🚀 go-pugleaf

In article <49043@ames.arc.nasa.gov> lamaster@pioneer.arc.nasa.gov (Hugh LaMaster) writes:
>
>Does anyone know what Grumman and the Naval Oceanographic and Atmospheric
>Research Laboratory are planning?  I just saw an announcement in
>Information Week that an 11 year contract (!) which could be worth
>as much as $204M (!)  was awarded to Grumman to supply NOARL 
>"with a supercomputer that will be used to preduct the behavior
>of the oceans and the atmosphere", and $41M will be used
>"...for Grumman to assemble the supercomputer."
>
>As always, I am sure the reality is rather more complex than a summary
>of a press release, but, I am surprised that I haven't heard of this
>before.  Does anyone have the details?
>
>  Hugh LaMaster, M/S 233-9,  UUCP ames!lamaster
>  NASA Ames Research Center  ARPA lamaster@ames.arc.nasa.gov

My guess is that this is the same contract which has been bouncing
around for about 3 years.  NOARL is the new conglomerate that used to
be INO (Institute for Naval Oceanography) and NORDA (Naval Ocean
Research and Development Activity) and maybe some other groups.
The machine was originally supposed to be for INO, but INO never
really got off the ground.  This was due to a combination of a director
that no one wanted to work for, a location that no one wanted to live
in, and a completely unrealistic expectation of the number of available
Ph.D. physical oceanographers/numerical modellers in circulation.

Back to the story....

The machine that they chose is a Cray Y/MP-8/864 as I recall.  The
bidding part was mostly about who was going to install and operate
it....  Another piece of the $240 Million is that there are to be two
identical machines -- one at NOARL (Mississippi) and one at Fleet
Numerical Oceanography Center (Monterrey) (I think the latter is the
correct outfit). Most of the money was for maintenance and management.

I was in on this about 2-3 years ago when one of my colleagues at Florida
State (a CDC employee) spent the summer in Minneapolis working on optimizing
one of the benchmark codes.  This particular global ocean model had a working
set of 96 MW in 64-bit precision!  ETA never could get the wall-clock time
to pass the requirement.  They probably could now since the paging software
was greatly improved last spring/summer, but it is a bit late!

The ironic thing is that this particular benchmark was proposed by an ETA
salescritter who had apparently been reading too much of the ETA marketing
propaganda!!!  Apparently the plan was to require the code to be run without
modifications, which would knock out the Cray Y/MP.  They were counting
on the Cray-2 not being fast enough to compete...

I later met the Cray analyst who implemented the out-of-core solver on 
the Y/MP, which did meet the wall-clock timing requirements.  I can't
remember his name, but he was visiting FSU when we were deciding what
sort of Cray to order to replace the ETA-10G.
-- 
John D. McCalpin                               mccalpin@vax1.udel.edu
Assistant Professor                            mccalpin@delocn.udel.edu
College of Marine Studies, U. Del.             mccalpin@scri1.scri.fsu.edu

Hi, I am looking for the e-mail address of consulting services at
NCAR (National Center for Atmospheric Research) to get information
on their supercomputers.

Thanks in advance.

The information that I eventually want to check is this:

What :NCAR \\CRAY \\Y-MP/832 
Storage \\(Gbytes):19.2 
Clock \\Cycle (ns):6.
Performance \\(MFLOPS):2700 
-- 
Olivier Schreiber (404)894 6147, Office of Computing Services
Georgia Institute of Technology, Atlanta Georgia, 30332
uucp: ...!{allegra,amd,hplabs,seismo,ut-ngp}!gatech!prism!ccsupos
ARPA: ccsupos@prism.gatech.edu