SGI & NASA Plan 10240-Processor Altix Cluster 202
green pizza writes "NASA has announced plans to cluster twenty 512-processor Silicon Graphics Inc Altix supercomputers connected to a 500-terabyte SGI InfiniteStorage SAN. The Altix uses Itanium2 CPUs running Linux atop an Origin 3000-derrived architecture. NASA and SGI scaled Linux to 512 CPUs late last year. There are also strong hints that SGI plans to bring its clustered ATI graphics to Altix in the near future. Lots of neat big iron project on the horizon!"
Surviving, but stock in a free-fall (Score:4, Interesting)
Good luck SGI, the Valley is rooting for its former star, and so are a lot of stock speculators.
Re:Good News for intel (Score:2, Interesting)
yes, for sure. they bought a congressman to make this happen. (no joke, trust me.)
and as usual, real science at nasa is going to suffer for a waste money on unneeded computing capacity just so the US can prove we have a bigger dick than the japanese.
-pissed off nasa worker
Homo Zapiens (Score:2, Interesting)
Re:Honest question: Why Linux? (Score:3, Interesting)
In essence, the BSD license would allow the creation of another Unix model where the core is identical or just similar, but the APIs would be used to lock users in. How would that solve IBM's problem? Or for that matter any Hardware vendors problem? It would not.
Finally an answer that doesn't involve ranting and raving about GPL/freedom/blah blah blah. Thanks for the simple common-sense answer to this question I wondered myself.
How much did they pay for this thing? (Score:4, Interesting)
SGI was supposed to be dead (Score:4, Interesting)
Now we see that there is a market for high-priced integrators as long as the underlying technology fits the market segment you target.
Re:Good News for intel (Score:4, Interesting)
Cheaper? Not likely, you'd have to buy the high-speed interconnect to make it worthwhile. And the Opterons perform fairly poorly in larger clusters, since they have the NUMA latency penalties locally on each node. Checking the Top500 list, a cluster of 256 Opteron 246 using Infiniband will perform worse than a cluster of 256 Xeon 2.8GHz using Infiniband. The scariest example is that a cluster of 256 P4's@3GHz using Gigabit Ethernet outperforms the Opteron cluster.....
Important to note is that the Linpack test doesn't stress the interconnect that much. The more a task stresses the interconnect, the more the Opteron cluster will be penalized. There's one exception though, and that's the Cray Octiga Bay systems.... And if you go that route, it costs _at_ _least_ as much as an Altix system
...well.... (Score:5, Interesting)
We always need machines. You could give me 1024 machines and I'd still need more.
For example, I study fluids currently. I may simulate 4,000,000 particles and it may take 3 weeks for my simulation to finish. If I had 10240 nodes, it may only take a day. Or perhaps I could simulate MORE particles for longer. There are all sorts of advantages to having this many machines hooked up.
One thing I can tell you for sure is that there most likely will not be *1* job that uses all of these at once. There are probably several researchers that are using it simultaneously and have a slice of the machines. Press releases like this are often time misleading because usually the CPUs are split between several jobs and researchers and research groups and what not.
Not to steal NASA's thunder -- a cluster this big is impressive.
Mike.
Re:How much did they pay for this thing? (Score:3, Interesting)
To counter all of their detractors, Itanium2's are pretty hot processors, and SGI has done an amazing job getting linux to run well on 500 processors. This will no doubt be one hell of a fast machine. I'm just amazed that sgi can stay in business selling at this price.
If anyone has been watching supercomputers lately, you might have noticed that cray is downsizing by 20% after selling 0 computers in their most recent quarter. It's a tough market for supercomputer makers. The clusters are good-enough-and-cheap. This box is being sold at very little more than cluster cost. Ick.