# Copyright (c) 2019 PaddlePaddle Authors. 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. import numpy as np from .... import core from ....framework import IrGraph from ....framework import IrNode __all__ = ['TransformForMkldnnPass'] class TransformForMkldnnPass(object): """ Convert QuantizationFreezePass generated IrGraph to MKL-DNN supported INT8 IrGraph. Following transformations did in this pass: 1. Convert int8 range weights with float32 data type, which are generated by the QuantizationFreezePass, to float32 range weights with float32 data type by using the corresponding scales. This conversion is because MKL-DNN INT8 conv2d kernel now only supports float32 weights input, will do weights quantization inside the conv2d kernel. 2. Create the new conv2d op with the converted weights and link its output to fake_dequantize_abs_max op's output and set conv2d's attribute "force_fp32 _output" as true 3. Transform fake_quantize_xx op to quantize op 4. Remove fake_dequantize_abs_max op """ def __init__(self, scope=None, place=None): """ Args: scope(fluid.Scope): scope is used to initialize the new parameters. place(fluid.CPUPlace): place is used to initialize the new parameters. Examples: .. code-block:: python # The original graph will be rewrite. import paddle.fluid as fluid from paddle.fluid.contrib.slim.quantization \ import TransformForMkldnnPass from paddle.fluid.framework import IrGraph from paddle.fluid import core graph = IrGraph(core.Graph(fluid.Program().desc), for_test=False) place = fluid.CPUPlace() mkldnn_pass = TransformForMkldnnPass(fluid.global_scope(), place) mkldnn_pass.apply(graph) """ self._scope = scope self._place = place self.quantize_type = [ 'fake_quantize_moving_average_abs_max', 'fake_quantize_range_abs_max' ] self.dequantize_type = ['fake_dequantize_max_abs'] self._quantizable_ops = ['conv2d', 'depthwise_conv2d', 'mul'] self._conv_ops = ['conv2d', 'depthwise_conv2d'] self.InScale = {} self.max_range = {} self.conv_new_output = {} self.s8_max = 127 # Temporary code for keeping the mul op as fake quantization #TODO Intel: Remove the following code when mul int8 mkldnn # kernel enabled self.mul_input_id = [] self.mul_output_id = [] def apply(self, graph): """ Quantize the graph for running MKL-DNN INT8 inference. According to activation quantization type, the graph will transform fake quantize ops to quantize ops and remove the fake dequantize ops. Args: graph(IrGraph): the applied graph. """ assert isinstance(graph, IrGraph), 'graph must be the instance of IrGraph.' ops = graph.all_op_nodes() persistable_vars = [p.name() for p in graph.all_persistable_nodes()] # Collect the InScales and max_range to calculate the new scales for MKL-DNN # INT8 conv2d for op_node in ops: if op_node.name() in self.dequantize_type: input_name = op_node.input("X")[0] scale_name = op_node.input("Scale")[0] self.InScale[input_name] = self._load_param(self._scope, scale_name)[0] self.max_range[input_name] = op_node.op().attr("max_range") self.conv_new_output[input_name] = op_node.output("Out")[0] # Temporary graph transform on keeping the mul op # TODO Intel: Remove following code elif op_node.name() in ['mul']: input_node = graph._find_node_by_name(op_node.inputs, op_node.input('X')[0]) output_node = graph._find_node_by_name(op_node.outputs, op_node.output('Out')[0]) self.mul_input_id.append(input_node.id()) self.mul_output_id.append(output_node.id()) for op_node in ops: if op_node.name() in self._conv_ops: self._transform_to_conv_mkldnn(graph, op_node) elif op_node.name() in self.quantize_type: self._transform_to_quantize_mkldnn(graph, op_node) elif op_node.name() in self.dequantize_type: self._remove_fake_dequantize_op(graph, op_node) self._remove_unused_var_nodes(graph) return graph def _transform_to_conv_mkldnn(self, graph, op_node): weight_name = op_node.input("Filter")[0] output_name = op_node.output("Output")[0] # Convert int8 range weights to fp32 range weights weight = self._load_param(self._scope, weight_name) w_fp32 = np.divide( np.multiply(weight, 127), self.max_range[output_name]) w_fp32 = w_fp32.reshape(weight.shape) self._restore_var(weight_name, w_fp32) input_var_node = graph._find_node_by_name(op_node.inputs, op_node.input("Input")[0]) weight_var_node = graph._find_node_by_name(op_node.inputs, weight_name) # Set fake_dequantize_abs_max's output as new output of conv2d output_var_node = graph._find_node_by_name( graph.all_var_nodes(), self.conv_new_output[output_name]) attrs = { name: op_node.op().attr(name) for name in op_node.op().attr_names() } conv_op_node = graph.create_op_node( op_type='conv2d', attrs=attrs, inputs={'Input': input_var_node, 'Filter': weight_var_node}, outputs={'Output': output_var_node}) # Based on the QAT's scales to calculate the scales of MKL-DNN INT8 conv2d scale_in = self.s8_max / self.InScale[output_name] scale_w = [] scale_w.append(self.max_range[output_name] / self.s8_max) conv_op_node.set_attr("Scale_weights", scale_w) conv_op_node.set_attr("Scale_in", scale_in) conv_op_node.set_attr("Scale_out", 1.0) conv_op_node.set_attr("use_mkldnn", 1) conv_op_node.set_attr("force_fp32_output", 1) graph.link_to(input_var_node, conv_op_node) graph.link_to(weight_var_node, conv_op_node) graph.link_to(conv_op_node, output_var_node) graph.safe_remove_nodes(op_node) def _transform_to_quantize_mkldnn(self, graph, op_node): """ Transform fake_quantize_xx op to quantize mkldnn op in the graph. """ input_var_node = graph._find_node_by_name(op_node.inputs, op_node.input("X")[0]) output_var_node = graph._find_node_by_name(op_node.outputs, op_node.output("Out")[0]) if output_var_node.id() in self.mul_input_id: return else: scale_in = self.s8_max / self._load_param( self._scope, op_node.input("InScale")[0])[0] quant_op_node = graph.create_op_node( op_type='quantize', attrs={ 'data_format': 'MKLDNNLAYOUT', 'use_mkldnn': 1, 'Scale': scale_in, 'is_negative_input': 1 }, inputs={'Input': input_var_node}, outputs={'Output': output_var_node}) graph.link_to(input_var_node, quant_op_node) graph.link_to(quant_op_node, output_var_node) graph.safe_remove_nodes(op_node) def _remove_fake_dequantize_op(self, graph, op_node): input_var_node = graph._find_node_by_name(op_node.inputs, op_node.input("X")[0]) if input_var_node.id() in self.mul_output_id: return else: graph.safe_remove_nodes(op_node) def _load_param(self, scope, param_name): return np.array(scope.find_var(param_name).get_tensor()) def _restore_var(self, name, array): tensor = self._scope.find_var(name).get_tensor() tensor.set(array, self._place) def _remove_unused_var_nodes(self, graph): all_used_vars = set() ops = graph.all_op_nodes() for op_node in ops: for input_node in op_node.inputs: all_used_vars.add(input_node) for output_node in op_node.outputs: all_used_vars.add(output_node) all_used_vars = {n.node for n in all_used_vars} all_unused_vars = { n for n in filter(lambda node: node.node not in all_used_vars, graph.all_var_nodes()) } graph.safe_remove_nodes(all_unused_vars)