1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
|
/*
* Copyright (c) 2017-2021, 2023 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
* Permission is hereby granted, free of charge, to any person obtaining a copy
* of this software and associated documentation files (the "Software"), to
* deal in the Software without restriction, including without limitation the
* rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
* sell copies of the Software, and to permit persons to whom the Software is
* furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in all
* copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
* AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
* SOFTWARE.
*/
#ifndef ACL_SRC_GPU_CL_OPERATORS_CLFULLYCONNECTED_H
#define ACL_SRC_GPU_CL_OPERATORS_CLFULLYCONNECTED_H
#include "arm_compute/core/TensorInfo.h"
#include "arm_compute/function_info/FullyConnectedLayerInfo.h"
#include "src/gpu/cl/ClCompileContext.h"
#include "src/gpu/cl/IClOperator.h"
#include "src/gpu/cl/kernels/ClMatMulLowpNativeKernel.h"
#include "src/gpu/cl/kernels/ClMatMulNativeKernel.h"
#include <memory>
namespace arm_compute
{
namespace opencl
{
// Forward declarations
class ClConvertFullyConnectedWeights;
class ClFlatten;
class ClGemm;
class ClGemmLowpMatrixMultiplyCore;
class ClTranspose;
// Kernel Forward Declarations
namespace kernels
{
class ClMatMulNativeKernel;
class ClMatMulLowpNativeKernel;
} // namespace kernels
/** Basic function to compute a Fully Connected layer on OpenCL. This function calls the following OpenCL kernels:
*
* -# @ref opencl::kernels::ClIm2ColKernel (called when the input comes from a convolutional layer)
* -# @ref CLTranspose (if @p are_weights_reshaped is set to false and transpose_weights is set to true ) (called once)
* -# @ref opencl::ClGemm or @ref CLGEMMLowpMatrixMultiplyCore (if quantized asymmetric)
*
* @note The fully connected layer accepts "weights" tensors only with 2 dimensions.
*/
class ClFullyConnected : public IClOperator
{
public:
ClFullyConnected();
~ClFullyConnected();
/** Set the input and output tensors.
*
* Valid data layouts:
* - NHWC
* - NCHW
*
* Valid data type configurations:
* |src0 |src1 |src2 |dst |
* |:--------------|:------------------|:------|:--------------|
* |F16 |F16 |F16 |F16 |
* |F32 |F32 |F32 |F32 |
* |QASYMM8 |QASYMM8 |S32 |QASYMM8 |
* |QASYMM8_SIGNED |QASYMM8_SIGNED |S32 |QASYMM8_SIGNED |
*
* @param[in] compile_context The compile context to be used.
* @param[in] src Source tensor. Data type supported: QASYMM8/QASYMM8_SIGNED/F16/F32.
* @param[in] weights Weights tensor. The weights must be 2 dimensional.
* If this function is called after a Convolution Layer, the (transposed) weights will have as many rows as the product of the first 3 input's dimensions.
* If it is called after another FullyConnected Layer, the (transposed) weights will have as many rows as the input's first dimension.
* Data type supported: Same as @p src.
* @param[in] biases Bias tensor. Can be nullptr. Data type supported:Same as @p src.
* @param[out] dst Destination tensor. Its shape should be equal to the output of a matrix multiplication between:
* - The output of im2col on the input and the (transposed) 2D weights, if the function is called after a Convolution Layer
* - The input tensor and the (transposed) 2D weights, if the function is called after another FullyConnected Layer.
* Data type supported: Same as @p src.
* @param[in] fc_info (Optional) Fully connected layer additional info
*/
void configure(const CLCompileContext &compile_context,
ITensorInfo *src,
ITensorInfo *weights,
ITensorInfo *biases,
ITensorInfo *dst,
FullyConnectedLayerInfo fc_info = FullyConnectedLayerInfo());
/** Static function to check if given info will lead to a valid configuration
*
* Similar to ClFullyConnected::configure()
*
* @return a status
*/
static Status validate(const ITensorInfo *src,
const ITensorInfo *weights,
const ITensorInfo *biases,
const ITensorInfo *dst,
FullyConnectedLayerInfo fc_info = FullyConnectedLayerInfo());
// Inherited methods overriden
void run(ITensorPack &tensors) override;
void prepare(ITensorPack &tensors) override;
experimental::MemoryRequirements workspace() const override;
private:
void configure_fc_fc(const CLCompileContext &compile_context,
ITensorInfo *src,
ITensorInfo *weights,
ITensorInfo *bias,
ITensorInfo *dst,
const FullyConnectedLayerInfo &fc_info);
void configure_conv_fc(const CLCompileContext &compile_context,
ITensorInfo *src,
ITensorInfo *weights,
ITensorInfo *bias,
ITensorInfo *dst,
const FullyConnectedLayerInfo &fc_info);
void configure_mm(const CLCompileContext &compile_context,
ITensorInfo *src,
ITensorInfo *weights,
ITensorInfo *bias,
ITensorInfo *dst,
const FullyConnectedLayerInfo &fc_info);
private:
enum AuxTensorIdx
{
TransposedWeights = 10,
ConvertedWeights = 11,
FlattenedSrc = 12,
Count = 13
};
std::unique_ptr<ClConvertFullyConnectedWeights> _convert_weights;
std::unique_ptr<ClFlatten> _flatten;
std::unique_ptr<ClTranspose> _reshape_weights;
std::unique_ptr<ClGemm> _mm_gemm;
std::unique_ptr<ClGemmLowpMatrixMultiplyCore> _mm_gemmlowp;
std::unique_ptr<kernels::ClMatMulNativeKernel> _matmul_native_kernel;
std::unique_ptr<kernels::ClMatMulLowpNativeKernel> _matmul_lowp_native_kernel;
experimental::MemoryRequirements _aux_mem{};
TensorInfo _flattened_src{};
TensorInfo _converted_weights{};
TensorInfo _reshaped_weights{};
TensorInfo _lhs_to_use{};
TensorInfo _weights_to_use{};
int _weights_to_use_idx{ACL_SRC_1};
bool _run_convert_weights{false};
bool _transpose_weights{false};
bool _dynamic_gemm{false};
bool _use_matmul{false};
bool _is_fc_after_conv{true};
bool _is_quantized{false};
bool _is_prepared{false};
#ifdef ARM_COMPUTE_ASSERTS_ENABLED
int _asrt_run_count{};
int _asrt_prepare_count{};
#endif // ARM_COMPUTE_ASSERTS_ENABLED
};
} // namespace opencl
} // namespace arm_compute
#endif // ACL_SRC_GPU_CL_OPERATORS_CLFULLYCONNECTED_H
|
