To: John Walliker who wrote (39959 ) 4/15/2000 2:42:00 PM From: Daniel Schuh Respond to of 93625
I don't know about exact, but the reported cause of the last minute delay on the 820 was that the boards wouldn't work reliably with 3 RIMMs, for obscure timing reasons. And that would very definitely be a Rambus related problem. That of course doesn't say anything about the fact that the 820 was already way late at the time the launch was aborted. Which, offhand, I would guess also had something very much to do with the 820 being "designed around Rambus". You disagree? Ok, you're entitled to your opinion. I don't imagine Intel will be publishing a history of the project anytime soon, and I sure wouldn't take any statements from the local Intel employees seriously. Who knows, maybe the 820 was way late because too many Intel engineers were spending all their time on SI. Yes, the 820 and 840 were designed around Rambus, and yes, I'd say that's a problem for Intel. Because Rambus memory is way expensive and provides no apparent offsetting benefit, unless you look at the correct carefully chosen and cooked benchmarks. And I don't particularly care, as long as there are alternatives available. If it were a question of Rambus either getting production costs down or dying on the market, it'd be fine if events just took their course. The thing that really annoys me about Rambus is this crap with the no-criticism allowed contracts with memory makers and all the silly patent suits. That, and Intel's "this is the way the world is going" line, and the investors who seem to have picked up on the Rambus company "sue the critics" line. Know what would make me happy? If Intel took the same "give the customers what they want" line on desktop systems as they're currently taking on server systems. Any idea why there's a distinction being made there, John? Cheers, Dan.