Profiling

• There are many easy ways to make your program run faster. One is to use your compiler's optimization switches: just add a -O, -O1, -O2, or -O3 to the command line. This will cause the compiler to spend more time compiling, with the result of faster and/or smaller generated code.

• The only problem with optimization is that it isn't very compatable with debugging. Most C compilers don't allow you to use the -O and -g switches at the same time, but gcc is an exception. If I may quote from the gcc man page, Unlike most other C compilers, GNU CC allows you to use `-g' with `-O'. The shortcuts taken by optimized code may occasionally produce surprising results: some variables you declared may not exist at all; flow of control may briefly move where you did not expect it; some statements may not be executed because they compute constant results or their values were already at hand; some statements may execute in different places because they were moved out of loops.

• Many compilers, including gcc, even allow you to turn on and off their specific optimizations. For example, -finline-functions, -ffast-math, and -funroll-loops are all valid gcc compile switches.

• Another way to speed up your program is to use profiling. Profiling records the ammount of time your program spends in its various parts (where part is usually defined as a function or basic block). To have your program generate profiling information as it runs, compile it with the -p (or -pg) option. This information is kept in a mon.out (or gmon.out) file as your program runs, and then can be interpreted by the prof (or gprof) programs.

  Many compilers (but not gcc) can use profiling information to fine-tune their optimizations. Profiling information is also useful in determining which parts of your program are worth speeding up and which parts aren't.