On 12 Jul 2001 22:00:30 +0200, Thierry Vignaud wrote:
> Blue Lizard <[EMAIL PROTECTED]> writes:
>
> > > P.S: "this feature is slow" isn't a bug report ;-)
> >
> > Obviously the code for new features is gonna be in debug mode for a
> > little while (ie, not aimed at efficiency) but would you care for us to
> > break with the 'only fix what you're gonna ship' policy and do some
> > profiling or other perf drudgery for you?
>
> any profiling report that show where cpu cycles are lost is welcome
>
>
Your toungue hath beat your head. Think of what you say. On this box,
here, the memory is not load balanced by thousands of individual
processors just as capable as the 'main' one. The memory is faster than
the swap file. The less data that needs to be read, the faster the read
is complete. The more hanging mallocs, the more redundant symbols, the
more magnets on my motherboard. All have an affect on speed. Measuring
and profiling can be done on inlined functions in case you've never
noticed. Getting thru the smoke?
