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
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
|
/*
* Copyright 2022 Patrick Stotko
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
#include <benchmark/benchmark.h>
#include <algorithm>
#include <limits>
#include <random>
#include <benchmark_utils.h>
#include <stdgpu/algorithm.h>
#include <stdgpu/memory.h>
#include <stdgpu/vector.cuh>
namespace
{
int*
create_values(const stdgpu::index_t N)
{
// Generate true random numbers
size_t seed = benchmark_utils::random_seed();
std::default_random_engine rng(static_cast<std::default_random_engine::result_type>(seed));
std::uniform_int_distribution<int> dist(std::numeric_limits<int>::lowest(), std::numeric_limits<int>::max());
int* host_values = createHostArray<int>(N);
std::generate(host_values, host_values + N, [&dist, &rng]() { return dist(rng); });
int* values = copyCreateHost2DeviceArray<int>(host_values, N);
destroyHostArray<int>(host_values);
return values;
}
template <typename T>
class push_back_vector
{
public:
push_back_vector(const stdgpu::vector<T>& pool, T* values)
: _pool(pool)
, _values(values)
{
}
STDGPU_DEVICE_ONLY void
operator()(const stdgpu::index_t i)
{
_pool.push_back(_values[i]);
}
private:
stdgpu::vector<T> _pool;
T* _values;
};
template <typename T>
class pop_back_vector
{
public:
explicit pop_back_vector(const stdgpu::vector<T>& pool)
: _pool(pool)
{
}
STDGPU_DEVICE_ONLY void
operator()([[maybe_unused]] const stdgpu::index_t i)
{
_pool.pop_back();
}
private:
stdgpu::vector<T> _pool;
};
} // namespace
void
stdgpu_vector_insert(benchmark::State& state, const stdgpu::index_t vector_size)
{
stdgpu::vector<int> v = stdgpu::vector<int>::createDeviceObject(vector_size);
int* values = create_values(vector_size);
v.insert(v.device_end(), stdgpu::device_begin(values), stdgpu::device_end(values));
v.clear();
for (auto _ : state)
{
v.insert(v.device_end(), stdgpu::device_begin(values), stdgpu::device_end(values));
state.PauseTiming();
v.clear();
state.ResumeTiming();
}
stdgpu::vector<int>::destroyDeviceObject(v);
destroyDeviceArray<int>(values);
}
void
stdgpu_vector_erase(benchmark::State& state, const stdgpu::index_t vector_size)
{
stdgpu::vector<int> v = stdgpu::vector<int>::createDeviceObject(vector_size);
int* values = create_values(vector_size);
v.insert(v.device_end(), stdgpu::device_begin(values), stdgpu::device_end(values));
for (auto _ : state)
{
v.erase(v.device_begin(), v.device_end());
state.PauseTiming();
v.insert(v.device_end(), stdgpu::device_begin(values), stdgpu::device_end(values));
state.ResumeTiming();
}
stdgpu::vector<int>::destroyDeviceObject(v);
destroyDeviceArray<int>(values);
}
void
stdgpu_vector_push_back(benchmark::State& state, const stdgpu::index_t vector_size)
{
stdgpu::vector<int> v = stdgpu::vector<int>::createDeviceObject(vector_size);
int* values = create_values(vector_size);
stdgpu::for_each_index(stdgpu::execution::device, vector_size, push_back_vector<int>(v, values));
v.clear();
for (auto _ : state)
{
stdgpu::for_each_index(stdgpu::execution::device, vector_size, push_back_vector<int>(v, values));
state.PauseTiming();
v.clear();
state.ResumeTiming();
}
stdgpu::vector<int>::destroyDeviceObject(v);
destroyDeviceArray<int>(values);
}
void
stdgpu_vector_pop_back(benchmark::State& state, const stdgpu::index_t vector_size)
{
stdgpu::vector<int> v = stdgpu::vector<int>::createDeviceObject(vector_size);
int* values = create_values(vector_size);
v.insert(v.device_end(), stdgpu::device_begin(values), stdgpu::device_end(values));
for (auto _ : state)
{
stdgpu::for_each_index(stdgpu::execution::device, vector_size, pop_back_vector<int>(v));
state.PauseTiming();
v.insert(v.device_end(), stdgpu::device_begin(values), stdgpu::device_end(values));
state.ResumeTiming();
}
stdgpu::vector<int>::destroyDeviceObject(v);
destroyDeviceArray<int>(values);
}
void
stdgpu_vector_clear(benchmark::State& state, const stdgpu::index_t vector_size)
{
stdgpu::vector<int> v = stdgpu::vector<int>::createDeviceObject(vector_size);
int* values = create_values(vector_size);
v.insert(v.device_end(), stdgpu::device_begin(values), stdgpu::device_end(values));
for (auto _ : state)
{
v.clear();
state.PauseTiming();
v.insert(v.device_end(), stdgpu::device_begin(values), stdgpu::device_end(values));
state.ResumeTiming();
}
stdgpu::vector<int>::destroyDeviceObject(v);
destroyDeviceArray<int>(values);
}
void
stdgpu_vector_valid(benchmark::State& state, const stdgpu::index_t vector_size)
{
stdgpu::vector<int> v = stdgpu::vector<int>::createDeviceObject(vector_size);
benchmark::DoNotOptimize(v.valid());
for (auto _ : state)
{
benchmark::DoNotOptimize(v.valid());
}
stdgpu::vector<int>::destroyDeviceObject(v);
}
#define STDGPU_REGISTER_BENCHMARK(function) \
BENCHMARK_CAPTURE(function, 1000, 1000)->Unit(benchmark::kMillisecond); \
BENCHMARK_CAPTURE(function, 100000, 100000)->Unit(benchmark::kMillisecond); \
BENCHMARK_CAPTURE(function, 10000000, 10000000)->Unit(benchmark::kMillisecond);
STDGPU_REGISTER_BENCHMARK(stdgpu_vector_insert)
STDGPU_REGISTER_BENCHMARK(stdgpu_vector_erase)
STDGPU_REGISTER_BENCHMARK(stdgpu_vector_push_back)
STDGPU_REGISTER_BENCHMARK(stdgpu_vector_pop_back)
// clear is significantly faster than non-measured insert
#if STDGPU_BACKEND != STDGPU_BACKEND_OPENMP
STDGPU_REGISTER_BENCHMARK(stdgpu_vector_clear)
#endif
STDGPU_REGISTER_BENCHMARK(stdgpu_vector_valid)
|
