One question: Since nbench is a cpu benchmark, shouldn't you have used
-O0 (no optimization) instead? Using -O2 could potentially skew
results as it would optimize the code which can work better on a
particular (class of) processor, but not much better on other
processors. Just a thought. Every benchmark I've ran before, I've
compiled with -O0 to avoid any skewing.
> port document:
http://weiffun.blogspot.com/2009/04/port-nbench-byte-223-to-minix314....
>
> Regards