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
180
181
182
183
184
185
186
187
|
/*
* Copyright (c) 2017-2018 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.
*/
#include "arm_compute/core/Types.h"
#include "arm_compute/runtime/NEON/functions/NEDirectConvolutionLayer.h"
#include "arm_compute/runtime/Tensor.h"
#include "arm_compute/runtime/TensorAllocator.h"
#include "tests/NEON/Accessor.h"
#include "tests/PaddingCalculator.h"
#include "tests/datasets/ShapeDatasets.h"
#include "tests/framework/Asserts.h"
#include "tests/framework/Macros.h"
#include "tests/framework/datasets/Datasets.h"
#include "tests/validation/Validation.h"
#include "tests/validation/fixtures/DirectConvolutionLayerFixture.h"
namespace arm_compute
{
namespace test
{
namespace validation
{
namespace
{
#ifdef __ARM_FEATURE_FP16_VECTOR_ARITHMETIC
constexpr AbsoluteTolerance<float> tolerance_fp16(0.01f); /**< Tolerance for half precision floating point tests */
#endif /* __ARM_FEATURE_FP16_VECTOR_ARITHMETIC */
constexpr AbsoluteTolerance<float> tolerance_fp32(0.001f); /**< Tolerance for floating point tests */
/** Direct convolution data set. */
const auto data_pad_f32 = concat(concat(combine(framework::dataset::make("PadX", { 0, 1 }),
combine(framework::dataset::make("PadY", { 0, 1 }),
framework::dataset::make("KernelSize", 3))),
combine(framework::dataset::make("PadX", { 0, 2 }),
combine(framework::dataset::make("PadY", { 0, 2 }),
framework::dataset::make("KernelSize", 3)))),
combine(framework::dataset::make("PadX", { 0, 3 }),
combine(framework::dataset::make("PadY", { 0, 3 }),
framework::dataset::make("KernelSize", 5))));
const auto data_f32 = combine(datasets::SmallDirectConvolutionShapes(),
combine(framework::dataset::make("StrideX", { 1, 3 }),
combine(framework::dataset::make("StrideY", { 1, 3 }),
combine(data_pad_f32,
framework::dataset::make("NumKernels", { 1, 4, 8, 16 })))));
/** Activation function Dataset*/
const auto ActivationFunctionsDataset = framework::dataset::make("ActivationInfo",
{
ActivationLayerInfo(),
ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU),
ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::BOUNDED_RELU, 0.5f),
ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::LU_BOUNDED_RELU, 0.5f)
});
} // namespace
TEST_SUITE(NEON)
TEST_SUITE(DirectConvolutionLayer)
// *INDENT-OFF*
// clang-format off
DATA_TEST_CASE(Validate, framework::DatasetMode::ALL, zip(zip(zip(zip(zip(zip(
framework::dataset::make("InputInfo", { TensorInfo(TensorShape(27U, 13U, 2U), 1, DataType::F32, 0), // Mismatching data type input/weights
TensorInfo(TensorShape(27U, 13U, 2U), 1, DataType::F32, 0), // Mismatching input feature maps
TensorInfo(TensorShape(27U, 13U, 2U), 1, DataType::F32, 0), // Unsupported kernel width
TensorInfo(TensorShape(27U, 13U, 2U), 1, DataType::F32, 0), // Non-rectangular weights dimensions
TensorInfo(TensorShape(27U, 13U, 2U), 1, DataType::F32, 0), // Invalid weights dimensions
TensorInfo(TensorShape(27U, 13U, 2U), 1, DataType::F32, 0), // Invalid stride
TensorInfo(TensorShape(27U, 13U, 2U), 1, DataType::F32, 0), // Invalid biases size
TensorInfo(TensorShape(27U, 13U, 2U), 1, DataType::F32, 0), // Invalid biases dimensions
TensorInfo(TensorShape(27U, 13U, 2U), 1, DataType::F32, 0), // Invalid output size
}),
framework::dataset::make("WeightsInfo",{ TensorInfo(TensorShape(3U, 3U, 2U, 4U), 1, DataType::F16, 0),
TensorInfo(TensorShape(3U, 3U, 3U, 4U), 1, DataType::F32, 0),
TensorInfo(TensorShape(9U, 9U, 2U, 4U), 1, DataType::F32, 0),
TensorInfo(TensorShape(5U, 3U, 2U, 4U), 1, DataType::F32, 0),
TensorInfo(TensorShape(3U, 3U, 2U, 4U, 3U), 1, DataType::F32, 0),
TensorInfo(TensorShape(3U, 3U, 2U, 4U), 1, DataType::F32, 0),
TensorInfo(TensorShape(3U, 3U, 2U, 4U), 1, DataType::F32, 0),
TensorInfo(TensorShape(3U, 3U, 2U, 4U), 1, DataType::F32, 0),
TensorInfo(TensorShape(3U, 3U, 2U, 4U), 1, DataType::F32, 0),
})),
framework::dataset::make("BiasesInfo",{ TensorInfo(TensorShape(4U), 1, DataType::F32, 0),
TensorInfo(TensorShape(4U), 1, DataType::F32, 0),
TensorInfo(TensorShape(4U), 1, DataType::F32, 0),
TensorInfo(TensorShape(4U), 1, DataType::F32, 0),
TensorInfo(TensorShape(4U), 1, DataType::F32, 0),
TensorInfo(TensorShape(4U), 1, DataType::F32, 0),
TensorInfo(TensorShape(3U), 1, DataType::F32, 0),
TensorInfo(TensorShape(4U, 2U), 1, DataType::F32, 0),
TensorInfo(TensorShape(4U), 1, DataType::F32, 0),
})),
framework::dataset::make("OutputInfo",{ TensorInfo(TensorShape(25U, 11U, 4U), 1, DataType::F32, 0),
TensorInfo(TensorShape(25U, 11U, 4U), 1, DataType::F32, 0),
TensorInfo(TensorShape(25U, 11U, 4U), 1, DataType::F32, 0),
TensorInfo(TensorShape(25U, 11U, 4U), 1, DataType::F32, 0),
TensorInfo(TensorShape(25U, 11U, 4U), 1, DataType::F32, 0),
TensorInfo(TensorShape(25U, 11U, 4U), 1, DataType::F32, 0),
TensorInfo(TensorShape(25U, 11U, 4U), 1, DataType::F32, 0),
TensorInfo(TensorShape(25U, 11U, 4U), 1, DataType::F32, 0),
TensorInfo(TensorShape(26U, 11U, 4U), 1, DataType::F32, 0),
})),
framework::dataset::make("ConvInfo", { PadStrideInfo(1, 1, 0, 0),
PadStrideInfo(1, 1, 0, 0),
PadStrideInfo(1, 1, 0, 0),
PadStrideInfo(1, 1, 0, 0),
PadStrideInfo(1, 1, 0, 0),
PadStrideInfo(3, 3, 0, 0),
PadStrideInfo(1, 1, 0, 0),
PadStrideInfo(1, 1, 0, 0),
PadStrideInfo(1, 1, 0, 0),
})),
framework::dataset::make("ActivationInfo",
{
ActivationLayerInfo(),
ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU)
})),
framework::dataset::make("Expected", { false, false, false, false, false, false, false, false, false })),
input_info, weights_info, biases_info, output_info, conv_info, act_info, expected)
{
bool is_valid = bool(NEDirectConvolutionLayer::validate(&input_info.clone()->set_is_resizable(false), &weights_info.clone()->set_is_resizable(false), &biases_info.clone()->set_is_resizable(false), &output_info.clone()->set_is_resizable(false), conv_info, act_info));
ARM_COMPUTE_EXPECT(is_valid == expected, framework::LogLevel::ERRORS);
}
// clang-format on
// *INDENT-ON*
//TODO(COMPMID-415): Configuration tests?
template <typename T>
using NEDirectConvolutionLayerFixture = DirectConvolutionValidationFixture<Tensor, Accessor, NEDirectConvolutionLayer, T>;
TEST_SUITE(Float)
#ifdef __ARM_FEATURE_FP16_VECTOR_ARITHMETIC
TEST_SUITE(FP16)
FIXTURE_DATA_TEST_CASE(Run, NEDirectConvolutionLayerFixture<half>, framework::DatasetMode::ALL, combine(combine(combine(data_f32, framework::dataset::make("DataType", DataType::F16)),
ActivationFunctionsDataset),
framework::dataset::make("DataLayout", DataLayout::NCHW)))
{
// Validate output
validate(Accessor(_target), _reference, tolerance_fp16);
}
TEST_SUITE_END()
#endif /* __ARM_FEATURE_FP16_VECTOR_ARITHMETIC */
TEST_SUITE(FP32)
FIXTURE_DATA_TEST_CASE(Run, NEDirectConvolutionLayerFixture<float>, framework::DatasetMode::ALL, combine(combine(combine(data_f32, framework::dataset::make("DataType", DataType::F32)),
ActivationFunctionsDataset),
framework::dataset::make("DataLayout", { DataLayout::NCHW, DataLayout::NHWC })))
{
// Validate output
validate(Accessor(_target), _reference, tolerance_fp32);
}
TEST_SUITE_END()
TEST_SUITE_END()
const auto QuantizedActivationFunctionsDataset = framework::dataset::make("ActivationInfo",
{
ActivationLayerInfo(),
ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU),
ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::LU_BOUNDED_RELU, 6.f)
});
TEST_SUITE_END()
TEST_SUITE_END()
} // namespace validation
} // namespace test
} // namespace arm_compute
|
