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: Tenchusatsu who wrote (50828)	2/24/1999 1:49:00 PM
From: Elmer	Respond to of 1572571

Re: "I think Socket 7 processors can do multiprocessing, if you have chipsets and motherboards to support it. I don't think anyone really pursued it, though. I don't know anything about the Socket 7 bus protocol, except that it's transaction-based (duh)." Actually I would call it cycle based where each memory/IO access locks the bus until it's completed. A transaction based protocol has a means of identifying the transaction by ID# and the agent it belongs to. This is how transactions can be completed out of the order they were issued. (I know I'm not telling you anything you don't already know, but others might like to understand). Socket7 has no such capability. There are some Socket7 dual processor boards out there but they only work with Intel processors. Intel holds the patent on the APIC Pentium uses and AMD/Cyrix processors don't conform. AMD and Cyrix jointly defined their own APIC protocol but because of the poor Socket7 performance nobody bothered to use it. EP

To: Tenchusatsu who wrote (50828)	2/24/1999 6:59:00 PM
From: TGPTNDR	Respond to of 1572571

Tenchusatsu, Multiprocessing can be done in socket 7 motherboards, but not very well. Prior to the Pentium Pro, there were several dual socket Pentium motherboards in manufacture. "Servers", in general, do one function, and come in lots of flavors, the most interesting of which, for Expensive CPUs are application, file, and database(which is really just a cross between application and file), and MUST be exceptionally well networked. Back in my mainframe days we had a 3090 (with multiple T-1 lines as input(I know, ancient technology - the transfer rate from your hard drive to your cpu is about 8 T-1 lines wide) dedicated to serving printers,(A version of File-Application server, since Files are input, the Application formats the files for the printer, and Files are output to the printer-spoolers) It is interesting, that the lone INTC holdout in the top 10 PC manufacturers, DELL promotes(and may manufacture) the NTAP(Network Appliance) file server line which is based on the CPQ(X-DEC) line of cpus( 21164 & 21264 - ALPHA for which AMD has manufacturing rights). tgptndr