
French connection
Quote: The project has been dubbed "Hyperion".
No doubt in tribute to the late 70's/early 80s "Ajile Hyperion" Home Computer. Was that French?
Anyone else remember those?
Mine's the anorak.
America's Lawrence Livermore nuclear bomb lab has teamed up with open-source computing heavyweights to build the next generation of Linux superclusters, ultimately scaling into the petaflop range. The project has been dubbed "Hyperion". "Hyperion represents a new way of doing business. Collectively we are building a system …
Mark Seagar is quite well-known at Livermore for coming up with outrageous ideas and having no clue at all how to make them work. Looking at the specs of this cluster it's pretty clear it's a hodge-podge of "cool ideas" just thrown into a blender and set to puree.
I pity the BOFH's there that will have to suffer another series of long weeks with little sleep trying to make this abomination work while Mark hovers nearby in Seagull Manager mode.
What's quickly adding up to pocket change[1] when detonating a bomb would be all-but-free? Not to mention the latter would send a message. To who, I don't know. The content, I don't know. I just like sending messages and blowing things up(and powerful computers).
[1]Remember, we're talking government here.
LLN does important research other than designing nukes - though it does do that too.
IIRC, about 8 or ten years ago, Livermore's children and I think also his widow sued the USA gov't to rename the facility to not include his first name. They said in their court brief that the man was not an advocate of nuclear weapons: if it had to have the name Livermore they wanted his first name removed.
LLN does climate modeling also. This is at least as important a topic as nukes and is at least as computationally intensive.
That said, the USA is about as adherent to nuke non-testing treaties as the Russians and both are more so than India, Pakistan and North Korea. Yet it still has several thousand nukes: Reagan and Gorbachev did not completely succeed. Whether the research is on new weapon designs or on understanding the stability of existing designs over time and degradation - would you prefer live tests & radiation or would you prefer computer simulations?
Of course no nukes across the board would be the best, but you gotta play with the hand you have.
I like Lewis' articles a lot but in this case hyperbole got the better of him.