<Well, Windows is a CPU sink. No matter how fast the processor, Microsoft will develop a version of Windows that will absorb every cycle so that it doesn't appear to run any faster than a 80386 running DOS. By then, Windows will also require 512+ MB to boot.

As long as I stay in the profession I'm in now, I'll be a major AutoCAD user. That ap' is, if anything, a bigger CPU hog than the OS it runs in. Rule of thumb: with each new version Autodesk puts out, the minimum RAM needed to run it is twice whatever Windows needs. And. of course, they hang more bells and whistles on each new version, taking advantage of increased clock speed... The end result: Acad2000 runs only slighly faster on this 867MHz/256MB PIII box parked at my left knee than AcadR14 did on the 133MHz/64MB PI at my last gig, a year ago.

Of course, with plenty of HDD I get a huge swap file, which means I can have two windows of Web browser open and a half-dozen other TSRs (ooops - dating myself!) running in background at the same time...

The question was about Joe Average, though... and I'm going to turn it around and ask: will the rest of the Web infrastructure - the throughput devices - improve at a rate to make a 10+GHz machine worth owning? ISPs are a major bottleneck - land lines are even worse. A lot of us Joe Average types live in the sticks, where there's no cable TV, the phone lines are antique and the switching centers, while converted to push-button dialing, are about as old as that tech is. Where I am, the mountains obscure the nearest cell phone tower. Anybody got a projection on the cost of a direct satellite uplink?

I trhink I'm sidling up to an answer of "No, Joe and Jane won't need a 10+GHz box". Their kids might, for stand-alone gaming, but anything involving interaction with a remote player or device will be limited by throughput speeds elsewhere in the system.

E'