The December 2022 issue of IEEE Spectrum is here!

Close bar

Mirror, Mirror, On the Wall, Who's the Fastest of Them All?

With Oak Ridge’s 20 petaflop ‘Titan’ grabbing the supercomputing title, re-ranks the field

2 min read
Titan, ORNL's new 20 pflop supercomputer
Oak Ridge National Laboratory

The unveiling of Oak Ridge National Laboratory’s Titan Cray XK7 supercomputer knocks every other computer in the world down one notch in the Petaflop Hall of Fame. The Top500 Supercomputer Sites website has just posted new rankings noting that there are now 23 systems with performance better than a petaflop per second, “just four-and-a-half years after the debut of Roadrunner, the world’s first petaflop/s supercomputer.” (Los Alamos National Laboratory's Roadrunner is now 22nd on the list.)

Top500’s 12 November announcement  includes an analysis of today’s trends in supercomputing. Some highlights:

  • Multi-core systems dominate, and more systems (including Titan) are using processor/co-processor architectures.
  • Intel provides 76 percent of the supercomputer processors, followed by AMD (12 percent) and IBM (10.6 percent).
  • The threshold for membership in the Top 100 list has moved up to 241.3 teraflop per second, up from 172.7 Tflop/s just six months ago.

So here are the world's 10 most powerful supercomputers—for now: 

  • Titan (Cray XK7)—Oak Ridge National Laboratory, USA:  560 640 cores, 17 590 teraflop per second, 8 209 kilowatts.
  • Sequoia (IBM BlueGene/Q)—Lawrence Livermore National Laboratory, USA:  1 572 864 cores, 16 325 Tflop/s, 7890 kW
  • K computer (Fujitsu K computer)—RIKEN Advanced Institute for Computational Science, Japan:  705 024 cores, 10 510 Tflop/s, 12 660 kW
  • Mira  (IBM BlueGene/Q )—Argonne National Laboratory, USA: 786 432 cores, 8162 Tflop/s, 3945 kW
  • JUQUEEN (IBM  BlueGene/Q)—Forschungszentrum Juelich (FZJ), Germany: 393 216 cores, 4141 Tflop/s, 1970 kW
  • SuperMUC (IBM iDataPlex DX360M4)—Leibniz Rechenzentrum, Germany:  147 456 cores, 2897 Tflop/s, 3423 kW
  • Stampede  (Dell PowerEdge C82207)—Texas Advanced Computing Center, USA: 204 900 cores, 2660 Tflop/s , (power n/a)
  • Tianhe-1A ( NUDT YH MPP)—National Supercomputing Center in Tianjin, China: 186 368 cores, 2566 Tflop/s,  4040 kW
  • Fermi  (IBM BlueGene/Q)—CINECA, Italy:  163 840 cores, 1725 Tflop/s, 822 kW
  • DARPA Trial Subset  (IBM Power 775)—IBM Development Engineering, USA:  63 360 cores, 1515 Tflop/s, 3576 kW

See the Top500 site for the full list, highlights, analysis and a poster showing the evolution of big iron.

Images: Titan, Oak Ridge National Laboratory (top). Sequoia, IBM

The Conversation (0)

Why Functional Programming Should Be the Future of Software Development

It’s hard to learn, but your code will produce fewer nasty surprises

11 min read
A plate of spaghetti made from code
Shira Inbar

You’d expectthe longest and most costly phase in the lifecycle of a software product to be the initial development of the system, when all those great features are first imagined and then created. In fact, the hardest part comes later, during the maintenance phase. That’s when programmers pay the price for the shortcuts they took during development.

So why did they take shortcuts? Maybe they didn’t realize that they were cutting any corners. Only when their code was deployed and exercised by a lot of users did its hidden flaws come to light. And maybe the developers were rushed. Time-to-market pressures would almost guarantee that their software will contain more bugs than it would otherwise.

Keep Reading ↓Show less