Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Technology Stocks : Intel Corporation (INTC) -- Ignore unavailable to you. Want to Upgrade?

To: Scumbria who wrote (67394)	10/26/1998 1:38:00 PM
From: Tony Viola	Read Replies (1) \| Respond to of 186894

Scumbria, >>>The most powerful supercomputer in the world is a massively parallel Pentium Pro machine at Sandia National Labs. <snip> So there really is no theoretical limit to the amount of K7's you can put in a system.<<< Yeah, but only one of those, maybe two, were ever built. Old expression in the business: "anyone can build just one." Do you really think that something besides SMP, which has become the multiprocessor standard for mainframes, Suns, and now NT servers, is going to get serious competition from something else in the foreseeable future? Tony

To: Scumbria who wrote (67394)	10/26/1998 1:45:00 PM
From: Tenchusatsu	Read Replies (1) \| Respond to of 186894

<So there really is no theoretical limit to the amount of K7's you can put in a system.> There's also no theoretical limit to the amount of K6 CPU's you can put into one system, as long as you build the crossbars and the network topology to support all of them. (Replace the name "K6" with "Pentium" or "486" in the last sentence and it would still be true.) I was only responding to Kash's weird claim that the K7 will support up to 8-way servers, when in reality, only one K7 can fit per P2P bus. The only difference is that the K7 will scale much better than the K6 when used in multiprocessor configurations. But if you want to build a 4-way K7 system in order to compete against a 4-way SMP Xeon system, you'll either have to: 1) Design the north bridge to have enough pins for four P2P connections. This leads to a very high pin count, which isn't trivial. 2) Connect the four K7's to a crossbar, which effectively acts as a multiprocessor bus medium between the K7's and the north bridge. This adds an extra component to the system and an extra layer of latency, which hurts performance. An 8-way K7 system would be even worse, while an 8-way Xeon system involves connecting two multiprocessor buses onto one north bridge. Take a wild guess as to which way is easier. Tenchusatsu