Nat! bio photo


Senior Mull.

Twitter RSS


G5: A real life test

I downloaded textutils2.1 from GNU/Linux :) and ran the ./configure ; make on it.
This is on a G5 running some NDA OS using two Seagate Barracuda drives RAIDed together in stripe configuration. The development environment on the G5 is gcc 3.3 and whatever came with the installed NDA toolkit.

I made a little shell script which I ran with time ./

#! /bin/sh
make distclean > log 2>&1 
./configure >> log 2>&1
make >> log 2>&1 

I had the clock in the menubar turned off and I set Terminal to not show the running process in the windows title, as this is bound to waste some cycles. I checked with top before, that the system otherwise appeared quiet. I ran it three times and used the lowest numbers

real 1m35.1 user 0m37.3 sys 1m4.1

Apparently the dual processor configuration is helping a little bit here, because the combined user and sys time is actually more than the real time.

it would be interesting to contrast this with a non-RAIDed Dual-G5 and of course with some Athlon Linux box. So if you have the time send me some results. This are the results of my Apple Cube, my main development machine for the last couple of years with a WDC WD205AA 20GB/5400 drive

real 5m15.1 user 2m34.9 sys 2m28.5