New IBM Super Computer

[Bob Ayers]

Jim, check this out....if you had one of these, we wouldn't be having the "server too busy" messages!


The sound barrier. The four-minute mile. The moon. So many milestones that once seemed insurmountable are now written in the record books of human and technological achievement. Now IBM has added another to that list: the petaflop.

A milestone in perspective:
IBM’s latest Roadrunner system, designed for the U.S. Department of Energy and its Los Alamos Lab, is the first supercomputer to achieve performance at the petaflop level. The accomplishment was announced today by the U.S. Department of Energy in conjunction with the U.S. National Nuclear Security Administration, TOP500.org and IBM.

Read what IBMers are saying about Roadrunner:
This type of breakthrough was just what Thomas Watson wanted for his company when he implemented the THINK motto in IBM in 1914. "Thought has been the father of every advance since time began," said Watson. That mantra is in part what drives IBMers to achieve technical greatness and why IBM technology has helped people walk on the moon, see surface pictures of Mars, map the human genome and achieve countless other breakthroughs.

Every IBMer can be proud of this supercomputing milestone. It speaks volumes to our heritage as a company that embraces possibility, inspiration and a culture of innovation. In the world of industry firsts, everyone remembers the company that makes a technical mark of such unprecedented scale.

Roadrunner is twice as fast as IBM's Blue Gene system at Lawrence Livermore National Lab, which has, at least until now, been the world's fastest supercomputer. Roadrunner is six times faster than our competition's systems.

How fast is it? One petaflop equals one thousand trillion flops or one quadrillion calculations per second. The fastest computer in the world 10 years ago was capable of one teraflop (one trillion calculations per second). Since that time, supercomputing power has increased by 1,000 times. In fact, a complex physics calculation that will take the Roadrunner system one week to complete would take the 1998 machine 20 years to finish, which means it would only be 50 percent complete today.

 

[snake]

Cool. I took a petaflop this morning though. Who comes up with this terminolgy? Do they make it up as they go, or has it been around forever?

 

[MAKG]

It's been around forever.

A FLOP is a floating-point operation per second.

I worked on teraflop computers some 10 years ago.

 

[snake]

Dayum!

 

[MAKG]

And for the record, I treat that IBM announcement with a really big yawn.

I guarantee that the exaflop boundary will be crossed, probably in about 10 more years, and probably at one of the nuclear weapons labs. There is a rule very similar to Moore's Law at work here.

Gotta love marketing.

 

[Bob Ayers]

And for the record, I treat that IBM announcement with a really big yawn.

I guarantee that the exaflop boundary will be crossed, probably in about 10 more years, and probably at one of the nuclear weapons labs. There is a rule very similar to Moore's Law at work here.

Gotta love marketing.

The Associated Press announcment doesn't sound like a big yawn:

http://ap.google.com/article/ALeqM5haM-6ZR7I0a1ENUjDHKN2RgUACnAD916MIJ00

 

[KELLY88]

I saw this story on CNN this morning. They said they were also hoping to use it to reduce pollution and find solutions to the global warming crisis. Hope it works for what they want.

 

[MAKG]

The Associated Press announcment doesn't sound like a big yawn:

AP didn't announce it. IBM did and AP picked it up.

And if you think it's important, why is it more important than the Columbia supercomputer? Never heard of it? It's the previous record holder, and it also does climate change work, plus a few other things like galaxy formation, solar models, reentry dynamics, and other nasty fluids calculations.

But it doesn't have the magic odometer rollover number.

THAT'S why it's a big yawn. The difference between a petaflop and 950 teraflops (for instancE) is not significant, but it WILL make the difference over whether AP pays attention.

And FYI, the reason this thing is at a DOE facility has NOTHING to do with solving the world's problems. It's for modelling nuclear weapons. If it weren't, it wouldn't be funded by DOE (ALL the science is NSF's domain), and wouldn't be at a ridiculously isolated weapons lab. It probably will do nonnuclear calculations, but that's not its reason for existence.

 

[Bob Ayers]

And if you think it's important, why is it more important than the Columbia supercomputer? Never heard of it? It's the previous record holder, and it also does climate change work, plus a few other things like galaxy formation, solar models, reentry dynamics, and other nasty fluids calculations.
.

WRONG!! IBM's Blue Gene is the previous record holder!!

Roadrunner is twice as fast as IBM's Blue Gene system at Lawrence Livermore National Lab, which has, at least until now, been the world's fastest supercomputer. Roadrunner is six times faster than our competition's systems.

The Columbia is only 88.88 Teraflops YAWN:

http://www.nas.nasa.gov/Resources/Systems/columbia.html

 

[Evan]

Too bad we don't have a machine like that at work. It would come in handy for de-duping contact databases...

 

[Bob Ayers]

I saw this story on CNN this morning. They said they were also hoping to use it to reduce pollution and find solutions to the global warming crisis. Hope it works for what they want.

Yes, CNN's announcement:

http://www.cnn.com/2008/TECH/06/09/fastest.computer.ap/index.html?iref=newssearch

 

[OilPatch197]

"Made in USA by Robots"

 

[Delford]

Hey OilPatch
If the users follow the original Manufacturer's Handbook, which contains basic instructions before leaving Earth, the computer can be a valuable tool. Let's pray they do!

Del

 

[fixizin]

Amazing performance, yet no details on the IC technology involved. Liquid/gas cooled?

Ironically, it's probably crunching these huge datasets using FORTRAN compilers and IMSL libraries with roots going back over 40 years... ye olde S/W tools are just so thoroughly debugged, i.e. Verified and Validated (V&V), and so focused on quad-precision floating-point and vector processing.

Sleep well America, secure in the knowledge that NO MicroSquish products are used when it really, REALLY counts.

 

[snake]

WHAT?!! You mean its not running vista pro? How can that be?!!