Unholy Matrimony? Microsoft and Cray 358
fetusbear writes with a ZDNet story that says "'Microsoft and Cray are set to unveil on September 16 the Cray CX1, a compact supercomputer running Windows HPC Server 2008. The pair is expected to tout the new offering as "the most affordable supercomputer Cray has ever offered," with pricing starting at $25,000.' Although this would be the lowest cost hardware ever offered by Cray, it would also be the most expensive desktop ever offered by Microsoft."
Desktop? (Score:3, Informative)
From the article:
If space is a problem, not to worry, itâ(TM)s compact enough to fit in a broom closet.
From the summary:
... it would also be the most expensive desktop ever offered by Microsoft.
I know, the summary was an attempt to bash MS.
Non-useless link (Score:5, Informative)
instead of bloggy blather, you can go to the source [cray.com].
Re:hardly news (Score:5, Informative)
Re:What's the frame rate and resolution? (Score:4, Informative)
you tell me [cray.com]
Re:The Microsoft Tax (Score:5, Informative)
Actually, if you go to the Cray site and configure a system, it is available with Red Hat Linux for no cost (getting HPC adds $469)
Re:Alternatively... (Score:3, Informative)
Cray have and have had their own custom UNIX distribution since before slashdot existed.
You can already get Linux on CRAY hardware -- the SGI Altix series. I haven't kept up on the offerings, but I beleive there are other *nix based offerings as well.
The value proposition of something like this is that people who are better at science than programming (you know, most super computer users) get something that makes them more productive than they'd otherwise be. The operating system on a super computer is almost irrelevant as it is customized so completely for the needs of the client. The value add in something like this is the developer productivity and toolchain on offer.
There are some seriously brainy people at MS working on the software side of HPC/scientific computing. Some of them are ex Cray employees.
Re:Alternatively... (Score:2, Informative)
Re:Poor Seymour (Score:3, Informative)
The current "Cray" is actually a new company that used to be called Tera Computer. Their connection with the original Cray is that in 2000 they bought some SGI assets that originated with Cray Research. One suspects that the only asset they really wanted was the Cray name. Ironically, when SGI owned Cray, they tried to phase out the Cray brand — with disastrous results.
Unlike the original Cray Research, Tera/Cray has always been moderately profitable. So this is not a dying gasp by any means.
Re:Finally (Score:2, Informative)
Re:Too dinosaurs working together. (Score:2, Informative)
Re:What's the frame rate and resolution? (Score:3, Informative)
Re:This thing... (Score:3, Informative)
Are you trying to imply that Cray the company is "in name only?" Because that's not at all the case.
It's true that Cray was a shadow of its former self after Tera bought it, but many of the Tera executives have left, and some of what Cray Research used to be has re-emerged.
Now, the CX1 really is Cray in name only. Don't make the mistake of thinking of Cray as a maker of itty bitty clusters. Oak Ridge has a >30,000 core Cray XT4 [nccs.gov], NERSC has an almost 20,000-core XT4 [nersc.gov], and of course Red Storm [sandia.gov] has over 26,000 cores.
How far can a company sink? (Score:3, Informative)
Serving up one of the most crappy and broken corporate websites I've seen lately, Cray bedazzles me. They can't be serious, can they? Running a high throughput, custom piece of hardware on Windows as the prime OS? ... Unbelievable.
What Oomph does this thing have anyway? 16 Quad-Core Xeons. 64GB per node. Doesn't sound like that much of a a big deal to me. What corners could Cray have cut with the system archiecture itself to justify the hype? Won't a smalish blade-box or something simular from Sun or IBM wipe the floor with this thing? ... Just wondering.
Anybody with deeper insights on this?
Re:What's the frame rate and resolution? (Score:3, Informative)
Re:What's the frame rate and resolution? (Score:3, Informative)
There's a reason for that. It doesn't deserve karma.
The quality of discussion is no longer assisted by the moderation system because it has been subject to gaming. Humorous comments should be highlighted for enjoyment, but that's what the "funny" mod is for--it doesn't mean the poster has anything valuable to contribute, which is what the mod points are for. It doesn't mean the poster is considered to provide constructive, valuable comments as a general rule, which is what karma is for.
Likewise, with the -1 mods all collapsing into "I disagree" or "this opinion has not been sanctioned by the hive", Slashdot has become an echo chamber and encouraged rabid zealotry. When anger boils over at opposition comments being smashed down, it actually creates flamebait and troll posts--which are then modded UP when they come into vogue.
Case in point: most comments about "Apple fanboys" or anything anti-IP, no matter how uninformed or ill-conceived. These are ignorant, incoherent posts but up they go because it's an acceptable viewpoint.
Re:What's the frame rate and resolution? (Score:3, Informative)
Funny enough, GCC has become orders of magnitude slower at compiling because it now supports much more sophisticated optimisations. Hardware has moved forward faster than GCC though, so they're well in the green.
Re:What's the frame rate and resolution? (Score:5, Informative)
More importantly, newer kernels *feel* faster. In particular the kernel preemption makes an enormous difference as far as perceived speed goes (for a desktop user).
When I upgraded from 2.4.24 to one of the early 2.6 releases I was astounded at how much faster things felt. On a very modest laptop (1.3 GHz Pentium-M, 512M RAM, 30G 5400 RPM hard drive) from a fresh boot I fired up OpenOffice, Konqueror, Eclipse, Firefox (might have still been Mozilla then, I forget) all at the same time, and the desktop was still liquid smooth and completely responsive. Needless to say, a similar task on 2.4 felt much slower, as actually getting the K menu to open again so I could select another program to start out of it took longer.
Newer kernels are actually faster in a lot of cases too, particularly with scalability, but lots of other optimizations have been done as well, as many kernel developers keep a very close eye on performance. Also, GCC has gotten better over time, and likely optimizes the kernel quite a bit better now than it could several years ago.