/. US DOE gets a $24.5 Million Linux Supercomputer

Greg Lindahl lindahl at keyresearch.com
Thu Apr 18 11:12:52 PDT 2002


On Thu, Apr 18, 2002 at 09:37:07AM -0500, Richard Walsh wrote:

>  True, but the dual-floating point units in the core are not
>  likely to be added to ...

I don't think that's a good guess. Not only are conventional cpus
getting wider over time, but the whole point of EPIC and VLIW in
general is that they potentially go _really_ wide. And most scientific
codes can use lots of functional units.

>  When I saw the posting, I was surprise how few IA-64 processors 
>  (even with the extras) were to be had for ~$25,000,000. The Pentium
>  4 looks a better deal at this level of analysis.

PNL's codes require 64-bit addressing. I don't think anyone was
willing to bid AMD's Hammer, although it was a possibility.

The problem with pricing this bid are that it's all forward-priced. HP
is committing to delivering Madison processors when McKinley isn't
even formally released. That means risk, and that means you need to
subtract off $$ to cover that risk. That skews all of your analysis.

BTW, the interconnect is next generation Quadrics. I've never seen any
specs for it, nor pricing.

greg




More information about the Beowulf mailing list