| Submit your comments on this article |
| Science |
| New supercomputer sets record |
| 2008-06-09 |
| An American military supercomputer, assembled from components originally designed for video game machines, has reached a long-sought-after computing milestone by processing more than 1.026 quadrillion calculations per second. The new machine is more than twice as fast as the previous fastest supercomputer, the IBM BlueGene/L, which is based at Lawrence Livermore National Laboratory in California. The new $133 million supercomputer, called Roadrunner in a reference to the state bird of New Mexico, was devised and built by engineers and scientists at IBM and Los Alamos National Laboratory, based in Los Alamos, New Mexico. It will be used principally to solve classified military problems to ensure that the nation's stockpile of nuclear weapons will continue to work correctly as they age. The Roadrunner will simulate the behavior of the weapons in the first fraction of a second during an explosion. |
| Posted by:gorb |
| #12 Can it count to infinity faster than Chuck Norris (PBUH)? |
| Posted by: Bright Pebbles 2008-06-09 22:10 |
| #11 Cue the "it'll do an infinite loop in fourteen seconds!" joke. |
| Posted by: Abdominal Snowman 2008-06-09 17:22 |
| #10 More "idle-loop" repetitions per second than EVER BEFORE! |
| Posted by: mojo 2008-06-09 16:01 |
| #9 Nope GIGO doesn't matter. It can be 99% accurate in Garbage out. The recursion inherent in climate models accentuates error. |
| Posted by: Bright Pebbles 2008-06-09 13:46 |
| #8 "They can run their models more quickly, and/or they can run their models at smaller scales; but it wont make them more accurate over a climate timescale." Probably more accruate from the point of view that they can consider more parameters in the models. However, there is always the GIGO (garbage in garbage out) factor to keep in mind. |
| Posted by: Kelly 2008-06-09 13:32 |
| #7 Already downloaded it. |
| Posted by: ed 2008-06-09 10:51 |
| #6 ed, take a look at CUDA, it is very helpful. I've been messing with it on my 8800gt, nothing serious, but it is very cool and so much easier than before as you don't have to map data structures to the texture units, etc (CUDA is high level C like language for the GPUs, so you can avoid mapping everything to DirectX or OpenGL). |
| Posted by: bombay 2008-06-09 10:27 |
| #5 and 10 pedaflops = 1 "Divac" |
| Posted by: Frank G 2008-06-09 09:00 |
| #4 Even more powerful (per chip) and interesting are the new programmable video cards from Nvidia and AMD. I recently bought an Nvidia 8000 series card just to learn how to program it for scientific tasks. |
| Posted by: ed 2008-06-09 08:59 |
| #3 Oh and 1527 Teraflops, as shown on the Stanford site, is a 1.527 pedaflop. |
| Posted by: Procopius2k 2008-06-09 08:49 |
| #2 The PS3 gaming community has been delivering 'spare' computing power to Stanford in analyzing human proteins mutations "folding" in pursuit of medical/biological studies. By downloading a program for their PS3s through the internet and then allowing it to run on the platform overnight or when not in use, each member of the community takes a portion of the process and crunches the numbers whose results are then uploaded for analysis. This PS3 network right now is producing a Teraflop or more of processing, and has been for most of this year. If this intrigues you, go here for more information. |
| Posted by: Procopius2k 2008-06-09 08:45 |
| #1 > Before it is placed in a classified environment, it will also be used to explore scientific problems like regime climate change. The greater speed of the Roadrunner will make it possible for scientists to test global climate models with higher accuracy. This is what we technically call a lie. They can run their models more quickly, and/or they can run their models at smaller scales; but it wont make them more accurate over a climate timescale. |
| Posted by: Bright Pebbles 2008-06-09 04:50 |