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
|
//
// parallel_sort.cpp
// ~~~~~~~~~~~~~~~~~
//
// Copyright (c) 2003-2023 Christopher M. Kohlhoff (chris at kohlhoff dot com)
//
// Distributed under the Boost Software License, Version 1.0. (See accompanying
// file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
//
#include <boost/asio.hpp>
#include <boost/thread/thread.hpp>
#include <boost/asio/experimental/parallel_group.hpp>
#include <algorithm>
#include <chrono>
#include <functional>
#include <iostream>
#include <random>
template <
typename Executor,
typename RandomAccessIterator,
BOOST_ASIO_COMPLETION_TOKEN_FOR(void()) CompletionToken>
auto parallel_sort(
Executor executor,
RandomAccessIterator begin,
RandomAccessIterator end,
CompletionToken&& token);
template <
typename Executor,
typename RandomAccessIterator>
void parallel_sort_impl(
Executor executor,
RandomAccessIterator begin,
RandomAccessIterator end,
std::function<void()> continuation)
{
std::size_t n = end - begin;
if (n <= 16384)
{
boost::asio::post(executor,
[=]
{
std::sort(begin, end);
continuation();
}
);
}
else
{
boost::asio::experimental::make_parallel_group(
[=](auto token)
{
return parallel_sort(executor, begin, begin + n / 2, token);
},
[=](auto token)
{
return parallel_sort(executor, begin + n / 2, end, token);
}
).async_wait(
boost::asio::experimental::wait_for_all(),
[=](std::array<std::size_t, 2>)
{
std::inplace_merge(begin, begin + n / 2, end);
continuation();
}
);
}
}
template <
typename Executor,
typename RandomAccessIterator,
BOOST_ASIO_COMPLETION_TOKEN_FOR(void()) CompletionToken>
auto parallel_sort(
Executor executor,
RandomAccessIterator begin,
RandomAccessIterator end,
CompletionToken&& token)
{
return boost::asio::async_compose<CompletionToken, void()>(
[=](auto& self, auto... args)
{
if (sizeof...(args) == 0)
{
using self_type = std::decay_t<decltype(self)>;
parallel_sort_impl(executor, begin, end,
[self = std::make_shared<self_type>(std::move(self))]
{
boost::asio::dispatch(
boost::asio::append(
std::move(*self), 0));
}
);
}
else
{
self.complete();
}
},
token
);
}
int main()
{
boost::asio::thread_pool pool(4);
std::vector<int> values(100'000'000);
std::random_device random_device;
std::mt19937 rng(random_device());
std::uniform_int_distribution<int> dist(1, 1'000'000);
std::generate(values.begin(), values.end(), [&]{ return dist(rng); });
std::cout << "starting sort\n";
auto begin = std::chrono::high_resolution_clock::now();
parallel_sort(
pool.get_executor(),
values.begin(),
values.end(),
boost::asio::use_future
).get();
auto end = std::chrono::high_resolution_clock::now();
auto duration = end - begin;
std::cout << "sort took "
<< std::chrono::duration_cast<std::chrono::microseconds>(duration).count()
<< " microseconds\n";
}
|
