source vs. binary benchmarks

rs · n00b Joined: 20 Jun 2002 Posts: 4

Hi!

Did anybody do some benchmarking comparing the speed of apps compiled by Gentoo versus apps coming from some binary distribution?

It would be interesting to know how big the speed increase really is. It would also be very interesting how much the different CPU-families (P3, P4, Athlon, PPC) benefit from optimizations, for example I would guess that P4 should show the biggest difference between pre-compiled and custom-compiled while P3 should show the smallest.

So if you've done some benchmarking (even the different boot-times would be interesting) please post them.

Thanks!

klieber · Posted: Thu Jun 20, 2002 12:27 pm Post subject:

I haven't heard of anyone doing such benchmarks, at least not on a gentoo system. You might run some yourself and post the results -- I'm sure folks here would be interested in them.

--kurt
_________________
The problem with political jokes is that they get elected

AutoBot · l33t Joined: 22 Apr 2002 Posts: 968 Location: Usually Out

I believe the more under-powered your processor is the better results you have compiling from source with optimizations.

Time is your friend, try timing gzipping a large sum of files with gentoo then on another distro on the same pc.
_________________
This message self destructed a long time ago.

delta407 · Posted: Thu Jun 20, 2002 9:51 pm Post subject:

Right, but the more underpowered your processor it is, the less likely system-specific optimizations will make a difference. (If binaries are distributed as i386, and you compile for i486, there won't be nearly as much of a difference.)

Of course, -fomit-frame-pointer tends to speed things up a bunch, but still.

AutoBot · l33t Joined: 22 Apr 2002 Posts: 968 Location: Usually Out

Well I didn't mean that underpowered, I was talking more along the lines of 500mhz to 800mhz celeron/pIII/duron based pc's.
_________________
This message self destructed a long time ago.