As Alan mentioned, Slashdot had a rare original article
on the subject, where admin CmdrTaco interviewed the
head of Apple Engineering.  Turns out that hyperthreading
was disabled on the PCs because it made them
run slower on the benchmarks.  SSE2 was turned on, not
off, according to the details of their test report.
They also discuss the decision
to use gcc, which is already optimized for x86 and not the G5,
as the app for the benchmark.  The engineer also cheerfully
offered to do the test again with a Windows compiler, while
admitting that the Intel compiler (not used) is significantly faster
than gcc.

Personally, as someone who's been using a 400MHz G4 as a cross-platform
file server, web server, and you name it, all with the consumer version
of Mac
OS X, for years (256MB 100mhz RAM!), with virtually zero downtime, I'm
amused by all this hardware mania.  It will be interesting to see what
happens
next year when they scale up to dual 3GHz and Windows still won't run
on any
64-bit platform at all....



Here's the interview link (and a long discussion)

http://apple.slashdot.org/apple/03/06/24/2154256.shtml?tid=126&tid=181