To: Charles R who wrote (80396 ) 11/18/1999 4:06:00 PM From: Burt Masnick Read Replies (1) | Respond to
Athlon and Coppermine and Willamette are very complex devices trying to essentially commoditize remarkable technological achievements. That's tough to do and very, very tough to do when an agressive schedule must be met. Thus Athlon and Coppermine had bumpy intros. For whatever reason, each intro was less than stellar. Putting aside which intro was worse, neither was textbook or smooth or desirable. In addition to just working at all, they have to work within manufacturable parameters so that they have reasonable yields and bin splits. Contrary to popular opinion, that's not a piece of cake, even for the very best designers and manufacturing enterprises. that old and I started grad school programming with punched cards in assembly code on an IBM 1620 with 4000 12 bit words in its' magnetic core memory. I think it did an instruction every 4 milliseconds. And it was in a glass enclosed air conditioned room on a raised platform. With a trained "operator". And it cost a million bucks. The big thrill as I was leaving grad school was being able to use a PDP5 with a paper tape entry that only cost $20,000. In Fortran. No more air conditioned room or full time "operator". The technology in the Athlon or the Coppermine just smashes what was available back then - pipelines, renaming registers, speculative decodes, multi-level caching. I designed a couple of primitive (by today's standards) embedded computers with multiple 74181 chips from Texas Instruments and thought they were hot stuff back then. AMD had the 29000 series that was pretty spiffy. And today's microprocessors don't just scream by comparison - they are in another universe.
