如何在Linux中创建一个高分辨率计时器来衡量程序的性能？ [英] How to create a high resolution timer in Linux to measure program performance?

问题描述

我想GPU比作CPU的性能。对于NVIDIA GPU我一直在使用cud​​aEvent_t类型得到一个非常precise时机。

I'm trying to compare GPU to CPU performance. For the NVIDIA GPU I've been using the cudaEvent_t types to get a very precise timing.

有关的CPU我一直在使用下面的code：

For the CPU I've been using the following code:

// Timers
clock_t start, stop;
float elapsedTime = 0;

// Capture the start time

start = clock();

// Do something here
.......

// Capture the stop time
stop = clock();
// Retrieve time elapsed in milliseconds
elapsedTime = (float)(stop - start) / (float)CLOCKS_PER_SEC * 1000.0f;

显然，那块code仅为code，如果你以秒​​为单位计数。此外，结果有时出来挺奇怪的。

Apparently, that piece of code is only code if you're counting in seconds. Also, the results sometime come out quite strange.

有谁知道的一些方法来在Linux中创建一个高分辨率计时器？

Does anyone know of some way to create a high resolution timer in Linux?

推荐答案

查看<一个href=\"http://www.kernel.org/doc/man-pages/online/pages/man2/clock_gettime.2.html\"><$c$c>clock_gettime,这是一个POSIX接口，高分辨率的计时器。

Check out clock_gettime, which is a POSIX interface to high-resolution timers.

如果，看了手册页，你在想关于 CLOCK_REALTIME 和 CLOCK_MONOTONIC 之间的区别，见<一href=\"http://stackoverflow.com/questions/3523442/difference-between-clock-realtime-and-clock-monotonic\">Difference CLOCK_REALTIME和CLOCK_MONOTONIC之间？

If, having read the manpage, you're left wondering about the difference between CLOCK_REALTIME and CLOCK_MONOTONIC, see Difference between CLOCK_REALTIME and CLOCK_MONOTONIC?

一个完整的示例，请参见以下页面：<一href=\"http://www.guyrutenberg.com/2007/09/22/profiling-$c$c-using-clock_gettime/\">http://www.guyrutenberg.com/2007/09/22/profiling-$c$c-using-clock_gettime/

See the following page for a complete example: http://www.guyrutenberg.com/2007/09/22/profiling-code-using-clock_gettime/

这篇关于如何在Linux中创建一个高分辨率计时器来衡量程序的性能？的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！