[ILUG] a quick gcc microoptimisation question

[P at draigBrady.com]

jm at jmason.org wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> 
> HiLUG,
> 
> quickie: when using gcc's -O switches, what is more efficient:
> 
>     if (condition) {
>       /* more common case */
>     } else {
>       /* less common case */
>     }
> 
> or 
> 
>     if (condition) {
>       /* less common case */
>     } else {
>       /* more common case */
>     }

I don't know what the default is but you can give branch
prediction hints with __builtin_expect() in gcc >= 3,
and playing with that and looking at the resulting assembly
should show the default behaviour.

I find the __[un]likely macros from the linux kernel
the nicest interface to this:

/* __likely etc. is to provide the compiler
    with branch prediction information.
    Search for builtin_expect in `info gcc` */
#if __GNUC__ < 3
     #define __builtin_expect(foo,bar) (foo)
     #define expect(foo,bar) (foo)
#else
     #define expect(foo,bar) __builtin_expect((long)(foo),bar)
#endif
#define __likely(foo) expect((foo),1)
#define __unlikely(foo) expect((foo),0)
 

Note I think tests for NULL (!something) don't need an unlikely
as gcc does that by default itself?

-- 
Pádraig Brady - http://www.pixelbeat.org
--- Following generated by rotagator ---

Changing priority of a process

The nice value (-20 to 19) represents the priority of a process.
The lower the value, the higher the priority (not as nice to others).
For non interactive processes (especially on multi-user systems) do:
   nice low_priority_command
You can retroactively give a process less priority like:
   renice 19 -p low_priority_pid
For e.g. this is useful at the start of long running shell scripts:
   renice 19 -p $$ > /dev/null
Only root can give processes more priority.
--