Three high-performance computers dedicated to life science research have debuted on the most recent listing of the world's fastest supercomputers.
Three life science computers — at MD Anderson Cancer Center, the Genome Science Center at the University of British Columbia, and the Laboratory for Systems Biology and Medicine at the University of Tokyo — that beat the 31.1-teraflop benchmark for inclusion in the 36th edition of the twice-yearly ranking.
The system at MD Anderson, a 48.1 teraflop, 8,064-core HP Cluster Platform, debuted at No. 169 on the list; the BC Genome Science Center machine, a 47.3-teraflop, 5,040-core IBM iDataPlex, came in at No. 178; and the University of Tokyo system, a 34.7-teraflop, 3,552-core Fujitsu Primergy, took the No. 377 spot.
The fastest life science system is a 97.1-teraflop, 18,176-core HP cluster at Pacific Northwest National Laboratory’s Environmental Molecular Sciences Laboratory, dubbed "Chinook," which holds the No. 75 spot on the current list — down from No. 57 in the June version of the ranking (BI 6/4/2010).
Three other life science systems remain from the June list: a 54-teraflop, 5,760-core Sun Microsystems blade system at the University of Tokyo's Human Genome Center that now holds the No. 138 spot, down from No. 96 six months ago; a 53.1-teraflop, 8,640-core system at the Georgia Institute of Technology's Center for the Study of Systems Biology that ranked at No. 98 on the June list and now holds the No. 140 spot; and a 35.8-teraflop, 4,000-core Dell system at the Howard Hughes Medical Institute's Janelia Farm Campus that debuted at the No. 208 spot in June but is now ranked No. 349.
One life science system in the June ranking, a 30.1-teraflop Dell system at Arizona State University and the Translational Genomics Research Institute, did not make the cutoff for the current ranking, bringing the total number of life science systems on the current list to seven — the most on the Top500 since 2005 (see Table 1, below, for details of the life science systems on the list).
Top500 Trends
A new system has claimed the title of the fastest supercomputer in the world — the Tianhe-1A system at China's National Supercomputer Center in Tianjin, which achieved a maximum performance level of 2.57 petaflops on the Linpack benchmark used to compile the ranking. It bumps the former number one system — a 1.75-petaflop Cray XT5 system at the US Department of Energy’s Oak Ridge Leadership Computing Facility — to second place.
IBM and HP remain the top supercomputing vendors, though Cray has made significant inroads with four of the top 10 machines — more than any other single manufacturer in the top 10.
IBM retains its lead over HP as the vendor with the most installations in the Top500, with 200, an increase of four systems over the June list. HP came in at second with 159 systems, a drop from 186 in June, while Cray gained eight placements since June to come in at third with a total of 29 installed systems on the Top500 (see Table 2, below, for details on manufacturer rankings).
Cray holds the number two spot in terms of market share measured in performance, ahead of HP, but behind IBM.
Clusters are still the most common architecture for systems in the Top500 list, though the number of clusters dropped to 415 from 424 in the June list. Massively parallel processing systems, meantime, gained a bit of ground with nine more systems in the current list than six months ago (see Table 3, below, for details).
Intel continues to dominate the market for high-performance processors, with 398 of all systems using Intel processors. Of those, 392 systems — or 78 percent of all systems in the Top500 — use chips from Intel's EM64T family. However, Intel's total has dropped slightly since June, when it claimed 406 systems.
AMD's Opteron family follows Intel with 57 systems, up from 47 six months ago, while IBM Power processors dropped to 40 systems from 42 in June (see Table 4, below, for details).
Quad-core processors are now used in 365 systems in the Top500, while 95 systems use processors with six or more cores.
In another trend, graphics processing units are making gains in the Top500, with three of the top 10 machines using Nvidia GPUs to accelerate computation. A total of 17 systems on the current ranking GPUs as accelerators, with six using the Cell processor, 10 using Nvidia chips, and one using ATI Radeon chips.
In terms of operating systems, Linux machines now claim 92 percent of all installations on the Top500 list, with 459 machines compared to 455 in the June list. Unix systems were the next most popular, with 19 systems on the current list, followed by "mixed" operating systems with 16 (see Table 5, below, for details).
The complete Top500 ranking is available here.
Have topics you'd like to see covered in BioInform? Contact the editor at btoner [at] genomeweb [.] com.