Since 1993, the TOP500 project has been publishing a semi-annual list of the 500 most powerful computer systems in the world, as a barometer of trends and accomplishments in high-performance computing. The latest list has just been released this month, and there is a new speed champion, the Jaguar Cray XT5 system at Oak Ridge National Laboratory in Oak Ridge, Tennessee. Unlike many previous Oak Ridge computers, which have been used to model nuclear explosions for military purposes, the Jaguar system is being used for civilian purposes, principally climate modeling.
The speed ratings for the systems on the list are based on 64-bit floating-point performance on the LINPACK benchmark, which solves a dense system of linear equations. Of course, a single rating cannot possibly capture all aspects of a system’s performance; but since systems in this class are usually employed to solve extremely computation-intensive problems, the LINPACK measure is a reasonable first approximation. The following table shows the top five systems, with their locations, speeds (in teraflops, 1 × 1012 floating-point operations per second), and basic processor technology:
Rank | System Name | Country | Speed | Technology |
---|---|---|---|---|
1 | Jaguar | USA | 1800 | Cray KT5 |
2 | Roadrunner | USA | 1042 | AMD / Cell |
3 | Kraken | USA | 831 | AMD Opteron |
4 | Jugene | Germany | 825 | IBM Blue Gene |
5 | Tianhe-1 | China | 563 | Intel / AMD GPU |
The TOP500 site allows you to generate charts and graphs of the systems categorized in various ways: by vendor, by country, and by processor architecture, for example. One of the more interesting categorizations, perhaps, is by operating system:
OS Family | No. of Systems |
---|---|
Linux | 446 |
UNIX | 25 |
MS Windows | 5 |
BSD | 1 |
Mixed / Misc. | 23 |
Clearly, Microsoft’s near monopoly on the desktop does not cut much ice in this market.
As far as I know, there’s no prize, other than bragging rights, awarded to the winner. But it is interesting to see the progress being made in really high-performance computing.