1 /*
2  * Copyright (c) 2019 Arm Limited.
3  *
4  * SPDX-License-Identifier: MIT
5  *
6  * Permission is hereby granted, free of charge, to any person obtaining a copy
7  * of this software and associated documentation files (the "Software"), to
8  * deal in the Software without restriction, including without limitation the
9  * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
10  * sell copies of the Software, and to permit persons to whom the Software is
11  * furnished to do so, subject to the following conditions:
12  *
13  * The above copyright notice and this permission notice shall be included in all
14  * copies or substantial portions of the Software.
15  *
16  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
17  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
18  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
19  * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
20  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
21  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
22  * SOFTWARE.
23  */
24 
25 #ifndef ARM_COMPUTE_GRAPH_BACKENDS_FUSED_CONVOLUTION_BATCH_NORMAZLIZATION_FUNCTION_H
26 #define ARM_COMPUTE_GRAPH_BACKENDS_FUSED_CONVOLUTION_BATCH_NORMAZLIZATION_FUNCTION_H
27 
28 #include "arm_compute/core/Types.h"
29 #include "arm_compute/runtime/IFunction.h"
30 
31 namespace arm_compute
32 {
33 namespace graph
34 {
35 namespace backends
36 {
37 /** Wrapper function to first apply {NE, CL}BatchNormalizationLayer on the weights and then run {NE, CL}ConvolutionLayer with the modified weights */
38 template <typename TargetInfo, typename FusedLayerTypes>
39 class FusedConvolutionBatchNormalizationFunction : public IFunction
40 {
41 public:
42     using TensorType         = typename TargetInfo::TensorType;
43     using TensorConcreteType = typename TargetInfo::TensorConcreteType;
44 
45     FusedConvolutionBatchNormalizationFunction(std::shared_ptr<IMemoryManager> memory_manager = nullptr)
_conv_layer(memory_manager)46         : _conv_layer(memory_manager), _fused_batch_norm_layer(), _fused_bias(), _is_prepared(false)
47     {
48     }
49 
50     /** Set the input and output tensors.
51      *
52      * @param[in]  input      Source tensor. 3 lower dimensions represent a single input [width, height, IFM],
53      *                        while every optional dimension from 4 and above represent a batch of inputs.
54      *                        Data types supported: QASYMM8/F16/F32.
55      * @param[in]  weights    Weights tensor. Weights are 4D tensor with dimensions [kernel_x, kernel_y, IFM, OFM]. Data type supported: Same as @p input.
56      * @param[in]  bias       Biases tensor. Shared biases supported. Biases are 1D tensor with dimensions [OFM].
57      *                        Data type supported: Should match @p input data type.
58      * @param[out] output     Destination tensor. 3 lower dimensions represent a single output [width, height, OFM], while the rest represent batch of outputs.
59      *                        Data types supported: Same as @p input.
60      * @param[in]  mean       Mean values tensor. 1 dimension with size equal to the feature maps [FM]. Data types supported: Same as @p input
61      * @param[in]  var        Variance values tensor. 1 dimension with size equal to the feature maps [FM]. Data types supported: Same as @p input
62      * @param[in]  beta       Beta values tensor info. 1 dimension with size equal to the feature maps [FM]. If not provided, default value for beta is 0. Data types supported: Same as @p input
63      * @param[in]  gamma      Gamma values tensor info. 1 dimension with size equal to the feature maps [FM]. If not provided, default value for gamma is 1. Data types supported: Same as @p input
64      * @param[in]  epsilon    Small value to avoid division with zero. Default value is 0.001f.
65      * @param[in]  conv_info  Contains padding and stride information described in @ref PadStrideInfo.
66      * @param[in]  num_groups Number of groups when performing a grouped convolution. num_groups != 1 is only supported for NCHW data layout
67      * @param[in]  fast_math  Enable fast math computation. In case this flag were set, the function could dispatch the fastest implementation
68      *                        available which may introduce a drop of accuracy as well. Default is false
69      * @param[in]  fused_act  Activation layer information in case of a fused activation.
70      *
71      */
configure(TensorType * input,TensorType * weights,TensorType * bias,TensorType * output,const TensorType * mean,const TensorType * var,const TensorType * beta,const TensorType * gamma,float epsilon,const PadStrideInfo & conv_info,unsigned int num_groups,bool fast_math,ActivationLayerInfo const & fused_act)72     void configure(TensorType       *input,
73                    TensorType       *weights,
74                    TensorType       *bias,
75                    TensorType       *output,
76                    const TensorType *mean,
77                    const TensorType *var,
78                    const TensorType *beta,
79                    const TensorType *gamma,
80                    float epsilon, const PadStrideInfo &conv_info, unsigned int num_groups, bool fast_math, ActivationLayerInfo const &fused_act)
81     {
82         // We don't run any validate, as we assume that the layers have been already validated
83         const bool        has_bias = (bias != nullptr);
84         const TensorType *bias_to_use;
85 
86         // We check if the layer has a bias. If yes, use it in-place. If not, we need to create one
87         // as batch normalization might end up with a bias != 0
88         if(has_bias)
89         {
90             _fused_batch_norm_layer.configure(weights, mean, var, nullptr, nullptr, bias, beta, gamma, epsilon);
91             bias_to_use = bias;
92         }
93         else
94         {
95             _fused_batch_norm_layer.configure(weights, mean, var, nullptr, &_fused_bias, nullptr, beta, gamma, epsilon);
96             bias_to_use = &_fused_bias;
97         }
98 
99         _conv_layer.configure(input, weights, bias_to_use, output, conv_info, WeightsInfo(), Size2D(1U, 1U), fused_act, fast_math, num_groups);
100 
101         if(!has_bias)
102         {
103             _fused_bias.allocator()->allocate();
104         }
105     }
106 
107     // Inherited methods overridden:
run()108     void run()
109     {
110         prepare();
111         _conv_layer.run();
112     }
113 
prepare()114     void prepare()
115     {
116         if(!_is_prepared)
117         {
118             _fused_batch_norm_layer.run();
119             _is_prepared = true;
120         }
121     }
122 
123 private:
124     typename FusedLayerTypes::ConvolutionLayer       _conv_layer;
125     typename FusedLayerTypes::FuseBatchNormalization _fused_batch_norm_layer;
126     TensorConcreteType                               _fused_bias;
127     bool                                             _is_prepared;
128 };
129 } // namespace backends
130 } // namespace graph
131 } // namespace arm_compute
132 
133 #endif /* ARM_COMPUTE_GRAPH_BACKENDS_FUSED_CONVOLUTION_BATCH_NORMAZLIZATION_FUNCTION_H */
134