Wuh-oh-- sounds like there might be some trouble a-brewin'. Unless you're pretty new to this game, you probably recall that when Apple unveiled the original Power Mac G4 back in 1999, Steve made a big deal about how it was the first personal "supercomputer," based on its unprecedented gigaflop ("one billion floating-point operations per second") performance. Indeed, two and a half years later, Apple is still singing the same tune; even now, its web site states that "the Dual 1 GHz Power Mac is a twin-engined supercomputer with a phenomenal peak performance of 15 gigaflops." And who's going to argue with that? After all, that sure is a whole lotta flops.

Well, we'll tell you who's going to argue with that: the well-respected but incredibly confusingly-named German technical journal c't, that's who. It just so happens that c't got its mitts on one of those new dual 1 GHz wonders and subjected it to a supposedly platform-independent benchmark called SPEC CPU2000, which is meant to be a test of raw processor speed. The results were not pretty. Apparently a 1 GHz G4 running Mac OS X in single user mode (i.e. no Aqua to slow things down) performed about as well as a 1 GHz Pentium III (yes, Pentium III) running Linux in the integer tests. Yecch.

But where things get really surprising is in the floating-point tests, where you might expect the G4 to excel; c't claims that the G4 only ran about half as quickly as the Pentium III, gigaflops notwithstanding. Keep in mind that the PIII is a year and a half old. The magazine acknowledges that "in theory the PowerPC FPU with its 32 registers ought to have been superior to the x86 FPU with its antiquated stack structure and eight registers only," but it wasn't-- leading c't to state that even the beefiest G4 currently available is "no 'supercomputer,' not by a long shot."

When informed of this bold statement, Apple's official response was "Is too." c't then issued a formal followup reply: "Is not." Soon thereafter, Apple stunned the industry by reiterating its position with a second "Is too." c't representatives have yet to issue a response, but industry experts strongly suspect that the magazine is currently drafting one along the general lines of "Is not, infinity." It would then fall to Apple to justify its position with an argument such as "Is too, infinity times infinity, plus your mom told us you were adopted from the monkey house at the zoo." At that point, most analysts feel that c't would then concede the debate by running home crying.

Of course, none of that changes the fact that the G4 got the pants beaten off it when it comes to the SPEC benchmark, but hey, everyone knows that benchmarks can be sketchy ways to measure true performance. If you're a graphic artist, we're going to go out on a limb and guess that you probably don't run SPEC CPU2000 all day-- you run Photoshop. And as Steve is so fond of demonstrating at every possible turn, the G4 totally kicks royal booty at running Photoshop, other benchmarks notwithstanding. We sure wouldn't mind hearing why the G4 lamed out so badly on those SPEC tests, though; after all, there's a reputation as a supercomputer at stake, here.