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
|
// Copyright Yahoo. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
#include <vespa/vespalib/util/cpu_usage.h>
#include <vespa/vespalib/util/benchmark_timer.h>
#include <vespa/vespalib/testkit/test_kit.h>
#include <thread>
using namespace vespalib;
bool verbose = false;
size_t loop_cnt = 10;
double budget = 0.25;
using Sampler = vespalib::cpu_usage::ThreadSampler;
//-----------------------------------------------------------------------------
void be_busy(duration d) {
if (d > 0ms) {
volatile int tmp = 123;
auto t0 = steady_clock::now();
while ((steady_clock::now() - t0) < d) {
for (int i = 0; i < 1000; ++i) {
tmp = (tmp + i);
tmp = (tmp - i);
}
}
}
}
std::vector<duration> sample(const std::vector<Sampler*> &list) {
std::vector<duration> result;
result.reserve(list.size());
for (auto *sampler: list) {
result.push_back(sampler->sample());
}
return result;
}
//-----------------------------------------------------------------------------
void verify_sampling(size_t thread_id, size_t num_threads, std::vector<Sampler*> &samplers, bool force_mock) {
if (thread_id == 0) {
TEST_BARRIER(); // #1
auto t0 = steady_clock::now();
std::vector<duration> pre_usage = sample(samplers);
TEST_BARRIER(); // #2
TEST_BARRIER(); // #3
auto t1 = steady_clock::now();
std::vector<duration> post_usage = sample(samplers);
TEST_BARRIER(); // #4
double wall = to_s(t1 - t0);
std::vector<double> load(4, 0.0);
for (size_t i = 0; i < 4; ++i) {
load[i] = to_s(post_usage[i] - pre_usage[i]) / wall;
}
EXPECT_GREATER(load[3], load[0]);
fprintf(stderr, "loads: { %.2f, %.2f, %.2f, %.2f }\n", load[0], load[1], load[2], load[3]);
} else {
int idx = (thread_id - 1);
double target_load = double(thread_id - 1) / (num_threads - 2);
auto sampler = cpu_usage::create_thread_sampler(force_mock, target_load);
samplers[idx] = sampler.get();
TEST_BARRIER(); // #1
TEST_BARRIER(); // #2
for (size_t i = 0; i < loop_cnt; ++i) {
be_busy(std::chrono::milliseconds(idx));
}
TEST_BARRIER(); // #3
TEST_BARRIER(); // #4
}
}
//-----------------------------------------------------------------------------
TEST_MT_F("require that dummy thread-based CPU usage sampling with known expected load works", 5, std::vector<Sampler*>(4, nullptr)) {
TEST_DO(verify_sampling(thread_id, num_threads, f1, true));
}
TEST_MT_F("require that external thread-based CPU usage sampling works", 5, std::vector<Sampler*>(4, nullptr)) {
TEST_DO(verify_sampling(thread_id, num_threads, f1, false));
}
TEST("measure thread CPU clock overhead") {
auto sampler = cpu_usage::create_thread_sampler();
duration d;
double min_time_us = BenchmarkTimer::benchmark([&d, &sampler]() noexcept { d = sampler->sample(); }, budget) * 1000000.0;
fprintf(stderr, "approx overhead per sample (thread CPU clock): %f us\n", min_time_us);
}
//-----------------------------------------------------------------------------
int main(int argc, char **argv) {
TEST_MASTER.init(__FILE__);
if ((argc == 2) && (argv[1] == std::string("verbose"))) {
verbose = true;
loop_cnt = 1000;
budget = 5.0;
}
TEST_RUN_ALL();
return (TEST_MASTER.fini() ? 0 : 1);
}
|
