Just to wander slightly off-topic for a minute, is it okay to celebrate a big win for the PowerPC architecture even if it's not directly Mac-related? Because the supercomputer race just keeps getting hotter and hotter, and while the 12.25 teraflop performance of Virginia Tech's System X, the only Mac-based cluster in the top 400 (UCLA has a small Xserve cluster ranked at #444), recently slid from third to seventh place, the bestest of the best is the non-Mac-but-still-PowerPC-based BlueGene/L system developed by IBM, whose performance hit a mind-blistering 70.72 teraflops last November-- over 36 percent faster than the silver medal entrant, the Intel-based Columbia cluster at NASA. Good times.

And they're only getting better! If the very thought of anything existing in this plane of reality flexing over 70 teraflops of raw computational beefcake had you hyperventilating into a paper bag, you may want to call the ambulance before you hear this, just to save some time. Remember how, even when BlueGene/L captured the top spot by an insanely wide margin, it was actually still under construction? Well, according to BBC News, IBM's worker ants have been busy; they recently doubled the number of processors in that puppy-- and the thing now spits out 135.5 teraflops, which clearly qualifies as just plain obscene.

Like we said, it's nothing to do with the Mac-- not directly, at least. But given that BlueGene/L is so far ahead of the rest of the pack and it's got PowerPCs working its mojo (the Beeb is overstating things a bit when it says that "the chips are the same as those found in high-end computers on the High Street," since the PowerPC 440s in BlueGene/L aren't quite what you'd find in a Power Mac), some people are bound to sit up and take notice of the fact that "Intel Inside" isn't necessarily all it's cracked up to be. Will it sell more Macs? Maybe not to the average shmoe on the street, but we can definitely see at least a few high-tech speed freaks getting interested in computers that just happen to be PowerPC-based, just like that BlueGene/L thing that spanked Intel's best so hard the welt's visible on the other side.

The sickest thing of all, of course, is that BlueGene/L still isn't complete. IBM's still busy glomming more and more processors onto that sucker, and reportedly when it's done, BlueGene/L will score somewhere in the neighborhood of-- ready for this?-- 360 teraflops, which will probably cause up to a third of the human population to keel over and die of utter disbelief when it finally happens, and anyone foolish enough to gaze upon the completed BlueGene/L while it does its thing will wind up with their faces melted off like the Nazis at the end of Raiders of the Lost Ark. But hey, isn't that a small price to pay to watch a few hundred thousand PowerPCs crank out over a third of a petaflop?