Re: build speeds with current vs netbsd-4

[Andrew Doran <ad%netbsd.org@localhost>]

On Tue, May 20, 2008 at 12:10:20PM +0200, Zafer Aydogan wrote:
> 2008/5/20 Simon Burge <simonb%netbsd.org@localhost>:
> > While trying to track down a problem, I've been booting netbsd-4
> > and -current on an 8 core amd64 box (dual quad-core).  I thought I'd
> > test the time to build an i386 -j16 release with build.sh.  Both
> > tests were done immediately after a reboot.
> >
> >  netbsd-4 building -current i386 release:
> >  5905.500u 15390.524s 49:52.07 711.7%    0+0k 5209+1342018io  4820pf+0w
> >
> >  -current (4.99.62 from 20080513) building -current i386 release:
> >  5363.027u  2573.479s 24:31.86 539.2%    0+0k 5254+ 452978io 55128pf+0w
> >
> > netbsd-4 is using GENERIC.MP
> > -current is using GENERIC + options DEBUG/DIAGNOSTIC/KMEMSTATS
> >
> > Most impressive difference in results, especially the system CPU time
> > component.  Looks like all the recent work that has been happening is
> > really paying off!
> >
> > Cheers,
> > Simon.
> >
> 
> I wonder what the impact of DEBUG is.
> Can you try once without DEBUG and DIAGNOSTIC ?

The debug options will cripple it!

Andrew