zoukankan      html  css  js  c++  java
  • Profile your program using GNU gprof

    Profiling is an indispensable measure for analyzing and optimizing the performance of your program. A typical profiler like GNU gprof will complete this task by running your program after a special compilation.

    There are 3 forms of the profiling analysis:

    • flat profile: how much time the program spent in each function and how many times the function was called.

    • call graph: for each function, which function called it and which other functions it called, and how many times.

    • annotated source: a listing of the program source code, which is labeled with the number of times each line of the program was executed.

    Compiler options for generating the code which can be profiled by gprof:

    • -pg: enable profiling by gprof, which is always mandatory.

    • -g: enable annotated source profiling.

    • -a: enable basic-block counting, which is required by the annotated source profiling. However, this option does not appear in gcc or g++.

    Options for gprof command

    • -p: generate flat profile.

    • -q: generate call graph.

    • -l: enable line-by-line profiling.

    • -s: summarize profile data from different files into gmon.sum.

    Steps for using gprof:

    1. Compile the program for gprof:

      g++ -std=gnu++11 -g -O2 -o output input.cpp -gp
    2. Run the program: ./output1. After execution, the profiling data will be stored in the file gmon.out.

    3. Call gprof command with the generated gmon file as argument to generate profiling reports.

  • 相关阅读:
    hdu--4027--不错的线段树
    hdu--3275--线段树<again>
    hdu--2795--又是线段树
    hdu--4407--一不留神就TLE了
    zoj--3822--第二题概率dp
    hdu--3911--线段树<我最近爱上她了>
    hdu--1710--二叉树的各种遍历间的联系
    hdu--1712--分组背包<如果你真的明白了背包..>
    hdu--4576--概率dp<见过最简单的概率dp>
    list remove object
  • 原文地址:https://www.cnblogs.com/peabody/p/15351661.html
Copyright © 2011-2022 走看看