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
|
#include <aocommon/recursivefor.h>
#include <aocommon/staticfor.h>
#include <atomic>
#include <set>
#include <mutex>
#include <unistd.h> // for sleep
#include <boost/test/unit_test.hpp>
using aocommon::RecursiveFor;
using aocommon::StaticFor;
using aocommon::ThreadPool;
BOOST_AUTO_TEST_SUITE(staticfor)
BOOST_AUTO_TEST_CASE(run) {
ThreadPool::GetInstance().SetNThreads(4);
StaticFor<size_t> loop;
std::mutex mutex;
std::vector<size_t> counts(10, 0);
loop.Run(0, 10, [&](size_t a, size_t b) {
for (size_t iter = a; iter != b; ++iter) {
std::unique_lock<std::mutex> lock(mutex);
counts[iter]++;
}
});
std::vector<size_t> ref(10, 1);
BOOST_CHECK_EQUAL_COLLECTIONS(counts.begin(), counts.end(), ref.begin(),
ref.end());
}
BOOST_AUTO_TEST_CASE(empty_loop) {
ThreadPool::GetInstance().SetNThreads(4);
StaticFor<size_t> loop;
std::atomic<bool> fail = false;
loop.Run(10, 10, [&](size_t a, size_t b) {
if (a != b) fail = true;
});
BOOST_CHECK(!fail);
}
BOOST_AUTO_TEST_CASE(single_threaded) {
ThreadPool::GetInstance().SetNThreads(1);
StaticFor<size_t> loop;
std::vector<size_t> counts(10, 0);
loop.Run(0, 10, [&](size_t a, size_t b) {
for (size_t iter = a; iter != b; ++iter) {
counts[iter]++;
}
});
std::vector<size_t> ref(10, 1);
BOOST_CHECK_EQUAL_COLLECTIONS(counts.begin(), counts.end(), ref.begin(),
ref.end());
}
BOOST_AUTO_TEST_CASE(resume_run) {
std::vector<size_t> counts(20, 0);
std::mutex mutex;
ThreadPool::GetInstance().SetNThreads(40);
StaticFor<size_t> loop;
loop.Run(0, 10, [&](size_t a, size_t b) {
for (size_t iter = a; iter != b; ++iter) {
std::unique_lock<std::mutex> lock(mutex);
counts[iter]++;
}
});
std::vector<size_t> ref(20, 0);
std::fill(ref.begin(), ref.begin() + 10, 1);
BOOST_CHECK_EQUAL_COLLECTIONS(counts.begin(), counts.end(), ref.begin(),
ref.end());
loop.Run(10, 20, [&](size_t a, size_t b) {
for (size_t iter = a; iter != b; ++iter) {
std::unique_lock<std::mutex> lock(mutex);
counts[iter]++;
}
});
ref = std::vector<size_t>(20, 1);
BOOST_CHECK_EQUAL_COLLECTIONS(counts.begin(), counts.end(), ref.begin(),
ref.end());
}
BOOST_AUTO_TEST_CASE(run_with_thread_id) {
ThreadPool::GetInstance().SetNThreads(4);
StaticFor<size_t> loop;
std::mutex mutex;
std::vector<size_t> counts(10, 0);
std::vector<size_t> threads(4, 0);
loop.Run(0, 10, [&](size_t a, size_t b, size_t t) {
for (size_t iter = a; iter != b; ++iter) {
std::unique_lock<std::mutex> lock(mutex);
counts[iter]++;
}
std::unique_lock<std::mutex> lock(mutex);
threads[t]++;
});
std::vector<size_t> ref(10, 1);
BOOST_CHECK_EQUAL_COLLECTIONS(counts.begin(), counts.end(), ref.begin(),
ref.end());
// Not all threads might actually be used, because if one thread
// finishes before a second thread is starting, the first
// thread is used to perform the next loop.
// Therefore all we can check is whether there weren't more than
// 4 blocks:
for (size_t i = 0; i != threads.size(); ++i) BOOST_CHECK_LT(threads[i], 5);
}
BOOST_AUTO_TEST_CASE(inside_recursive_for) {
const size_t kNThreads = 7;
ThreadPool::GetInstance().SetNThreads(kNThreads);
RecursiveFor recursive_for;
std::mutex mutex;
std::map<std::pair<size_t, size_t>, size_t> counter;
recursive_for.Run(0, 10, [&](size_t outer, size_t) {
StaticFor<size_t> nested_for;
nested_for.Run(0, 10, [&, outer](size_t start, size_t end) {
for (size_t inner = start; inner != end; ++inner) {
std::lock_guard lock(mutex);
counter[std::make_pair(outer, inner)]++;
}
});
});
BOOST_CHECK_EQUAL(counter.size(), 100);
for (const std::pair<const std::pair<size_t, size_t>, size_t>& count :
counter) {
BOOST_CHECK_LT(count.first.first, 10); // outer
BOOST_CHECK_LT(count.first.second, 10); // inner
BOOST_CHECK_EQUAL(count.second, 1); // count
}
}
BOOST_AUTO_TEST_CASE(constrained_run_without_id) {
ThreadPool::GetInstance().SetNThreads(4);
StaticFor<size_t> loop;
std::mutex mutex;
std::vector<size_t> counts(10, 0);
std::set<std::thread::id> thread_ids;
const size_t kMaxThreads = 3;
loop.ConstrainedRun(0, 10, kMaxThreads, [&](size_t a, size_t b) {
for (size_t iter = a; iter != b; ++iter) {
std::scoped_lock<std::mutex> lock(mutex);
counts[iter]++;
}
std::scoped_lock<std::mutex> lock(mutex);
thread_ids.emplace(std::this_thread::get_id());
});
std::vector<size_t> ref(10, 1);
BOOST_CHECK_EQUAL(thread_ids.size(), kMaxThreads);
BOOST_CHECK_EQUAL_COLLECTIONS(counts.begin(), counts.end(), ref.begin(),
ref.end());
}
BOOST_AUTO_TEST_CASE(constrained_run_with_id) {
ThreadPool::GetInstance().SetNThreads(4);
StaticFor<size_t> loop;
std::mutex mutex;
std::vector<size_t> counts(10, 0);
std::set<std::thread::id> thread_ids;
std::set<size_t> thread_indices;
const size_t kMaxThreads = 3;
loop.ConstrainedRun(0, 10, kMaxThreads,
[&](size_t a, size_t b, size_t thread_index) {
for (size_t iter = a; iter != b; ++iter) {
std::scoped_lock<std::mutex> lock(mutex);
counts[iter]++;
}
std::scoped_lock<std::mutex> lock(mutex);
thread_ids.emplace(std::this_thread::get_id());
thread_indices.emplace(thread_index);
});
BOOST_CHECK_EQUAL(thread_ids.size(), kMaxThreads);
BOOST_CHECK_EQUAL(thread_indices.size(), kMaxThreads);
const std::set<size_t> expected_indices{0, 1, 2};
BOOST_CHECK_EQUAL_COLLECTIONS(thread_indices.begin(), thread_indices.end(),
expected_indices.begin(),
expected_indices.end());
std::vector<size_t> expected_counts(10, 1);
BOOST_CHECK_EQUAL_COLLECTIONS(counts.begin(), counts.end(),
expected_counts.begin(), expected_counts.end());
}
BOOST_AUTO_TEST_CASE(conditional_nested_run) {
// Test the special case in which no RecursiveFor exists and a static for
// is nested nevertheless, but constrained to one thread max. This allows
// nested parallelization and the making of a RecursiveFor to be done
// conditional, which for certain cases is useful to use as many threads as
// possible.
size_t iteration_count = 0;
std::mutex mutex;
BOOST_CHECK(aocommon::RecursiveFor::GetInstance() == nullptr);
aocommon::RunStaticFor<size_t>(0, 1000, [&](size_t start, size_t end) {
for (size_t outer_index = start; outer_index != end; ++outer_index) {
aocommon::RunConstrainedStaticFor<size_t>(
0, 10, 1, [&](size_t nested_start, size_t nested_end) {
std::lock_guard lock(mutex);
iteration_count += nested_end - nested_start;
});
}
});
BOOST_CHECK_EQUAL(iteration_count, 10000);
}
BOOST_AUTO_TEST_SUITE_END()
|
