Commit 8baa8cf8 authored by gabime's avatar gabime

fixed latency test

parent a776a774
......@@ -32,7 +32,7 @@ void bench_mt(int howmany, std::shared_ptr<spdlog::logger> log, int thread_count
int main(int , char *[])
{
std::srand(std::time(nullptr)); // use current time as seed for random generator
int howmany = 50;
int howmany = 10000;
int queue_size = howmany + 2;
int threads = 10;
int file_size = 30 * 1024 * 1024;
......@@ -91,12 +91,6 @@ int main(int , char *[])
return EXIT_SUCCESS;
}
void rand_sleep()
{
auto sleep_ms= std::chrono::milliseconds(std::rand() % 50);
std::this_thread::sleep_for(sleep_ms);
}
void bench(int howmany, std::shared_ptr<spdlog::logger> log)
{
......@@ -113,7 +107,6 @@ void bench(int howmany, std::shared_ptr<spdlog::logger> log)
log->info("Hello logger: msg number {}", i);
auto delta_nanos = chrono::duration_cast<nanoseconds>(high_resolution_clock::now() - start);
total_nanos+= delta_nanos;
rand_sleep();
}
auto avg = total_nanos.count()/howmany;
......@@ -129,20 +122,17 @@ void bench_mt(int howmany, std::shared_ptr<spdlog::logger> log, int thread_count
cout << log->name() << "...\t\t" << flush;
vector<thread> threads;
nanoseconds::rep total_nanos = 0;
std::atomic<nanoseconds::rep> total_nanos{0};
for (int t = 0; t < thread_count; ++t)
{
threads.push_back(std::thread([&]() {
nanoseconds thr_total_nanos = nanoseconds::zero();
for (int j = 0; j < howmany / thread_count; j++)
{
auto start = high_resolution_clock::now();
log->info("Hello logger: msg number {}", j);
auto delta_nanos = chrono::duration_cast<nanoseconds>(high_resolution_clock::now() - start);
thr_total_nanos+= delta_nanos;
rand_sleep();
total_nanos+= delta_nanos.count();
}
total_nanos+= thr_total_nanos.count();
}));
}
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment