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
|
// Copyright Yahoo. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
#include <vespa/searchlib/tensor/tensor_buffer_type_mapper.h>
#include <vespa/searchlib/tensor/tensor_buffer_operations.h>
#include <vespa/eval/eval/value_type.h>
#include <vespa/vespalib/gtest/gtest.h>
using search::tensor::TensorBufferOperations;
using search::tensor::TensorBufferTypeMapper;
using vespalib::eval::ValueType;
const vespalib::string tensor_type_sparse_spec("tensor(x{})");
const vespalib::string tensor_type_2d_spec("tensor(x{},y{})");
const vespalib::string tensor_type_2d_mixed_spec("tensor(x{},y[2])");
const vespalib::string float_tensor_type_spec("tensor<float>(y{})");
const vespalib::string tensor_type_dense_spec("tensor(x[2])");
constexpr double grow_factor = 1.02;
struct TestParam
{
vespalib::string _name;
std::vector<size_t> _array_sizes;
std::vector<size_t> _large_array_sizes;
vespalib::string _tensor_type_spec;
TestParam(vespalib::string name, std::vector<size_t> array_sizes, std::vector<size_t> large_array_sizes, const vespalib::string& tensor_type_spec)
: _name(std::move(name)),
_array_sizes(std::move(array_sizes)),
_large_array_sizes(std::move(large_array_sizes)),
_tensor_type_spec(tensor_type_spec)
{
}
TestParam(const TestParam&);
~TestParam();
};
TestParam::TestParam(const TestParam&) = default;
TestParam::~TestParam() = default;
std::ostream& operator<<(std::ostream& os, const TestParam& param)
{
os << param._name;
return os;
}
class TensorBufferTypeMapperTest : public testing::TestWithParam<TestParam>
{
protected:
ValueType _tensor_type;
TensorBufferOperations _ops;
TensorBufferTypeMapper _mapper;
TensorBufferTypeMapperTest();
~TensorBufferTypeMapperTest() override;
std::vector<size_t> get_array_sizes();
std::vector<size_t> get_large_array_sizes();
void select_type_ids();
};
TensorBufferTypeMapperTest::TensorBufferTypeMapperTest()
: testing::TestWithParam<TestParam>(),
_tensor_type(ValueType::from_spec(GetParam()._tensor_type_spec)),
_ops(_tensor_type),
_mapper(GetParam()._array_sizes.size(), grow_factor, &_ops)
{
}
TensorBufferTypeMapperTest::~TensorBufferTypeMapperTest() = default;
std::vector<size_t>
TensorBufferTypeMapperTest::get_array_sizes()
{
uint32_t max_small_subspaces_type_id = GetParam()._array_sizes.size();
std::vector<size_t> array_sizes;
for (uint32_t type_id = 1; type_id <= max_small_subspaces_type_id; ++type_id) {
auto num_subspaces = type_id - 1;
array_sizes.emplace_back(_mapper.get_array_size(type_id));
EXPECT_EQ(_ops.get_buffer_size(num_subspaces), array_sizes.back());
}
return array_sizes;
}
std::vector<size_t>
TensorBufferTypeMapperTest::get_large_array_sizes()
{
auto& large_array_sizes = GetParam()._large_array_sizes;
uint32_t max_large = large_array_sizes.size();
TensorBufferTypeMapper mapper(max_large * 100, grow_factor, &_ops);
std::vector<size_t> result;
for (uint32_t i = 0; i < max_large; ++i) {
uint32_t type_id = (i + 1) * 100;
auto array_size = mapper.get_array_size(type_id);
result.emplace_back(array_size);
EXPECT_EQ(type_id, mapper.get_type_id(array_size));
EXPECT_EQ(type_id, mapper.get_type_id(array_size - 1));
if (array_size == large_array_sizes.back()) {
EXPECT_EQ(0u, mapper.get_type_id(array_size + 1));
} else {
EXPECT_EQ(type_id + 1, mapper.get_type_id(array_size + 1));
}
}
return result;
}
void
TensorBufferTypeMapperTest::select_type_ids()
{
auto& array_sizes = GetParam()._array_sizes;
uint32_t type_id = 0;
for (auto array_size : array_sizes) {
++type_id;
EXPECT_EQ(type_id, _mapper.get_type_id(array_size));
EXPECT_EQ(type_id, _mapper.get_type_id(array_size - 1));
if (array_size == array_sizes.back()) {
// Fallback to indirect storage, using type id 0
EXPECT_EQ(0u, _mapper.get_type_id(array_size + 1));
} else {
EXPECT_EQ(type_id + 1, _mapper.get_type_id(array_size + 1));
}
}
}
/*
* For "dense" case, array size for type id 1 is irrelevant, since
* type ids 0 and 1 are not used when storing dense tensors in
* TensorBufferStore.
*/
INSTANTIATE_TEST_SUITE_P(TensorBufferTypeMapperMultiTest,
TensorBufferTypeMapperTest,
testing::Values(TestParam("1d", {8, 16, 32, 40, 64}, {1760, 10880, 76896, 555248, 4020512}, tensor_type_sparse_spec),
TestParam("1dfloat", {4, 12, 20, 28, 36}, {1728, 11104, 79168, 572128, 4143664}, float_tensor_type_spec),
TestParam("2d", {8, 24, 40, 56, 80}, {1600, 9184, 63872, 460416, 3332976}, tensor_type_2d_spec),
TestParam("2dmixed", {8, 24, 48, 64, 96}, {1984, 11472, 79824, 575504, 4166208}, tensor_type_2d_mixed_spec),
TestParam("dense", {8, 24}, {}, tensor_type_dense_spec)),
testing::PrintToStringParamName());
TEST_P(TensorBufferTypeMapperTest, array_sizes_are_calculated)
{
EXPECT_EQ(GetParam()._array_sizes, get_array_sizes());
}
TEST_P(TensorBufferTypeMapperTest, type_ids_are_selected)
{
select_type_ids();
}
TEST_P(TensorBufferTypeMapperTest, large_arrays_grows_exponentially)
{
EXPECT_EQ(GetParam()._large_array_sizes, get_large_array_sizes());
}
TEST_P(TensorBufferTypeMapperTest, avoid_array_size_overflow)
{
TensorBufferTypeMapper mapper(400, 2.0, &_ops);
EXPECT_GE(30, mapper.get_max_small_array_type_id(1000));
}
GTEST_MAIN_RUN_ALL_TESTS()
|
