Slashdot Mirror
SGI & NASA Plan 10240-Processor Altix Cluster
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!"
What would you do with 10k processors hooked up to 500 terabytes? Sounds like you could replace every machine Nasa has with an account on this thing.
Sounds quite insane, I'd love to see the practical reasons for this.
This is great news for intel. They will double the number of itanics shipped in a single deal!
Stick Men
I'm guessing that NASA found out Doom 3 has a software renderer and are buying the minimum specs.
The reason? The License. While BSD License really is the most free, it would allow IBM to put a lot of effort into it, and then have MS swope in, modify it, and sell with a sorts of closed APIs, etc.
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.
I prefer the "u" in honour as it seems to be missing these days.
Good luck SGI, the Valley is rooting for its former star, and so are a lot of stock speculators.
So NASA is planning to upgrade to Longhorn then?
to fake a human settlement on Mars.
Laws are for people with no friends.
Just curious. My guess is that Intel keeps pumping money into SGI to get Altix systems out and those who have them (LLNL and ...?) got them at practically no charge to run Linpack and look good on the Top500 list.
I'm glad to see that SGI has regained its legs and is back in the high-end computing market again. The gamble they made in embracing Linux has paid off. Other folks had counted them dead because they came to the WinNT game late and were, therefore, fated to be high-priced integrators. Their days were numbered by the low-end market forces like Dell and HP.
Now we see that there is a market for high-priced integrators as long as the underlying technology fits the market segment you target.
"Rocky Rococo, at your cervix!"
I have always liked SGI hardware. And congradulations are in order for them to have a single Linux kernel running across 512 CPUs.
In SGIs press release they state that they hope to get the top spot on the 500 list. As all know IBM is expecting Blue Gene http://www.research.ibm.com/bluegene/ to take the top spot in 2005.
It looks like SGIs architecture for the Altix is better than the Blue Gene, but 10,240 intel CPUs is just going to be outpowered by the 65,536 PowerPC CPUs in Blue Gene.
Now the ultimate machine would have SGIs architecute (memory) and #CPUs per node using the PowerPC CPU. We know that IBM and SGI would never colaborate on something like this, but can't a geek dream!
More blue gene specs: http://sc-2002.org/paperpdfs/pap.pap207.pdf
thought that SGI sold a lot of their graphics IP (including many of their top graphics engineers) to NVIDIA a while back, and still have agreements with them. Their IRIX systems sell with VPRO graphics cards, which I believe are repackaged NVIDIA chips with a few extras..
Or did I miss something?
Yes, The Vpro series only resembles Nvidia chipes because after it was completed, most of the team went to work for Nvidia and created the geforce with lots of the same ideas behind it. So the original GeForce chips were more like cut down Vpro's than the VPro's were soupped up GeForces if that makes any sense.
I don't know what NASA would do with this, but I know what our group would do with it.
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.
Mmmm......sacrelicious.