AMD XP 3200+ benchmarks disable Intel hyperthreading

CHIP FIRMS always put the best spin possible on their benchmark results, but the latest figures AMD has printed on its web site go a spin too far, we'd suggest. Although to be absolutely fair about this, Intel hasn't exactly covered itself in glory in the past.

Like the lion and the unicorn - the unkinder comparison is Tweedledum and Tweedledee -- Intel and AMD are engaged in a continued and bitter struggle to claim the performance crown, particularly on desktop microprocessors.

The figures and graphs on the AMD web site this morning, here, compare the Pentium 4 3.0GHz with hyperthreading disabled, against the XP 3000+ and the just released 3200+.

They are "normalised" to the Pentium 4 3GHz with hyperthreading disabled for three benchmarks - Etesting, ZD and Bapco Sysmark 2001 Internet Content Creation.

3D Gaming tests also disable hyperthreading.

Now we've never been entirely sure what benefits hyperthreading actually gives, as Intel has been oddly reluctant to provide us with anything other than benchmarks of a non-hyperthreaded P4 against a hyperthreaded P4, so perhaps Chipzilla is hoist on its own petard by this one.

But at the bottom of the AMD page there's a link to the configuration it uses to reach its figures. Here we see it compares the P4 3GHz chip using its own Canterwood D875PBZ board - not the most exciting i875 board on the planet. The AMD board is a far more exciting Asus A7N8X 2.00 motherboard. And AMD uses its own software which is unavailable to the public.

Some would argue that the "performance crown" is made of tinsel, not gold, these days.

Both the chip vendors, we'd suggest, are not only tilting at windmills, but aren't really giving any objective view of why each thinks it deserves this tinsel crown. µ

See Also
Why CPU benchmarks are good and bad
Did AMD's Athlon XP 3000+ earn its rating?
On benchmarks, on bias, on the Beeb, on Linux, on the Xbox Letters