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
|
// Copyright (c) 2015-2016 The Bitcoin Core developers
// Distributed under the MIT software license, see the accompanying
// file COPYING or http://www.opensource.org/licenses/mit-license.php.
#include "bench.h"
#include "perf.h"
#include <assert.h>
#include <iostream>
#include <iomanip>
benchmark::BenchRunner::BenchmarkMap &benchmark::BenchRunner::benchmarks() {
static std::map<std::string, benchmark::BenchFunction> benchmarks_map;
return benchmarks_map;
}
benchmark::BenchRunner::BenchRunner(std::string name, benchmark::BenchFunction func)
{
benchmarks().insert(std::make_pair(name, func));
}
void
benchmark::BenchRunner::RunAll(benchmark::duration elapsedTimeForOne)
{
perf_init();
if (std::ratio_less_equal<benchmark::clock::period, std::micro>::value) {
std::cerr << "WARNING: Clock precision is worse than microsecond - benchmarks may be less accurate!\n";
}
std::cout << "#Benchmark" << "," << "count" << "," << "min(ns)" << "," << "max(ns)" << "," << "average(ns)" << ","
<< "min_cycles" << "," << "max_cycles" << "," << "average_cycles" << "\n";
for (const auto &p: benchmarks()) {
State state(p.first, elapsedTimeForOne);
p.second(state);
}
perf_fini();
}
bool benchmark::State::KeepRunning()
{
if (count & countMask) {
++count;
return true;
}
time_point now;
uint64_t nowCycles;
if (count == 0) {
lastTime = beginTime = now = clock::now();
lastCycles = beginCycles = nowCycles = perf_cpucycles();
}
else {
now = clock::now();
auto elapsed = now - lastTime;
auto elapsedOne = elapsed / (countMask + 1);
if (elapsedOne < minTime) minTime = elapsedOne;
if (elapsedOne > maxTime) maxTime = elapsedOne;
// We only use relative values, so don't have to handle 64-bit wrap-around specially
nowCycles = perf_cpucycles();
uint64_t elapsedOneCycles = (nowCycles - lastCycles) / (countMask + 1);
if (elapsedOneCycles < minCycles) minCycles = elapsedOneCycles;
if (elapsedOneCycles > maxCycles) maxCycles = elapsedOneCycles;
if (elapsed*128 < maxElapsed) {
// If the execution was much too fast (1/128th of maxElapsed), increase the count mask by 8x and restart timing.
// The restart avoids including the overhead of this code in the measurement.
countMask = ((countMask<<3)|7) & ((1LL<<60)-1);
count = 0;
minTime = duration::max();
maxTime = duration::zero();
minCycles = std::numeric_limits<uint64_t>::max();
maxCycles = std::numeric_limits<uint64_t>::min();
return true;
}
if (elapsed*16 < maxElapsed) {
uint64_t newCountMask = ((countMask<<1)|1) & ((1LL<<60)-1);
if ((count & newCountMask)==0) {
countMask = newCountMask;
}
}
}
lastTime = now;
lastCycles = nowCycles;
++count;
if (now - beginTime < maxElapsed) return true; // Keep going
--count;
assert(count != 0 && "count == 0 => (now == 0 && beginTime == 0) => return above");
// Output results
// Duration casts are only necessary here because hardware with sub-nanosecond clocks
// will lose precision.
int64_t min_elapsed = std::chrono::duration_cast<std::chrono::nanoseconds>(minTime).count();
int64_t max_elapsed = std::chrono::duration_cast<std::chrono::nanoseconds>(maxTime).count();
int64_t avg_elapsed = std::chrono::duration_cast<std::chrono::nanoseconds>((now-beginTime)/count).count();
int64_t averageCycles = (nowCycles-beginCycles)/count;
std::cout << std::fixed << std::setprecision(15) << name << "," << count << "," << min_elapsed << "," << max_elapsed << "," << avg_elapsed << ","
<< minCycles << "," << maxCycles << "," << averageCycles << "\n";
std::cout.copyfmt(std::ios(nullptr));
return false;
}
|
