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.

Politics : Formerly About Advanced Micro Devices -- Ignore unavailable to you. Want to Upgrade?

To: Jim McMannis who wrote (50820)	2/24/1999 11:18:00 AM
From: Elmer	Read Replies (1) \| Respond to of 1572507

Re: "What's the approx % of multiprocessor low cost/low end servers? What is the official definition of a "Server" and why is the Pentium II/III a better server chip than the Celeron? Or is it?" Don't know what the number is. Hope someone else can provide it. I don't think there is an official definition of server but some of the main characteristics are that it is networked and holds most of the applications software and main database to serve out to the attached network nodes. The operating system (example NT Server) has differing capabilities to handle tasks a single workstation wouldn't face. A server would have no need for graphics hence Intel's 450xx series of chipsets have no AGP port (isn't this correct Ten?). Also Intel announced this week a new server board based on the 440GX chipset. What's interesting here is that it has 2 66mhz PCI slots, even though the 440GX has no 66mhz PCI capability. It would appear that Intel has used the AGP port as a PCI 2.1 66Mhz expansion because the AGP port has no graphics use in a server. What's the better server chip? Customers are willing to pay $$BIG$$ bucks for Xeon processors with their greater caching capability. This is very significant in a multiprocessor system because it cuts way down on bus traffic, allowing much higher thruput. The Socket7 systems will still be at a major disadvantage, even with large on die L2, because of the bus protocol they use. Socket7 locks the bus for the duration of a memory cycle. Intel's bus protocol allows for deferred transactions and multiple agents can issue transactions before even the first one is complete. In addition there are major cache coherency issues Socket7 was never intended to handle. I'm sure that the K7 will not suffer from the Socket7 handicap but it will add time to the multiprocessor validation effort. EP