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: Paul Engel who wrote (115535)11/1/2000 2:47:57 PM
From: Tony Viola  Read Replies (3) | Respond to of 186894
 
Paul, Tony, not Ten here. Anyway, the Cornell supercomputer article is interesting. The techniques used to build up supercomputer power out of the building blocks they mention, clustering, has been around for a while. The term scaling out, to describe how the processing power is built up is mentioned below. This is vs. scale up, where you add processors to a single footprint SMP machine, like a Sun UE1000. Scale out-add servers via a Giganet type switch technology, horizontally, onto the cluster.

This method for building super computers, known as clustering, uses industry-standard computer equipment like Dell servers, Intel Pentium® Processors, Windows® 2000 from Microsoft, and networking technology from Giganet® as building blocks to group computers into clusters for increased power, performance, and reliability of the individual systems. The advantage of this industry standard approach is that it ends up significantly less expensive than building one large, custom-engineered system. Another benefit is that owners of these newer clustered systems can increase their power simply by adding another computer to the cluster.

``This approach to building high-end computing systems is known as scaling out and will make it possible for more organizations to develop these systems and possibly accelerate new breakthroughs,'' added Coleman.


In this way, of course, our industry standard servers, cheaper storage, etc., can be used to get processing power that can exceed the highest "standard" Sun or even IBM servers, and lots cheaper. Those "tpm/C" systems we see every time someone breaks the record are these "scaled out", clustered systems.

Doesn't say, but sounds like Cornell is probably stretching the scaleout part and building one or two that exceed the biggest tpm/C's we've seen.

All good stuff, helps sell servers.

Ten probably laid the groundwork for these systems when he was at Cornell!

Tony



To: Paul Engel who wrote (115535)11/1/2000 3:34:05 PM
From: ratan lal  Read Replies (2) | Respond to of 186894
 
re: Supercomputers.

So CRAY will end up in the dustbin?

And what about US sanctions against exporting supercomputers?? Will they stop the export of PC's??