Slashdot Mirror
Award of $200M Supercomputer To IBM Proving Controversial
An anonymous reader writes "According to documents accidentally placed on a federal government Web site for a short time last week the national science foundation (NSF) will award the contract to buy a $200M supercomputer in 2011 to IBM. The machine is designed to perform scientific calculations at sustained speed of 1 petaflop. The award is already proving controversial however, with questions being raised about the correctness of the bidding procedure. Similar concerns have also been raised about the award of a smaller machine to Oak Ridge national lab, which is a Department of Energy laboratory, not a site one would expect to house an NSF machine."
No-one ever got fired for buying IBM!
...to stop Zonk from posting boring, pointless articles?
I'd prefer if HAL didn't open the pod bay doors, if Zonk were out there.
Seems a good bet it's a Blue Gene/P.7 91.wss
http://www-03.ibm.com/press/us/en/pressrelease/21
--- I've completed diagnosis of your problem and can classify it as a YOYO...You're On Your Own
I don't know that having one of the machines at Oak Ridge is that big of a deal. One simple explanation is that the NSF is going to share time on the mainframe with the DoE, and in exchange, the DoE foots the energy bills and finds a place to put it. I'd rather have the agencies sharing multi-million dollar computers than buying them and not using them to capacity.
The question is why not IBM? Who else can beat it (BlueGene) at that price? Seems like a pretty good deal to me. Although, government procedures are never optimal. Free market works far better and far more efficient.
1 Petaflop?
Were any animal rights activists harmed in the design or manufacture of this computer?
"According to documents accidentally placed on a federal government Web site for a short time last week (a punctuation mark maybe?) the national science foundation (NSF) will award the contract to buy a $200M supercomputer in 2011 to IBM. The machine is designed to perform scientific calculations at sustained speed(s?) of 1 petaflop. The award is already proving controversial however, with questions being raised about the correctness of the bidding procedure. Similar concerns have also been raised about the award of a smaller machine to Oak Ridge national lab, which is a Department of energy laboratory, not a site one would expect to house an NSF machine."
Come on editors!
The government can't save you.
If the government was interested in a machine from a company who has consistently shown it knows how to build these things, then who else would they choose?
IBM has consistently dominated the fastest supercomputer list:
http://www.top500.org/
And as for it's location... why would the government want to keep putting all their eggs in the same basket? Also, it's not like you need a keyboard and mouse and operator directly attached to this machine... so housing it elsewhere in a facility that can house it makes sense.
Sounds more like a bunch of people grumbling that they arent going to have access to what they thought would be their newest toy. In addition, it indicates possible collaboration between the DOE and NSA which should only be a good thing.
StarTrekPhase2 - The Five Year Mission Continues!
A NSF machine? I wonder what a Not Safe for Work machine would have on it? PORN!
Given that IBM's strategy is to try to run linux on all its hardware, there's a good chance that the supercomputer will be running a version of linux. Expect Ballmer to be sending several chairs by express courier to the NSF in short order.
"Dude, you're gettin' a Cray!"
I just read Slashdot for the articles.
Aw c'mon, we all know it will only output 42.
The
I think you hit the nail on the head. Show me a $200M government contract award that WASN'T challenged by the folks who didn't get it...
E pluribus unum
As I recall, the number of NSF-funded supercomputer centers was drastically cut maybe 10 years ago or so, and something like only three emerged intact (NCSA, Pittsburg, and SDSC). I presume those are the "Illinois," "California" and "Pennsylvania" options mentioned in this otherwise utterly clueless piece of journalistic malpractice. Although maybe "California" refers to Livermore or LANL.
That NCSA might win the contract with a proposal that IBM build the machine is about as uncontroversial and "safe" a result as one could possible imagine, given the very long track record each institution has in this area. So I suspect the "widespread concern" is probably widespread concern for somebody's job or research grant. My impression is that the NSF has been steadily decreasing the amount of money it spends on supercomputer projects, because the community of people who think you can do something with a supercomputer that you really can't with a cluster is steadily shrinking. What that means is people in the business are being brutally squeezed as the research money dries up, and some are being squeezed right out of the field. Harsh, and it's not surprising if the harshest thing they say in response -- as life dreams go up in smoke -- is that they aren't sure the NSF was totally "fair" when they awarded the mongo grant to some other center...
The bit about an "extraordinary shift in the balance of computing power between military and scientific computing centers" -- a shift toward the scientific side and away from the military -- is so strange a thing for the New York Times to bitch about, with their knee-jerk loathing of all things military, that it's hard not to believe the reporter totally misunderstood Simon's statement (which is probably not much more than him saying Livermore has always had badass computers for designing nukes, and he thinks they still should, notwithstanding the fact that the nuke design business has been a bit slow lately).
Finally, the bit about another computer going to ORNL demonstrates more cluelessness on the part of the reporter. The fact that UT runs ORNL under contract to DOE doesn't prevent the PIs at ORNL from getting NSF grants -- and plenty folks there have them, I believe. I can't see any reason why folks at ORNL couldn't submit a successful proposal for a big computer to the NSF. They do a lot of unique materials research (since they have those great neutron and X-ray sources), and materials research is a good place to do massive simulations.
I wonder who, with what personal axe to grind, submitted this curiously vapid and pointless article to slashdot?
IBM has at least 3 different systems that this could be (x86-ish linux cluster, power* aix, bluegene), but the article doesn't say which.
Other competitors would have been Sun (linux or solaris), SGI (Altix), Cray, etc. Apparently the USGov won't consider Japanese machines so Hitachi/NEC are out.
PS Japan is building a 10x faster machine in the same timeframe.
PPS The top500 is heavily biased toward clustery machines; certain types of science codes do not run well on such systems. Not saying it's a wrong metric, just that the machines on the top500 are not good at every kind of problem.
The summary is misleading. RTFA. The real issue here as academic prestige via the selection site. This has nothing to do with the selection of IBM. Mascarading under the cloak of "California" and "Pennsylvania", I'd guess we actually have a couple of ugly stepsisters in the form of the supercomputing facilities (and the universities themselves) at UC Berkeley and Carnegie-Mellon. They seem to be shocked and somewhat put out that one of them was not selcted (instead of NCSA/UIUC). So the response now is to try and CREATE a "cloud of suspicion" by complaining that there might be a cloud of suspicion.
Harumph! What were they thinking? I mean, how suspicious that someone would put a new supercomputer at someplace called the National Center for Supercomputing Applications. Obviously there's something wrong with that selection!
Another point to consider, anytime a large contract (in this case a huge one in several respects) gets awarded, Miss Congeniality and Miss Second Runner-up are going to protest. Happens all the time. This go around however the academic community is trying to BS their way through the issue.