US reclaims top spot on Top500 supercomputing list
The U.S. Department of Energy's Sequoia is number 1, but Europe has more systems in the top 10
IDG News Service - The U.S. once again has the most powerful supercomputer in the world, thanks to the U.S. Department of Energy's Sequoia, according to the latest edition of the Top500 supercomputer list, ending Asia's hold on the top spot. Sequoia's 1.57 million processor cores can perform 16.32 petaflops (quadrillion floating-point calculations per second).
The U.S. lost the supercomputer throne to the Chinese Tianhe-1A supercomputer in November 2010. That, in turn, was surpassed by the Japanese K Computer, which held on to the number one spot for two editions of the twice yearly list. The K Computer, able to perform 10.51 petaflops on its 705,024 Sparc64 processing cores, is now in second place, while Tianhe-1A has slipped to fifth place.
Sequoia is an IBM BlueGene/Q system powered by Power BQC 16-core processors running at 1.6GHz. It runs Linux. The DOE's National Nuclear Security Administration announced Sequoia's construction in February 2009. The computer is installed at the Lawrence Livermore National Laboratory, and has been upgraded since it appeared at 17th place in last November's ranking.
Following Sequoia and the K Computer is a new U.S. system, the Argonne National Laboratory's Mira. It is another IBM BlueGene/Q system with 786,432 cores, capable of 8.16 petaflops.
The top U.S. system on the previous list, Jaguar, is now in sixth place.
The latest list is a big win for the U.S. But it also marks a return of European systems in force, according to the list's authors. Europe has more supercomputers in the top ten than any other continent, with four systems that are all making their debut on the list.
The fastest European supercomputer is Germany's SuperMUC, capable of 2.9 petaflops using 147,456 processor cores. It is installed at the Leibniz Rechenzentrum. In fourth place, it is the highest-placed system on the list to use Intel processors.
The other European supercomputers on the top ten are Italy's Fermi (seventh place), Germany's JuQUEEN (eighth) and France's Curie thin nodes (ninth).
For China, the latest Top500 list is less uplifting; the Tianhe-1A system and the Nebulae have dropped from second and fourth place to fifth and 10th.
The U.S. is still the leading continent with 252 of the 500 systems on the list, followed by Asia with 121 and Europe with 106 systems. The latter two grew their shares, while the U.S share dropped by 11 systems compared to the November list.
Overall, the Top500 list saw a big boost in performance, with a total computing capacity of 123.4 petaflops, up from 74.2 petaflops on the November list and 58.7 petaflops one year ago.
But more performance often means more power consumption too. The average power consumption of a Top500 system is 671.3 kW, up from 634 kW six months ago and 543 kW one year ago. Sequoia is not only the fastest of the supercomputers, but also one of the most energy-efficient, at around 2 Gflops/watt, a characteristic it shares with the other BlueGene/Q systems on the list.
Niether Sequoia nor the five other newcomers in the top ten use graphics processors to boost performance, but the popularity of such accelerators or co-processors is growing, with 58 systems using them, up from 39 on the previous list. Of those, 53 use Nvidia chips, two use Cell processors, two use ATI Radeon chips and one new system uses Intel's MIC architecture.
IBM and Hewlett-Packard continue to sell the bulk of the systems at all performance levels of the Top500. IBM kept its lead and has now 213 systems on the list, compared to HP with 138 systems.
Send news tips and comments to firstname.lastname@example.org
- Workload Change: The 70 Percent of Your Business DevOps Forgot Adding WLA early in the development process ensures that the benefits of DevOps accrue for all applications, including your batch services. This paper...
- Oracle EBS and RAC Performance & Elasticity on Vblock™ Systems VCE Vblock™ Systems provide simple, low-risk solutions for migrating from a physical to virtual environment.
- Backup and Recovery of Oracle EBS on VCE Vblock™ Systems This solution architecture describes the key features and benefits of backup and recovery in the virtualized Oracle EBS on Vblock Systems with EMC...
- Zero Downtime Migration for Oracle EBS R12 with RAC on Vblock™ Systems The validation testing of a typical customer configuration outlined in this white paper, utilizes a virtualized Oracle EBS environment. The results illustrate that...
- Building Tomorrow's Data Center with Converged Technologies A number of forces are converging: the cloud, converged infrastructure, big data and fabric architectures to name a few.
- Responding to New SSL Cybersecurity Threat The featured Gartner research examines current strategies to address new SSL cybersecurity threats and vulnerabilities. All Data Center White Papers | Webcasts
Our new bimonthly Internet of Things newsletter helps you keep pace with the rapidly evolving technologies, trends and developments related to the IoT. Subscribe now and stay up to date!