Slashdot Mirror
LLNL/RPI Supercomputer Smashes Simulation Speed Record
Lank writes "A team of computer scientists from Lawrence Livermore National Laboratory and Rensselaer Polytechnic Institute have managed to coordinate nearly 2 million cores to achieve a blistering 504 billion events per second, over 40 times faster than the previous record. This result was achieved on Sequoia, a 120-rack IBM Blue Gene/Q normally used to run classified nuclear simulations. Note: I am a co-author of the coming paper to appear in PADS 2013."
Was i the only one who thought for a second that this was about a raspberry pi cluster?
I was already running Warp 3 in 1995! :-)
(OS/2 Warp 3, to be exact)
The Tao of math: The numbers you can count are not the real numbers.
I clicked hoping to read the paper, but the actual paper doesn't seem to be posted, only the abstract. The ACM copyright policy explicitly allows authors to "Post the Accepted Version of the Work on ... the Author's home page", so there is no legal barrier to the authors putting a PDF online. Doing so would of course increase readership of the paper, so ought to benefit everyone.
10 PRINT CHR$(205.5+RND(1)); : GOTO 10
No, those events are Who. Simulating is How. What is calculated.
Well.. maybe. Or Maybe not. But Definitely not sort of.
It let's us get by with fewer bombs. Fewer bombs means less chances for mistakes. And I am very glad that the superpower states cannot fight each other directly because indirect war is bad enough.
Cats.
The US military budget is about as much as the next 10 biggest national military budgets *combined.* The US isn't one player in a delicate balance of superpowers; it is a massive unilateral force, driven by greed and paranoia to utterly irrational levels of military spending. No matter how much the US has, war hawks clamor for more. "Fewer bombs" is a sick joke in the context of the ridiculous number of bombs the US has. Scrap 90% of our military, and we'd still be an untouchable superpower.
While that is true (or true-ish; there is no reason to believe the PRC's public budget numbers), the US spends its defense money so inefficiently that it doesn't have as much strength as the next ten national militarizes combined, or anywhere close to that.
This is a simulation, events. The summary doesn't say what the events are, but probably more complicated than just testing a key.
Besides, brute-forcing a key wouldn't be best done on general-purpose or even GPU. An ASIC would be the fastest, and you can be confident such chips would be easily within the capability of any major and a lot of not-to-major governments. So you're looking at a chip that can do, as a back-of-the-envelope, a key every cycle and clocked at 1.2GHz - standard for a lot of systems, as a sort of performance-per-watt peak. Times 64 cores per chip, times eight chips per PCI-e card, times eight processor cards per 2U case, times 42/3=14 systems per rack (leave space for cooling and switch), that's 1.2 * 64 * 8 * 8 * 14 = 68812 GK/s per rack.
This article is about discrete event simulation, not something as embarrassingly easy to paralleled as brute-forcing a block cipher.
The computer is performing a musical?
The Tao of math: The numbers you can count are not the real numbers.
I wonder how much money you could make mining bitcoins on that for one minute.
'You earn $400,000 by using this computer for 12 seconds.'
Headline is incorrect: Sequoia is at LLNL, not RPI.
Cats in space =)
,that's 1.2 * 64 * 8 * 8 * 14 = 68812 GK/s per rack.
That gives you about 2**40 keys/(rack*second)
If you bought 1000 racks (of this highly specialized hardware) you are still looking at a 32 year wait.
I think that the keys are probably still safe.
I'd be interested in seeing if this system could run our full Poliovirus simulations (consisting of around 3.5 million atoms). I've run our simulations on the BlueGene/Q at VLSCI using 32,768 cores (65,536 threads) and have been getting a very respectable 11.2 nanoseconds per day of simulation data using NAMD. Some data on our full virus simulations can be found here... (VIDRL supercomputer simulation page). Hey Lank, maybe you can help me figure out a way to crack the millisecond mark for our full-virus sims??? Great work and cheers from down under :-)
Maybe we could program them to scour the internet for comments that take a completely unrelated tangent to the original article, for the purposes of expounding on whatever irrelevant social agenda/issue the commentator has stuck up his behind? And then it could automatically delete them.
A liberal coming to terms with building a new nuclear power plant in the US
Good leaders run toward problems, bad leaders hide from them.
Its just that our stupidity is smarter than theirs!
Good leaders run toward problems, bad leaders hide from them.
The US military budget is about as much as the next 10 biggest national military budgets *combined.* The US isn't one player in a delicate balance of superpowers; it is a massive unilateral force, driven by greed and paranoia to utterly irrational levels of military spending. No matter how much the US has, war hawks clamor for more. "Fewer bombs" is a sick joke in the context of the ridiculous number of bombs the US has. Scrap 90% of our military, and we'd still be an untouchable superpower.
I think that the sad truth is that the primary purpose of the military budget is to serve as a welfare program whereby congresscritters can hand out jobs to their constituents and pretend that it's not "wasteful gummint spending" because it's FREEDOM, DAMMIT!
An awful lot of money gets spent on horribly expensive military toys that the Pentagon claims not to want or need just because someone in Congress could get facilities opened back home to make and/or service them. You could replace quite a few bridges - and the Interstate highways connecting them - for the price of a single Osprey, if I have my numbers right.
Not enough to pay the electric bill.
This experiment didn't perform any useful computation - they just ran PHOLD, a benchmark that sends messages between nodes in a random pattern. It's a benchmark that's specifically tailored to perform well with the Time Warp synchronization algorithm for parallel discrete event simulation. Although Time Warp performs great in theory, it relies on rolling back program state when it detects a synchronization error, and is notoriously difficult to implement in practice for large simulations.
Furthermore, these big machines are going to be used mostly for continuous (i.e. physics) simulations; this test was a discrete event simulation. Keep your eyes peeled for the winner of the Bellman-Ford prize, which is awarded to the highest sustained supercomputer throughput *when working on a real science problem*.
Miss Piggy Does Not Like. It's Pigs in Space. Hyyuhh!
Mod me up/Mod me down: I wont frown as I've no crown
and the operating system it runs is?
The title to this piece is wrong. The supercomputer in question was Sequoia, the Blue Gene/Q supercomputer located at Lawrence Livermore National Laboratory. Some preliminary work was done on a smaller RPI BG/Q machine, however. (I am a coauthor of the paper.)
So new super computer managed to "achieve a blistering 504 billion events per second". All the summery says is the computer normally does "classified nuclear simulations". So These events are what? What is is simulating?
The summary said the "coordinated 2 million cores", without saying where they were, or why they needed two labs from opposite sides of the country to do so. The summary seems long on self promotion and short on details if you ask me.
Sig Battery depleted. Reverting to safe mode.
Gustafson would be proud.
Enlightenment is the elimination of that which is unnecessary.
We came to this state by basically unilaterally assuming responsibility for the defense of Europe. It was complicated, and had to do with the question of German rearmament after WWII (and, in no small measure, Vietnam before the US was really involved). But the short answer for why the US spends so much on defense is that we have chosen to carry all our allies. Could Taiwan support ten carriers at sea? No... but will they need ten carriers to wage an effective campaign if the day ever comes? Yes. And we have chosen to commit to providing them. It's not like we don't get anything from Taiwan in return. Or ROK... or back in the day, FRG and the rest of Western Europe. Part of all that is maintaining a capable nuclear arsenal, and we use computer simulations rather than live tests to assure the efficacy of our stockpile. I think that's a perfectly good reason to buy lots of computers and employ lots of scientists. I guess you've got some specific, well-funded, noble alternative to retask all these resources on?
The simulation was a well-known parallel discrete event benchmark called PHold. It is not a model of any particular physical system, but is more of a stress and scalability test for the simulator, in this case the ROSS simulator developed at RPI. PHold has particularly fine-grained events, which stresses the synchronization mechanism known as Time Warp, implemented ROSS with support for reverse computation. It stresses the scalability of the Global Virtual Time commitment mechanism (used for I/O, error detection, storage management, and termination detection). And because PHold has no locality in its communication, it greatly stresses the underlying communication layer, MPI. The general idea is that a simulator that can achieve high performance on PHold at very large parallel scale can achieve high performance on just about any realistic, load balanced discrete event simulation at that scale.
I guess you've got some specific, well-funded, noble alternative to retask all these resources on?
Well, nuclear anti-proliferation is a pretty nice start, which just requires enough funding to yank the plug out of the wall. But, since you've already presumably got funding for operation and research personnel for the bomb-maintenance tasks, you could just re-task that along with the computer (not designing bombs is a good start in itself). I don't currently have any personal pet projects that need a supercomputer, but perhaps I could refer you to the poster "aussie.virologist" further down the thread noting this could be handy for viral simulations. Researching new antiviral drugs (and releasing the results free to the world so anyone can manufacture them) seems like a pretty "noble alternative" to assuring we can initiate global nuclear holocaust at a minute's notice.
+++ATH0 +++ATH0 +++ATH0 If you're still here, you're not on an imitation Rockwell modem.
I was promised a flying car. Where is my flying car?
Commentator is not a word. HTH. HAND.
-- Nate
From Oxford Dictionary of English:
commentator |ËkÉ'mÉ(TM)nteÉtÉ(TM)|
noun
a person who comments on events or on a text.
â a person who commentates on a sports match or other event.
Commenter may have been more appropriate in the circumstances, I'll grant you.
Laugh all you like, the guy's onto something.
To have a right to do a thing is not at all the same as to be right in doing it
You mean, like assuring our WMD supply is always in tip-top shape for deterring mass murder?
Because that's what nukes have been doing for the last 60 years.
Slashdot still doesnâ(TM)t support Unicode after it was added to the HTML standard in 1997.
Third base.
RETURN without GOSUB in line 1050