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 : Silicon Graphics, Inc. (SGI) -- Ignore unavailable to you. Want to Upgrade?


To: Joseph E. Caiazzo who wrote (4075)1/20/1998 4:33:00 AM
From: Edward Smyth  Read Replies (1) | Respond to of 14451
 
Message from Joseph E. Caiazzo on Jan 19 1998 11:46PM EST
>Cray was in its death throes when SGI bought it and
>they have done nothing to revive it. The Japanese now
>produce massively parallel supercomputers at a far
>cheaper price than SGI.

I disagree. 1995 was bad for Cray because they replaced their
entire product line. 1994 was their best year ever.

The big advantage of Cray over its supercomputer competitors is
the OS. My work involves using a J90, T3D, T3E and IBM SP2.
The Cray systems are far ahead of the IBM on many things including tape management and batch queues.

T3E's now have the ability to automatically checkpoint/restart
parallel jobs. They can migrate jobs from one set of processors
to another to optimize system efficiency (bit like de-fragmenting
a disk).

I don't think any other MPP system can do these things. Plus the
T3E's interprocessor comms system is one of the best in the world.

>As a general purpose machine Cray is simply too costly, thats
>why I said SGI's future is only in the limited (albeit huge)
>graphic intensive field....animation, molecular modeling,
>design engineering, etc...

Yes, Cray J90 and T90 vector machines and the T3E are too costly
for a general purpose machine, but the Origin isn't. Plus,
if you build the Onyx "workstation" for the market you describe,
the development of the Origin "server" is free! :-)

>Very large enterprise computing needs are migrating to service >providers like IBM who provide mainframe and all logistices.

Mainframes are a completely different market from the traditional
Cray market. SGI will probably want to target mainframe customers with the Origin 2000.

Ed

PS:
>Cray was in its death throes when SGI bought it

People said the same about MIPS at the time.



To: Joseph E. Caiazzo who wrote (4075)1/20/1998 10:45:00 AM
From: Justin Banks  Respond to of 14451
 
Joseph -

I think Cray was in its death throes when SGI bought it and they have done nothing to revive it.

SGI gained a substantial amount of technical expertise from the Cray aquisition, as well as access to the vast majority of a $1B yearly market. Cray branded machines are quite hot sellers, and the R&D for them is 100% done. All SGI has to do is build them and bank the checks.

The Japanese now produce massively parallel supercomputers at a far cheaper price than SGI.

Actually, the Japanese don't produce MPP or vector machines more cheaply, they sell them at a loss. That's illegal, no matter which way you slice it.

As a general purpose machine Cray is simply too costly

AFAIK, there has never been a Cray branded machine designed for general pupose use (although the Cray branded O2000s come close). It's impossible to build a machine that performs well on specific types of vector and MPP code that also performs well on general purpose apps. For a decent discussion of this, see the recent thread about VM (virtual memory) on Usenet comp.sys.super newsgroup.

SGI's future is only in the limited (albeit huge) graphic intensive field....animation, molecular modeling, design engineering, etc...

On the contrary, SGI provides solutions for most types of problems now, and the problem space that we are capable of addressing will only grow with the introduction of our NT machines.

Very large enterprise computing needs are migrating to service providers like IBM who provide mainframe and all logistices.

All the (NDA) numbers I've seen seem to indicate otherwise. Sorry I can't really refute this, but if you've got statistics to back up what you're saying, I'd be interested in seeing them.

-justinb

BTW : I didn't work for Cray before the SGI aquisition.
BTW2 : As always, speaking only for me.