summaryrefslogtreecommitdiffstats
path: root/src/boost/libs/compute/perf/perf_partition_point.cpp
blob: 91261aa1d2f34f926c2aa3c844042f4ca42a0450 (plain)
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
//---------------------------------------------------------------------------//
// Copyright (c) 2014 Roshan <thisisroshansmail@gmail.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
//
// See http://boostorg.github.com/compute for more information.
//---------------------------------------------------------------------------//

#include <algorithm>
#include <iostream>
#include <numeric>
#include <vector>

#include <boost/compute/system.hpp>
#include <boost/compute/lambda.hpp>
#include <boost/compute/algorithm/partition.hpp>
#include <boost/compute/algorithm/partition_point.hpp>
#include <boost/compute/container/vector.hpp>

#include "perf.hpp"

int rand_int()
{
    return static_cast<int>((rand() / double(RAND_MAX)) * 25.0);
}

int main(int argc, char *argv[])
{
    perf_parse_args(argc, argv);
    std::cout << "size: " << PERF_N << std::endl;

    // setup context and queue for the default device
    boost::compute::device device = boost::compute::system::default_device();
    boost::compute::context context(device);
    boost::compute::command_queue queue(context, device);
    std::cout << "device: " << device.name() << std::endl;

    // create vector of random numbers on the host
    std::vector<int> host_vector(PERF_N);
    std::generate(host_vector.begin(), host_vector.end(), rand_int);

    // create vector on the device and copy the data
    boost::compute::vector<int> device_vector(PERF_N, context);
    boost::compute::copy(
        host_vector.begin(), host_vector.end(), device_vector.begin(), queue
    );

    using boost::compute::_1;
    boost::compute::partition(
        device_vector.begin(), device_vector.end(), _1 < 20, queue
    );


    perf_timer t;
    for(size_t trial = 0; trial < PERF_TRIALS; trial++){
        t.start();
        boost::compute::partition_point(
            device_vector.begin(), device_vector.end(), _1 < 20, queue
        );
        queue.finish();
        t.stop();
    }
    std::cout << "time: " << t.min_time() / 1e6 << " ms" << std::endl;

    return 0;
}