auto &projection_weights_obj = _operands.at(projection_weights_index); + registerCopyInitializer(projection_weights_index, projection_weights_obj); + + const auto &projection_bias_index = node.getInputs().at(ir::operation::LSTM::PROJECTION_BIAS); + const auto &projection_bias_obj = _operands.at(projection_bias_index); + registerCopyInitializer(projection_bias_index, projection_bias_obj); +} + +void ConstantInitializer::visit(const ir::operation::RNN &node) +{ + const auto &weights_index = node.getInputs().at(ir::operation::RNN::WEIGHTS); + const auto &weights_obj = _operands.at(weights_index); + registerCopyInitializer(weights_index, weights_obj); + + const auto &recurrent_weights_index = node.getInputs().at(ir::operation::RNN::RECURRENT_WEIGHTS); + const auto &recurrent_weights_obj = _operands.at(recurrent_weights_index); + registerCopyInitializer(recurrent_weights_index, recurrent_weights_obj); + + const auto &bias_index = node.getInputs().at(ir::operation::RNN::BIAS); + const auto &bias_obj = _operands.at(bias_index); + registerCopyInitializer(bias_index, bias_obj); +} + +void ConstantInitializer::visit(const ir::operation::SpaceToBatchND &node) +{ + const auto &block_size_index = node.getInputs().at(ir::operation::SpaceToBatchND::BLOCK_SIZE); + const auto &block_size_obj = _operands.at(block_size_index); + + if (block_size_obj.isConstant()) + { + _init_map[block_size_index] = [](const ir::Operand &model_obj, backend::operand::ITensor &obj) { + const auto &shape = model_obj.shape(); + const auto base = reinterpret_cast<const int32_t *>(model_obj.data().base()); + assert(model_obj.shape().rank() == 1); + obj.access([&](::neurun::backend::operand::ITensor &tensor) { + for (size_t i = 0; i < shape.num_elements(); ++i) + { + const int32_t value = base[shape.num_elements() - i - 1]; + int32_t *into = reinterpret_cast<int32_t *>(tensor.buffer() + + tensor.calcOffset({static_cast<int32_t>(i)})); + *into = value; + } + }); + }; + } + + const auto &paddings_index = node.getInputs().at(ir::operation::SpaceToBatchND::PADDINGS); + const auto &paddings_obj = _operands.at(paddings_index); + if (paddings_obj.isConstant()) + { + _init_map[paddings_index] = [](const ir::Operand &model_obj, backend::operand::ITensor &obj) { + const auto &shape = model_obj.shape(); + const auto base = reinterpret_cast<const int32_t *>(model_obj.data().base()); + assert(model_obj.shape().rank() == 2); + assert(obj.dimension(0) == 2); + obj.access([&](::neurun::backend::operand::ITensor &tensor) { + for (auto i = 0; i < shape.dim(0); ++i) + { + for (auto j = 0; j < shape.dim(1); ++j) + { + const int32_t value = base[i * 2 + j]; + int32_t *into = reinterpret_cast<int32_t *>( + tensor.buffer() + tensor.calcOffset({shape.dim(0) - i - 1, j})); + *into = value; + } + } + }); + }; + } +} + +void ConstantInitializer::visit(const ir::operation::TransposeConv &node) +{ + const auto &kernel_index = node.getInputs().at(ir::operation::TransposeConv::KERNEL); + const auto &kernel_obj = _operands.at(kernel_index); + registerPermuteInitializer(kernel_index, kernel_obj); +} + +} // namespace acl_cl +} // namespace backend +} // namespace neurun |