Blue Gene/L Tops Its Own Supercomputer Record

DIY News writes "Lawrence Livermore National Laboratory and IBM unveiled the Blue Gene/L supercomputer Thursday and announced it's broken its own record again for the world's fastest supercomputer. The 65,536-processor machine can sustain 280.6 teraflops. That's the top end of the range IBM forecast and more than twice the previous Blue Gene/L record of 136.8 teraflops, set when only half the machine was installed."

Comments

hmmm

[jigjigga]

lets put folding@home (http://folding.stanford.edu/) on that mother!

Re:hmmm

[DrEldarion]

Seriously, it's going to be used for nuclear weapons simulations and whatnot. I'd much rather see it going towards protein folding than that.

Re:hmmm

[VENONA]

Yeah, right. That's very sensitive of you, I'm sure.

We don't do chemical or biological warfare. All we have is nuclear weapons. So how do we respond, if and when some nutcase state hits us with a nuclear, biological, or chemical attack? Harsh language? Throw some nicely-folded proteins?

Personally, I prefer that we have some assurance that our nukes will work if we ever need them.

You can make any sort of argument you'd care to about our messed up foreign policy. I'd probably agree with quite a bit of it. But I still want our stockpiles to work if we ever need them.

It's a harsh world. Sensitivity and political correctness will only take you so far in dealing with it.

Re:hmmm

[Silverstrike]

Because, when you're in agony dying of radiation poisoning, its really going to make your day a lot better to be assured that somewhere else in the world another person, equally as removed from the political context of the nuclear conflict as you are, will be in just as much agony as you are.

Now, just maybe, the presence of these weapons can be called a deterant, so its possible that possessing them is a necessary evil. However, to be quite honest, if we ever "need" them -- I really do hope they fail to work.

Re:hmmm

[Anonymous Coward]

Yes, it is also sad that while other countries use thier supercomputing power mostly to investigate protien folding and earthquake propagation and other purposes generally recognized as peaceful we mainly use ours for simulation nuclear weapons designs; but it is not all bad.

One might consider that other countries can use their supercomputing power for such purposes precisely because the US uses devotes a relatively larger percentage of its computing power to weapons development. My god is it any secret at all that the rest of the free world has been "free-riding" on US military power since the end of WWII?

Now I'm not complaining about this, or suggesting that we should have thrown them to the wolves, or even suggesting that they're anything less than grateful. But I do get tired of hearing about how militaristic the US is "compared to other [more enlightened] nations" when the US largely provided the very freedom they have to devote significant budgetary resources to matters other than self-defense.

Re:Results?

[catch23]

What useful utility has top sports athletes contributed to society? We probably pay the top athletes the cost of several super computers and all they do with all this money is throw orange plastic balls pumped with air into little string baskets hoisted on a pole. Sometimes I wonder what is the purpose of paying some man to hit a tiny ball with dimples into a hole far far away on a grassy playing field. And every time someone breaks someone else's record of hitting the tiny white ball, they get large sums of cash. But when these super computer people break records, all they get is a little pat on the back. BlueGene/L should be signing contracts with Nike and Pepsi.

Re:compiler?

[Aardpig]

On something like this, they would probably be programming in High Performance Fortran or Fortran w/ OpenMP -- or some similar dialect that supports massively parallel execution. I'm sure IBM develop an in-house compiler for the language.

Re:Notice that performance had increased per cpu..

[mj_1903]

That probably only means that they have optimised the architecture over time as would be expected. Things like improved resource management, a slimmer kernel for each CPU, a better compiler, etc. can easily make up for that small performance gain.

Re:Notice that performance had increased per cpu..

[Anonymous Coward]

Just because the summary says when the machine was halfway done at 138 GFlops, that definitely does not mean the other half was just putting in the other half of the processors. They probably also meant that the optimizations were not finished as well. Pure linear growth is a dream enough already. If I saw exponential growth I'd crap my pants.

Re:Notice that performance had increased per cpu..

[Bloater]

Linux is substantially more scaleable now than it was even just 6 months ago (not the vanilla, but quite well tested scaleability patches). This could account for the improvement. I suspect if they ran just half of it now, they'd get a little bit over half the performance (but not much over half - that is how good Linux is these days).

Re:Notice that performance had increased per cpu..

[clacour]

Sorry, but I have to disagree with your conclusion that this represents exponential growth.

The effect you speak of (doubling the number of processors giving less than double the final "power") is due to additional overhead - various processors coordinating their work with each other, deciding things like "Should I split this 2 ways or 4?" and so on - and that sort of stuff inevitably increases with the number of processors.

You can use improved algorithms, special-purpose hardware, etc, etc, to minimize this "friction", but it will always exist, and the percentage of processing that is "overhead" will inevitably climb as you increase the number of processors.

It's far more likely that either the earlier number resulted from some inefficiencies that existed then (due to it not being built as designed yet, perhaps), or there have been improvements in the algorithms or infrastructure which give greater efficiencies.

If it's the latter case, if you unplugged the 2nd half of the CPUs and made the measurement again, you'd probably get 150 GFlops or so.

Basically, you could write the equation for total power something like:

X - O - i**x, where X is the number of processors, O is the basic overhead (for doing things like I/O, for example), and c is the incremental cost of adding each processor.

To have what you describe would require that i**x be a negative number, which is like saying that you can have 10 individual conversations in less time than you can have five. Ain't gonna happen.