SI
SI
discoversearch

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: SilentZ who wrote (152255)12/12/2001 2:08:47 PM
From: Tenchusatsu  Read Replies (2) | Respond to of 186894
 
Z, let me bring up Combjelly's original statement:

But much of the actual cost is tied up in things like the memory subsystems, switching fabrics for NUMA systems and the low volume engineering to create those things. Look at Sun's Starfire, for example. The high speed crossbar switches used to create the NUMA topology are very expensive. SledgeHammer with a HT switch can do the same thing for a fraction of the cost, and have lower latencies to boot.

This whole statement is correct, except for the last sentence. HyperTransport does not define a comprehensive standard for crossbar switches, NUMA topology, switching fabric, etc. All it defines is a protocol for an interface that could potentially be used at part of such a fabric. The implementation of (the NUMA version of) HyperTransport will still cost big bucks because of the low volumes and the need to custom-tailor each implementation to different needs. And the latency argument is bunk because it is highly dependent on the implementation. HyperTransport has nothing in it that will magically reduce the latency of high-end switching fabrics.

And heck, even all of this is getting way ahead of where HyperTransport is at right now. In its current form, HyperTransport is only a replacement for Intel's HubLink or VIA's V-Link interfaces. (It is also finding uses in communications, but that's another subject.) It does not yet have any of the features required for the high-end system stuff that Sun, IBM, and others are doing.

There you have it, as briefly as I can make it. I feel like I'm doing you a disservice by not going into more detail, but I hope you forgive me for that.

Tenchusatsu