For those of us not in the know, "supercomputer" conjures up few images. One might be the 30 Rock episode where Jenna mistakes a prop from the failed 1975 series Supercomputer for creepy head page Donny. Or possibly a giant Star-Trek-like room emitting boops and beeps. Those in the know, like Horst Gietl, an executive consultant for the International Supercomputing Conference, will tell you that a supercomputer is simply a computer that has the most advanced processing speed and capability right now.
The computer you're using to read this is more powerful than supercomputers of yore. But today's supercomputers are so fast, the cool kids call them "slow." Take the processor in a laptop, multiply it by thousands, stack them in racks, link them through a lightning-fast network and you have a supercomputer.
"If an application on your laptop needs years to produce valuable results, then a supercomputer may solve the same problem in minutes," Gietl says. They have the ability to handle massive amounts of data for climate analysis, molecular modeling and even nuclear explosion simulation. This week, the annual list of the world's 500 most powerful machines was announced at the ISC in Dresden, Germany. Here are the top ten:
The SGI Altix system in Pau, France, has a capacity of 106.1 teraflops (one teraflop = one trillion operations per second). This supercomputer is run by Total Exploration Production and is the largest system housed with an industrial customer, according to Top500.org. Total is a gas and oil company that uses its supercomputer to do seismic depth imaging in order to locate underground hydrocarbon reservoirs. The company reports that the massive heat from the computer is being used to warm some of the building at the center.
The BlueGene/P Solution system at the Institut du Développement et des Ressources en Informatique Scientifique in Orsay, France, is just one of many IBM systems on the list. IDRIS works in partnership with another supercomputer center in Montpellier to offer its capabilities to the national scientific community.
8. EKA
This is the second year a supercomputer in India has broken the top 10. EKA, which means number one in Sanskrit, runs on a Hewlett-Packard system at Computational Research Laboratories, a subsidiary of Tata Sons in Pune, India. Tata Group is the largest conglomerate in India, bringing in $55 billion annually. CRL is focused entirely on high-performance computing. EKA's power makes it ideal for molecular simulations, fluid dynamics computations and crash simulations.
7. Encanto
This system, at the New Mexico Computing Applications Center (NMCAC) in Rio Rancho, New Mexico, was built by SGI. According to an article in the United Kingdom's Register earlier this year, Encanto was being housed at an Intel facility (Intel made most of the processors in the Top500 list). Apparently, politicians in New Mexico are considering making the 133.2 teraflops supercomputer available for businesses and academic institutions to rent.
6. JUGENE
This computer, at the Jülich Research Centre (FZJ for short in German), in Nordrhein-Westfalen, Germany is yet another that runs on IBM's BlueGene system. Back in the day, Jülich had three nuclear reactors for research, but they have all since been closed. Now the center has shifted its focus to broader scientific subjects and is participating in several grid computing projects in the European Union.
5. Jaguar
The Jaguar supercomputer at Oak Ridge National Laboratory in Oak Ridge, Tenn., was built by Cray. If the name sounds familiar, that's because Seymour Cray dominated the computing world from the 1960s through the 1980s. His Cray-2 was the fastest computer in the world for four years in the '80s. Earlier this month, Jaguar got a makeover, upping its speed to 260 teraflops. Like many other supercomputers on the list, Jaguar is used to conduct a variety of security and scientific research.
4. Ranger
Unlike the other supercomputers at the top of this list, Ranger is a system intended to be open. Sun Microsystems worked with the Texas Advanced Computing Center at the University of Texas and a team of academic institutions tobuild Ranger, which has one-half a petaflop (the next step up from teraflop) capacity. Researchers at academic institutions in the United States whose hearts beat faster at the possibility of crunching "parallel algorithms" and doing "scalable visualization," can submit proposals to have the system run their numbers.
Another Big Blue system is the BlueGene/P at Argonne National Laboratory in Argonne, Ill. Argonne, along with Los Alamos, Oak Ridge, and Lawrence Berkeley, was part of the Manhattan Project, which developed the nuclear bomb in the 1940s. It's the first national laboratory in the country and one of the largest. Argonne still conducts nuclear research, but is also conducting research in environmental management, energy resources, and a variety of scientific fields.
2. BlueGene/L
IBM's BlueGene/L system at Lawrence Livermore National Laboratory in Livermore, Calif., is nothing to sneeze at. It can do 478.2 trillion operations in a second, which it does in collaboration with Los Alamos and Sandia. BlueGene/L can run nuclear computer simulations, replacing underground testing. Thankfully.
1. Roadrunner
Earlier this month, Roadrunner became the first computer ever to reach the one petaflop per second level. Translation: it can do one thousand trillion calculations in the blink of an eye. That's 15 zeroes. That's so fast, even analogies can't touch it. Roadrunner, named after New Mexico's state bird, is operated by Los Alamos National Laboratory. IBM designed and built the record-breaking system, which will be used to do energy, astronomy, climate, human genome research, and to keep the nation's nuclear stockpile safe. Beep-beep.
0 comments:
Post a Comment