Machine performance

[Steve Shockley]

What's the biggest bottleneck when compiling the ports tree?  I've got 
two machines available; the first is a P4-2.4, 1gb RAM, with two 15k 
disks on an Adaptec 39160 SCSI (ports tree on its own spindle), and a 
quad Xeon 500/2mb, 3gb EDO RAM, 5x 10k disks on an LSI RAID controller 
with 128mb cache.

I guess the questions boil down to: which benefits compiling ports more, 
disk or processor?  Would compiling ports actually use four processors?