In a flurry of articles, writers across the Web are finding issue
with the recently released benchmarks for Apple's new Power Mac G5.
I've read all of these (ignoring all the parts too technical for a
mere Mac user to understand) and come up with a summary of all the
things Apple would have to do to prove to the Wintel world that Macs
are truly faster. Apparently, double the speed on Quake, Photoshop, and
Mathematica just isn't enough; when you're in the minority, you have to
do everything twice a good just to stay even in the game
. . . uh, wait a minute.
Anyway, here is the Lite Side's
Benchmarks to Finally Convince Wintel Pundits that Macs are
Faster
First, we'll have to level the playing field by letting each company
write its own compiler, making the results essentially unique to each
machine and not comparable across platforms.
The pundits say that Apple cheated.
Solution: To convince the Wintel Pundits, the Mac group has
to write their compiler in the dark, using only their left hands, in
less than 30 minutes. The Dell group, not operating under a time limit,
must be twice as large as the Mac group and get free donuts for life.
Then, if the Mac is faster, we'll know it isn't because Apple cheated
on the benchmarks.
Second, to be fair, shouldn't we compare the latest Wintel hardware
to Apple's prototypes? The Pundits weigh in, "Isn't it unfair that
Apple used a non-shipping machine while Dell's machine wasn't even
using the newest 3.2 GHz P4?"
Solution: When a year-old Mac can beat a prototype Wintel
machine just off the design table, then maybe the Wintel Pundits will
admit Macs are fast. Maybe.
What about other non-Intel processors? Where are the Athlon
benchmarks? The pundits respond that Athlon kicks Intel's butt, and now
you know why Apple ignores the Athlon.
Solution: Wintel pundits will only admit Macs are fast when
Macs are faster than every other computer in existence. No exceptions!
Including distributed computing networks and supercomputer facilities
with massive parallel processors. Get on that, willya Steve?
What about the fact that in the Photoshop, Mathematica, and Quake
bake-offs, the Mac outperformed the PC by a factor of two? The Pundits
respond, "Stand back, ye puny mortal, whilst I wield my mighty shield
of Altivex."
Solution: When the Mac is designed to allow AltiVec to run
backwards, making everything that calls on it twice as slow as when it
is activated - and the Mac is still faster than the PC - then maybe
we'll hear the end of this.
But Mathematica and Quake aren't optimized for AltiVec, are they?
The Pundits respond, "Stand back, ye puny mortal, whilst I wield my
mighty shield of technobabble. GCC! SPEC! Compiler!
Hyperthreading!"
Solution: As we all know from Star Trek, the only thing that
defeats technobabble is . . . more technobabble! We need to
introduce these guys to a plasma leak right quick.
In conclusion, this will bang around the office for a while, but
when early production models arrive, it'll resurface as we see some
real-world benchmarks. Wintel Pundits will not believe these benchmarks
until they can talk their powers that be into providing a test
machine.
Look at it this way: If the Mac is slower, they get kudos from the
publishers of their respective rags (Apple lies about
benchmarks! Turn to page 11). If the Mac really is faster, they get the
world's fastest computer to play with. It's a win-win scenario.
Put simply: If Apple wants Wintel Pundits to provide an "unbiased"
point of view, a few early prototypes sent to the review departments of
various PC magazines wouldn't hurt.
