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
|
//
// Copyright © 2020, 2023-2024 Arm Ltd and Contributors. All rights reserved.
// SPDX-License-Identifier: MIT
//
#pragma once
#include <tensorflow/lite/c/common.h>
#include <armnn/BackendId.hpp>
#include <half/half.hpp>
#include <doctest/doctest.h>
using Half = half_float::half;
namespace
{
/**
* Based on the compilation options capture subcases for the available backends. If "onlyTheseBackends" is NOT empty
* then we'll ignore any backend NOT listed in it.
*
* @param onlyTheseBackends limit the number of backends considered for sub casing. If empty all are considered.
* @return vector of backends that have been captured for sub casing.
*/
std::vector<armnn::BackendId> CaptureAvailableBackends(const std::vector<armnn::BackendId>& onlyTheseBackends)
{
std::vector<armnn::BackendId> availableBackends;
#if defined(ARMNNREF_ENABLED)
// Careful logic here. An empty onlyTheseBackends means we always evaluate.
if (onlyTheseBackends.empty() || (std::find(onlyTheseBackends.begin(), onlyTheseBackends.end(),
armnn::Compute::CpuRef) != onlyTheseBackends.end()))
{
SUBCASE("CpuRef")
{
availableBackends.push_back({ armnn::Compute::CpuRef });
}
}
#endif
#if defined(ARMCOMPUTENEON_ENABLED)
// Careful logic here. An empty onlyTheseBackends means we always evaluate.
if (onlyTheseBackends.empty() || (std::find(onlyTheseBackends.begin(), onlyTheseBackends.end(),
armnn::Compute::CpuAcc) != onlyTheseBackends.end()))
{
SUBCASE("CpuAcc")
{
availableBackends.push_back({ armnn::Compute::CpuAcc });
}
}
#endif
#if defined(ARMCOMPUTECL_ENABLED)
if (onlyTheseBackends.empty() || (std::find(onlyTheseBackends.begin(), onlyTheseBackends.end(),
armnn::Compute::GpuAcc) != onlyTheseBackends.end()))
{
SUBCASE("GpuAcc")
{
availableBackends.push_back({ armnn::Compute::GpuAcc });
}
}
#endif
CAPTURE(availableBackends);
return availableBackends;
}
} // namespace
namespace armnnDelegate
{
constexpr const char* FILE_IDENTIFIER = "TFL3";
/// Can be used to compare bool data coming from a tflite interpreter
/// Boolean types get converted to a bit representation in a vector. vector.data() returns a void pointer
/// instead of a pointer to bool. Therefore a special function to compare to vector of bool is required
void CompareData(std::vector<bool>& tensor1, std::vector<bool>& tensor2, size_t tensorSize);
void CompareData(bool tensor1[], bool tensor2[], size_t tensorSize);
/// Can be used to compare float data coming from a tflite interpreter with a tolerance of limit_of_float*100
void CompareData(float tensor1[], float tensor2[], size_t tensorSize);
/// Can be used to compare float data coming from a tflite interpreter with a given percentage tolerance
void CompareData(float tensor1[], float tensor2[], size_t tensorSize, float percentTolerance);
/// Can be used to compare int8_t data coming from a tflite interpreter with a tolerance of 1
void CompareData(int8_t tensor1[], int8_t tensor2[], size_t tensorSize);
/// Can be used to compare uint8_t data coming from a tflite interpreter with a tolerance of 1
void CompareData(uint8_t tensor1[], uint8_t tensor2[], size_t tensorSize);
/// Can be used to compare int16_t data coming from a tflite interpreter with a tolerance of 1
void CompareData(int16_t tensor1[], int16_t tensor2[], size_t tensorSize);
/// Can be used to compare int32_t data coming from a tflite interpreter with a tolerance of 1
void CompareData(int32_t tensor1[], int32_t tensor2[], size_t tensorSize);
/// Can be used to compare Half (Float16) data with a tolerance of limit_of_float*100
void CompareData(Half tensor1[], Half tensor2[], size_t tensorSize);
/// Can be used to compare TfLiteFloat16 data coming from a tflite interpreter
void CompareData(TfLiteFloat16 tensor1[], TfLiteFloat16 tensor2[], size_t tensorSize);
/// Can be used to compare Half (Float16) data and TfLiteFloat16 data coming from a tflite interpreter
void CompareData(TfLiteFloat16 tensor1[], Half tensor2[], size_t tensorSize);
/// Can be used to compare the output tensor shape
/// Example usage can be found in ControlTestHelper.hpp
void CompareOutputShape(const std::vector<int32_t>& tfLiteDelegateShape,
const std::vector<int32_t>& armnnDelegateShape,
const std::vector<int32_t>& expectedOutputShape);
/// Can be used to compare the output tensor values
/// Example usage can be found in ControlTestHelper.hpp
template <typename T>
void CompareOutputData(std::vector<T>& tfLiteDelegateOutputs,
std::vector<T>& armnnDelegateOutputs,
std::vector<T>& expectedOutputValues)
{
armnnDelegate::CompareData(expectedOutputValues.data(), armnnDelegateOutputs.data(), expectedOutputValues.size());
armnnDelegate::CompareData(tfLiteDelegateOutputs.data(), expectedOutputValues.data(), expectedOutputValues.size());
armnnDelegate::CompareData(tfLiteDelegateOutputs.data(), armnnDelegateOutputs.data(), expectedOutputValues.size());
}
} // namespace armnnDelegate
|
