Re: Speeding up debian ... ?
On Mon, Jun 09, 2003 at 09:03:58AM -0500, Ray wrote:
> On Saturday 07 June 2003 16:04, Chris Metzler wrote:
> > On Sat, 7 Jun 2003 15:43:47 +0200
> >
> > aradorlinux@yahoo.es wrote:
> > > On Sat, 07 Jun 2003 03:29:37 +0100
> > >
> > > David selby <debian@pusspaws.net> wrote:
> > >> Would there be much of a speed increase, enough to warrent doing it ?
> > >
> > > Yes.
> > >
> > > (You can listen people saying "no"; but those people can't prove
> > > how its posible that a optimized compilation of some apps seems to make
> > > a diference in the real world)
> >
> > Since you're so confident that it makes a significant difference for
> > typical workstation users (who are rarely constrained by CPU speed,
> > instead being mostly constrained by I/O speed), I eagerly await your
> > numbers from a battery of real-world tests.
> >
>
> i recompiled my vorbis-tools, libogg0, libvorbis0.
> when playing oggs it was taking up 3-5% cpu, now its taking 0-1%
> i didn't look at what it was doing before when encoding, but it 'feels'
> faster, encoding each cd in 15 minutes (after ripping), before i was encoding
> & ripping in the same step, and it was doing it in about an hour each cd, but
> that isn't a fair test since they where in different settings.
Rebuilding Mozilla with optimizations tuned to my CPU seems to have
reduced its mean load time from ~6.3 seconds to ~4.9, and makes a very
complex html page (CSS and lots of nested proportional tables) render
in an average of ~3.2 seconds instead of ~4.7.
The above are on a 2.0GHz Athlon 2400 with 1 gig of PC2700 RAM,
running 2.4.20-k7 under sid. The timings were taken manually with a
stopwatch, but all samples but one were within .3 seconds of the
median.
Before timing, I loaded and exited Mozilla twice to help ensure all
Mozilla files were cached and to ensure that I knew what a completed
load looked like. The directory with the html and images was on disk,
and I refreshed twice before taking timings.
I don't know how other packages or architectures fare, and there's
probably a better way to benchmark this, but Mozilla seems to benefit
with these tests on my system.
When I was using an Amiga 3000 with a 68040 accelerator earlier this
year, recompiling ssh &c knocked about 28 seconds off the time to
negotiate a session, by a Mississippi count. I don't have the Amiga
anymore to test that more thoroughly, and I have a feeling the mostly
x86-centric folks are rolling their eyes about now. :-)
Reply to: