Re: [PATCH] (x)memcoll: performance improvement when input is known to be NUL delimited.

[Pádraig Brady]

On 10/03/10 07:05, Chen Guo wrote:
> Hi Bruno,
> I tried with the checks for the presence of that last NUL byte like you 
> suggested,
> and to my surprise on the remote machine I was testing on it was consistently
> (as in every run for 20 runs) faster, and thus I've included it in the patch.
> This is not what I expected at all

How much faster?
Could it be due to cache prefetching?
I would try different sized strings to test (32,65,1025,...)

Also I think that when you do
   if (!something)
gcc will do
   if (__unlikely(!something))
http://www.pixelbeat.org/programming/gcc/c_c++_notes.html#branch_predict

cheers,
Pádraig.