{"id":86,"date":"2012-05-13T14:47:00","date_gmt":"2012-05-13T14:47:00","guid":{"rendered":"http:\/\/finaldie.com\/wordpress\/?p=86"},"modified":"2012-11-08T09:33:19","modified_gmt":"2012-11-08T09:33:19","slug":"profile-with-callgrind","status":"publish","type":"post","link":"https:\/\/finaldie.com\/blog\/profile-with-callgrind\/","title":{"rendered":"Profile with callgrind"},"content":{"rendered":"

SUMMARY<\/h1>\n

There are still three articles about introducing the way to analyse\u00a0a program. But today, I’ll show you another member: callgrind<\/strong>. It’s a tool which belongs to\u00a0valgrind<\/a>.<\/p>\n

WHY<\/h1>\n

Sometimes maybe you encounter the issue as same as me: I don’t want to re-compile the program and the oprofile cannot work in my system as well, so.. uh…. what should I do?<\/p>\n

Don’t worry guys, the valgrind can save you :D. Why I say that? Because when you write a program in C\/C++, you must choose a kernel which can support valgrind working well. As we know, valgrind can detect memory errors easily, it’s a great helper for us. So, if all profilers cannot work well, at least, we have the valgrind. :D<\/p>\n

HOW<\/h1>\n

The way is simple:<\/p>\n

    \n
  1. valgrind –tool=callgrind .\/yourbin<\/li>\n
  2. gprof2dot.py<\/a> -f callgrind callgrind.out.xxx | dot -Tpng -o output.png<\/li>\n<\/ol>\n

    THE SAMPLE<\/h1>\n

    Here is a sample of analysis mread<\/a> program, which written by cloudwu.<\/p>\n

    \"call<\/a><\/p>\n

    GUI TOOL<\/h1>\n

    There is GUI front end tool — kcachegrind<\/a>, if you want to analyse the program more\u00a0meticulous, you can try it :D<\/p>\n

    THE LAST<\/h1>\n

    In the end, there are four articles for introducing how to analyse a C\/C++ program:<\/p>\n