Vintage Computer Festival wrote: agg. someone hit me with a ball peen hammer for adding to this thread. (heh - I used 026 card punches dammit. to feed a cdc no less. used to be hot at making drum cards. who knew? can you run dos/vs in simulation yet? i miss the 370.) I think the "purist" argument is that the x86 architecture has too much baggage. Certainly most would agree that working in 16 bit mode is often painful. 32 bit mode is better, but the instruction orthogonality is a little off and the register motion is a little off and well, it could use more registers. Most of this comes from it's heritage. They've done a fantastic job keeping it alive but the legacy baggage gets in the way. The PPC architecture is a bit fresher, and more "risc-y". More registers, smoother transition into the 64 bit world, etc.. And, most would say, much easier to program. Plus, there's the whole endian religion. I "go both ways" and like them both but I will say big endian is often easier to deal with and less confusing. But I've written a large number of production software packet routers and so I'm a bit biased in that space. In then end, the comment about cache sizes is probably the most interesting/relivant. IMHO i-cache size is the most important thing. Bursting from memory into L2 is always the bottleneck. I personally think it doesn't matter what cpu apple uses. I would like to see them make pc's - they would probably be more fun to use than most generic pc's. I mean, look at the mini. Is that not one of the coolest machines you've seen in the last 10 years? (but don't get me wrong Mr. Jobs - I'm *still* pissed at you for yelling at me about the HyperDrive! :-) -brad