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
|
// SPDX-License-Identifier: BSD-3-Clause
/* Copyright 2020, Intel Corporation */
#ifndef THREAD_HELPERS_COMMON_HPP
#define THREAD_HELPERS_COMMON_HPP
#include <condition_variable>
#include <functional>
#include <mutex>
#include <thread>
#include <vector>
template <typename Function>
void
parallel_exec(size_t concurrency, Function f)
{
std::vector<std::thread> threads;
threads.reserve(concurrency);
for (size_t i = 0; i < concurrency; ++i) {
threads.emplace_back(f, i);
}
for (auto &t : threads) {
t.join();
}
}
/*
* This function executes 'concurrency' threads and provides
* 'syncthreads' method (synchronization barrier) for f()
*/
template <typename Function>
void
parallel_xexec(size_t concurrency, Function f)
{
std::condition_variable cv;
std::mutex m;
std::unique_ptr<size_t> counter =
std::unique_ptr<size_t>(new size_t(0));
auto syncthreads = [&] {
std::unique_lock<std::mutex> lock(m);
(*counter)++;
if (*counter < concurrency)
cv.wait(lock, [&] { return *counter >= concurrency; });
else
/*
* notify_call could be called outside of a lock
* (it would perform better) but drd complains
* in that case
*/
cv.notify_all();
};
parallel_exec(concurrency, [&](size_t tid) { f(tid, syncthreads); });
}
/*
* This function executes 'concurrency' threads and wait for all of them to
* finish executing f before calling join().
*/
template <typename Function>
void
parallel_exec_with_sync(size_t concurrency, Function f)
{
parallel_xexec(concurrency,
[&](size_t tid, std::function<void(void)> syncthreads) {
f(tid);
syncthreads();
});
}
#endif /* THREAD_HELPERS_COMMON_HPP */
|
