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
135
136
137
138
139
140
141
142
143
|
// Copyright Vespa.ai. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
#include <vespa/vespalib/util/time.h>
#include <vespa/vespalib/util/require.h>
#include <vespa/vespalib/util/executor.h>
#include <vespa/vespalib/util/threadstackexecutor.h>
#include <vespa/vespalib/locale/c.h>
#include <thread>
#include <cmath>
using namespace vespalib;
using namespace std::literals;
uint32_t doStuff(uint32_t input) {
char buf[128];
for (uint32_t i = 0; i < sizeof(buf); ++i) {
buf[i] = ((input + i) * i) & 0xff;
}
uint32_t result = 0;
for (uint32_t i = 0; i < sizeof(buf); ++i) {
result += ((buf[i] * i) + input) & 0xff;
}
return result;
}
struct CPUTask : public Executor::Task {
uint32_t taskSize;
uint32_t &result;
CPUTask(uint32_t size, uint32_t &res) : taskSize(size), result(res) {}
void run() override {
uint32_t res = 0;
for (uint32_t i = 0; i < taskSize; ++i) {
res += doStuff(i);
}
result += res;
}
};
struct SyncTask : public Executor::Task {
Gate &gate;
CountDownLatch &latch;
SyncTask(Gate &g, CountDownLatch &l) : gate(g), latch(l) {}
void run() override {
latch.countDown();
gate.await();
}
};
class Test
{
private:
uint32_t _result;
public:
Test() : _result(0) {}
uint32_t calibrate(double ms);
void stress(Executor &executor, uint32_t taskSize, uint32_t numTasks);
void eat_value(uint32_t result) { _result += result; }
};
uint32_t
Test::calibrate(double wanted_ms)
{
uint32_t n = 0;
vespalib::steady_time t0;
vespalib::steady_time t1;
{ // calibration of calibration loop
uint32_t result = 0;
t0 = vespalib::steady_clock::now();
vespalib::duration dur;
do {
result += doStuff(++n);
t1 = vespalib::steady_clock::now();
dur = t1 - t0;
} while (dur < 1s);
_result += result;
}
{ // calibrate loop
t0 = vespalib::steady_clock::now();
uint32_t result = 0;
for (uint32_t i = 0; i < n; ++i) {
result += doStuff(i);
}
_result += result;
t1 = vespalib::steady_clock::now();
}
std::chrono::duration<double,std::milli> ms = t1 - t0;
double size = (double(n) / ms.count()) * wanted_ms;
return (uint32_t) std::round(size);
}
int main(int argc, char **argv) {
if (argc != 4) {
fprintf(stderr, "Usage: %s <threads> <ms per task> <tasks>\n",
argv[0]);
return 1;
}
Test test;
uint32_t threads = atoi(argv[1]);
double ms_per_task = locale::c::strtod(argv[2], 0);
uint32_t tasks = atoi(argv[3]);
fprintf(stderr, "threads : %u\n", threads);
fprintf(stderr, "ms per task: %g\n", ms_per_task);
fprintf(stderr, "tasks : %u\n", tasks);
{
fprintf(stderr, "calibrating task size...\n");
uint32_t taskSize = test.calibrate(ms_per_task);
fprintf(stderr, "calibrated task size: %u\n", taskSize);
ThreadStackExecutor executor(threads, 5000 + threads);
{
Gate gate;
CountDownLatch latch(threads);
for (uint32_t i = 0; i < threads; ++i) {
Executor::Task::UP res
= executor.execute(Executor::Task::UP(
new SyncTask(gate, latch)));
REQUIRE(res.get() == 0);
}
latch.await();
gate.countDown();
executor.sync();
fprintf(stderr, "all threads have been accounted for...\n");
}
{
vespalib::Timer t0;
fprintf(stderr, "starting task submission...\n");
uint32_t result = 0;
for (uint32_t i = 0; i < tasks; ++i) {
Executor::Task::UP t(new CPUTask(taskSize, result));
t = executor.execute(std::move(t));
while (t) {
std::this_thread::sleep_for(10ms);
t = executor.execute(std::move(t));
}
}
executor.sync();
double ms = vespalib::count_ms(t0.elapsed());
fprintf(stderr, "total execution wall time: %g ms\n", ms);
test.eat_value(result);
}
}
return 0;
}
|
