WASHINGTON, D.C. - The Department of Energy’s National Nuclear Security Administration (NNSA) announced today a contract with IBM to bring world-leading supercomputing systems to its Lawrence Livermore National Laboratory to help continue to ensure the safety and reliability of the nation’s aging nuclear deterrent.
“The longstanding partnership of NNSA, Lawrence Livermore National Laboratory, and IBM is ushering in an era of multi-petaflops computing,” said NNSA Administrator Thomas D’Agostino. “These powerful machines will provide NNSA with the capabilities needed to resolve time-urgent and complex scientific problems, ensuring the viability of the nation’s nuclear deterrent into the future. This endeavor will also help maintain U.S. leadership in high performance computing and promote scientific discovery.”
The Rest of the Story
“The longstanding partnership of NNSA, Lawrence Livermore National Laboratory, and IBM is ushering in an era of multi-petaflops computing,” said NNSA Administrator Thomas D’Agostino. “These powerful machines will provide NNSA with the capabilities needed to resolve time-urgent and complex scientific problems, ensuring the viability of the nation’s nuclear deterrent into the future. This endeavor will also help maintain U.S. leadership in high performance computing and promote scientific discovery.”
The Rest of the Story
Comments
When you look at the success that IBM brought to the Blue, White and Purple platforms while LANL going a different route for their Blue machine and then the Q failure, I think LLNL/IBM has a proven track record. Is it a smooth road we've traveled with nary a pothole, no. Perhaps SGI/Cray, HP, Sun - will develop something but until someone shows me that they have a proven product that gives you faith they can produce again in the future I think you keep dancing with the partner that got you here.
Blue had serious problems with cpu connections, white had problems with uptime, and purple multiple disk problems. Yet IBM was never penalized per contract and Comp leaders even announced they were going to lie about meeting benchmarks so IBM could get their money on time. People aren't that generous without something going on under the table.
And talk to the IG - you're kidding aren't you?
You point out the failures we had in the Blue/White/Purple platforms, and those points are valid. But we need to put them in context. Blue Pacific performed better than Blue Mountain at LANL. LANL did not have a White machine and in fact did milestone work on White. The Q machine was a disaster for LANL and Comp predicted it would be because we had tried a smaller machine of that architecture an LLNL and battled with it.
The "Too Big to fail" is not without merit. If you push a company hard enough to meet specifications and find that it has no resources/reserves to call upon, it can fold. The BBN Butterfly and the Meiko come to mind.
I understand your statement and your POV. However....
Talking to the IG would be no different than talking than an lab omsbud. Useless, unless you want to get fired or worse. Even before LLNS, employees would be hunted down for saying the wrong thing or evidence created after the fact to justify an action. LLNL was sued more than once for such actions, including Comp management. Despite losing those suits nothing ever changed in Comp, except the managers involved getting promoted.
The reason LC did better than LANL had more to do with the line workers than the machines. Before IBM's dominance, NERSC and LC staff each made machines from Cray, SGI, HP, DEC, Sun, BBN, Meiko all work for a good long time. They helped re-design Blue's file system to make it better. LC's system administators and programers turned a lot of IBM's sow ears into silk purses through huge amounts of hard work.
This all despite scum like MS. MS never took the blame for entire compute clusters that had massive problems with power supplies or contracts with inadequate time-of-response clauses. He just screamed/cursed/threatened his "minions" and after they pulled off a minor miracle he'd make a beaming announcement taking credit and earn another peer-1 ranking. And then go off to make another bone-headed deal with the Devil. Either people like him are stupid or bribed. He's certainly not stupid enough to have gotten himself fired so what does that leave?