% fortune -ae paul murphy

PPC: IBM's ace is Games King

The world's top three dedicated gaming platforms, by sales volume, are Sony's playstations, Nintendo's gamecubes, and Microsoft's Xboxes. All are now powered by next generation PPC products made by IBM: Microsoft's X360 uses a three core, G5 derived, 3.2Ghz Xenon; Sony uses a G5 based, 3.2Ghz cell processor; and Nintendo's next generation is powered by a G3 derived "broadway" CPU running at 729Mhz.

Intel isn't in the game - the Pentium IV "extreme edition" isn't remotely in the PPC's performance league and costs gamers more then ten times what Microsoft pays for the Xenon.

More interestingly, it's the PC programming community that really isn't in the game.

Sony's people are succeeding with Cell -they're a long way from success, but they're getting there and even the first products now in beta already have uniquely cell enabled features including better identity management for multi-player games and on-the-fly avatar image generation and manipulation. Similarly, Nintendo's developers are bringing a whole new interface into a seriously resource constrained environment while maintaining their competitive position on graphics and sound. Both feats are technically very difficult, very bleeding edge; but both companies are succeeding because they have bench depth on the programming side.

But what does Microsoft have, besides deep pockets and a problem? The X360 hardware is the simplest of the three: a straight forward brute force approach to delivering somewhere around ten times the throughput potential of a Xeon but without significantly changing (except with respect to the use of the Altivec short array processor) the programming model. And yet, the few new games we've seen for the thing are repetitive at best and boring at worst, while the improvements you do see - in graphics and networking - are driven from the hardware, not the software.

So why? I think it's obvious that everybody in the volume games business has gone PPC for the same reason: much better cost/performance. Compare Xenon to Xeon and you get ten times the potential performance at one tenth the cost: a no brainer in anybody's book - and Cell is significantly better than that.

But it's also obvious that both Nintendo and Sony have well established software supply channels: not so much deep pockets as deep expertise -and that's what Microsoft doesn't have and that's why Xbox is in trouble.

In effect what's happening to Microsoft is exactly what used to happen to Apple with respect to games: people thinking in terms of the Intel programming model designing hardware and software for PPC - to produce boring, inefficient, brute force, solutions that heat rooms more than imaginations.

And yet Xenon is Microsoft's future: that's the plan: move Office to the X360, add a (BSD based!) network OS, and then migrate the combination into the office -ultimately freeing Microsoft from the costs and other consequences of the design and marketing decisions that went into DOS and Windows on x86.

It's a grand strategic vision, but as long PC programming expertise continues to direct X360 development the winner will continue to be IBM - meaning that the mob rule strategy used so effectively by Microsoft and Intel to get them where they are, may well have produced market forces that now lock them out of smarter, more technically savvy, markets.


Paul Murphy wrote and published The Unix Guide to Defenestration. Murphy is a 25-year veteran of the I.T. consulting industry, specializing in Unix and Unix-related management issues.