I dislike nobody, now that Woodrow Wyatt is dead - James Branch Cabell
|
|
|
NCSA cancels its petaflop supercomputer deal with IBM
THE US NCSA - National Center for Supercomputing Applications - has pulled out of its contract with IBM for a supercomputer that can sustain a petaflop.
NCSA's Blue Waters project got under way back in 2007 after the organisation selected IBM to supply it with a supercomputer that could sustain a petaflop of computing power. However the technology IBM proposed to meet those goals was "more complex and required significantly increased financial and technical support by IBM beyond its original expectations", meaning it would run over budget.
The NCSA said it tried to keep IBM's participation in the Blue Waters project alive "but could not come to a mutually agreed-on plan concerning the path forward". The NCSA said IBM will return the money it has received for the project and the NCSA will return the equipment IBM delivered.
Although the termination of IBM's contract is a blow to the NCSA's dream of having a supercomputer that can sustain a petaflop, it still believes the goal is reachable. In its statement, the organisation said, "NCSA is confident that its goal of building a sustained petascale supercomputer remains achievable in a timely manner. NCSA is coordinating with the National Science Foundation to ensure project continuity and that the goals of the project are achieved."
Back in June, the top 10 super computers in the Top 500 list all had peak performance greater than a petaflop. That raises the question of whether building a supercomputer capable of delivering a sustained petaflop is a worthy goal, given that Japan's K-Computer aims to run at 10 petaflops.
It's unlikely the NCSA will give up on its search for a petaflop machine but the fact that IBM couldn't deliver one within budget brings into question whether Big Blue can still compete at this level. Currently the firm that in many ways symbolises US computing technology is at number 10 in the Top 500 list with the ageing Roadrunner cluster. µ
Tags: Hardware
Share this:
Share
Ironic - the article above this notes "China Builds PetaFlop COmputer with its own parts" - and here IBM can't deliver a 1PF machine while the world has moved on to 10PF. And now we turn to the President to get basic drugs for health care, can't figure out how to employ people and consume energy at twice the rate of our European and East Asian competitors. While this may be timing and sensational journalism, the reality is that the American fabric is frayed and our can do attitude got up and went. The good news is that NCSA canceled instead of continuing with a proejct that would be over priced, delivered late and under perform.
http://www.theregister.co.uk/2011/06/20/intel_mic_knights_update/
The Forschungszentrum Juelich (FZJ) and Leibniz Rechenzentrum (LRZ) labs in Germany, Centre Européenne pour la Recherche Nucléaire (CERN) in Switzerland, the National Center for Supercomputing Applications (NCSA) in the United States, and the (Korea Institute of Science and Technology Information in South Korea are all playing around with prototype Knights Ferry machines, testing their codes.
SUPERCOMPUTERS REPRESENT THE CREME DE CREME OF GIBBERISH COMPUTING, THATS BECAUSE THE UNDERLYING ASSUMPTIONS ARE ALWAYS IMPERFECTLY UNDERSTOOD, SO THEY TAKE THAT MILD ERROR AND EXTRAPOLATE IT TO A TOTALLY UNREASONABLE DEGREE TO PRODUCE THE FINEST RESULTS THAT HAVE NO POSSIBILITY WHATSOEVER OF HAVING ANYTHING TO DO WITH REALITY.
BUT HEY HO, IT IS JOLLY FUN TO GO SWINGING AROUND THE BIGGEST DICK IN THE LAND, EH?
"That raises the question" No it doesn't. The top500 is based on the linpack benchmark, which is not representative of scientific applications. Most applications expect to get 5-10% of peak performance, linpack is usually upwards of 70. For a machine with 1 petaflop sustained to come online in 2011, even early 2012, would make perfect sense.
Blue Waters should have reached sustained performance from 10 to 20 petaflops...
Mr. Mariner!
You get it in one! Mediocrity? Leave that to the Europeans!
We, those who get things done, do not care for the whiny, mud-huggers amongst us!
Inovation comes from daring! So those in the way... Get out of the way!
As to IBM, I suspect that the specification for the machine has been changed to such a degree that they were could no longer meet the software performance requirements without quite a bit of new development. I would believe, from the hardware specificaitons, that that portion was within their reach.
After the wake-up call with the last Space Shuttle flight, which marked the end of 40 years of US instensity in manned space flight, the IBM contract cancellation (which appears to be because of performance) is disturbing.
I grew up in an America that worked hard to be the best and brightest technically and we seem to have lost that edge. The operative word here is "seem" because internationally a lot of this is perception. Some could perceive that the US now believes less that exceptional performance draws us all up, and instead wants to take a break and divide up the wealth that the technical achievements have brought.
In case anybody is listening, I think that is stupid, counterproductive, and I will fight it.