Highlights: November 2011

Highlights

  • Fujitsu’s “K Computer” installed at the RIKEN Advanced Institute for Computational Science (AICS) in Kobe, Japan, is the No.1 system on the TOP500. Now fully build it achieved an impressive 10.51 Petaflop/s on the Linpack benchmark using 705,024 SPARC64 processing cores. The K computer partially built had taken the No. 1 position six month ago and is the first system ever to exceed the 10 Petaflop/s mark. Contrary to many other recent very large system it does not utilize graphics processors or other accelerators. The K Computer is also one of the most energy efficient systems on the list.
  • The Chinese Tianhe-1A system enjoyed the No. 1 on the TOP500 only for a single edition last November and is now the No. 2 with 2.57 Petaflop/s performance.
  • The largest U.S. system, Jaguar, installed at the Oak Ridge National Laboratory, is No. 3 with 1.75 Petaflop/s Linpack performance.
  • Roadrunner, the first system to break the petaflop barrier in June 2008, is now listed at No 10.
  • There are 10 systems listed which achieved greater than one Petaflop/s.
  • The two Chinese systems at No. 2 and No. 4 and the Japanese Tsubame 2.0 system at No. 5 are all using NVIDIA GPUs to accelerate computation and a total of 39 systems on the list are using GPU technology.
  • The most powerful system in Europe is a Bull system at the French CEA at No. 9.
  • China keeps increasing its number of systems to 75 and is now clearly the No. 2 country, as a user of HPC, ahead of Japan, UK, France, and Germany.
  • Intel continues to provide the processors for the largest share (76.8 percent) of TOP500 systems.
  • Intel’s Westmere processors increased their presence in the list strongly with 244 systems, compared with 175 in the last list.
  • Already 62 percent of the systems use processors with six or more cores.
  • 39 systems use GPUs as accelerators (up from 17 six month ago), 35 of these use NVIDIA chips, two use Cell processors, and two uses ATI Radeon.
  • The No. 5 system Tsubame 2.0 can run a Windows OS and achieve almost identical performance doing so.
  • Thanks to the size of the K computer system, Fujitsu captured the No. 2 spot in market share by total performance slightly ahead of Cray, but IBM stays well ahead of either.
  • The Cray’s XT system series remains very popular for big research customers with three systems in the TOP10.

Highlights from the Top 10:

  • Fujitsu’s “K Computer” installed at the RIKEN Advanced Institute for Computational Science (AICS) in Kobe, Japan, is the No.1 system on the TOP500. Now fully build it achieved an impressive 10.51 Petaflop/s on the Linpack benchmark using 705,024 SPARC64 processing cores. The K computer partially built had taken the No. 1 position six month ago and is the first system ever to exceed the 10 Petaflop/s mark. Contrary to many other recent very large system it does not utilize graphics processors or other accelerators. The K Computer is also one of the most energy efficient systems on the list.
  • The Chinese Tianhe-1A system enjoyed the No. 1 on the TOP500 only for a single edition last November and remains at the No. 2 with 2.57 Petaflop/s performance.
  • The largest U.S. system, Jaguar, installed at the Oak Ridge National Laboratory, is the No. 3 with 1.75 Petaflop/s Linpack performance.
  • Roadrunner, the first system to break the Petaflop/s barrier in June 2008, is now listed at No 10.
  • There are 10 systems listed which achieved greater than one Petaflop/s.
  • The U.S. is tops in Petaflop/s with five systems performing at the Petaflop/s level, Japan and China have each two and France has one.
  • The two Chinese systems at No.2 and No. 4 and the Japanese Tsubame 2.0 system at No. 5 are all using NVIDIA GPUs to accelerate computation and a total of 39 systems on the list are using GPU technology.
  • The most powerful system in Europe is a Bull system at the French CEA at No. 9. General highlights from the TOP500 since the last edition:
  • Already 62 percent of the systems use processors with six or more cores.
  • The entry level to the list moved up to the 50.9 Teraflop/s mark on the Linpack benchmark, compared to 39.1 Teraflop/s six months ago.
  • The last system on the newest list was listed at position 305 in the previous TOP500 just six months ago.
  • Total combined performance of all 500 systems has grown to 74.2 Petaflop/s, compared to 58.7 Petaflop/s six months ago and 43.7 Petaflop/s one year ago.
  • The entry point for the TOP100 increased in six months from 88.92 Teraflop/s to 115.9 Teraflop/s.
  • The average concurrency level in the TOP500 is 18,383 cores per system, up from 15,520 six months ago and 13,071 one year ago.
  • A total of 384 systems (76.8 percent) are now using Intel processors. This is slightly down from six months ago (386 systems 77.2 percent).
  • They are now followed by the AMD Opteron family with 63 systems (12.6 percent), down from 66.
  • The share of IBM Power processors has stabilized for now with 49 systems (9.8 percent), up from 45.
  • Intel’s Westmere processors increased their presence in the list strongly with 244 systems, compared with 175 in the last list.
  • 39 systems use GPUs as accelerators (up from 17 six month ago), 35 of these use NVIDIA chips, two use Cell processors, and two uses ATI Radeon.
  • Gigabit Ethernet is still the most-used internal system interconnect technology (223 systems, down from 230 systems), due to its widespread use at industrial customers, followed by InfiniBand technology with 213 systems, up from 208 systems.
  • However, InfiniBand-based systems account for almost twice as much performance (28.7 Pflop/s) than Gigabit Ethernet ones (14.2 Pflop/s).
  • IBM and Hewlett-Packard continue to sell the bulk of the systems at all performance levels of the TOP500.
  • IBM kept its lead in systems and has now 223 systems (44.6 percent) compared to HP with 140 systems (28.0 percent). HP had 146 systems (29.2 percent) six months ago, compared to IBM with 218 systems (43.6 percent).
  • IBM remains the clear leader in the TOP500 list in performance with 27.3 percent of installed total performance (up from 26.4 percent). Fujitsu gained the second spot due to the impressive performance of the No. 1 K Computer with 14.7 percent unchanged from 14.8 percent. Cray follows closely in third place in this category with 14.3 percent down from 15.5 percent. HP is fourth with 12.9 percent down from 14.0 percent.
  • In the system category, Cray, SGI, and Bull, and Appro follow with 5.4 percent, 3.3 percent, 3 percent, and 2.6 percent respectively.
  • IBM (161) and HP (121) together sold 282 out of 287 systems at commercial and industrial customers and have had this important market segment clearly cornered for some time now.
  • The U.S. is clearly the leading consumer of HPC systems with 263 of the 500 systems (up from 251). The European share (103 systems – down from 126) is lower than the Asian share (118 systems – up from 103).
  • Dominant countries in Asia are China with 74 systems (up from 61), Japan with 30 systems (up from 26).
  • In Europe, UK, France, and Germany, are almost equal with 27, 23, and 20 respectively.

Power consumption of supercomputers

TOP500 now tracks actual power consumption of supercomputers in a consistent fashion.

  • 29 systems on the list are confirmed to use more than 1 megawatt (MW) of power.
  • The No. 1 system, the K computer also reports the highest total power consumption of 12.66 MW.
  • Average power consumption of a TOP500 system is 634 kW (up from 543 kW six months ago and 447 kW one year ago).
  • Average power efficiency is 282 Mflops/watt (up from 248 Mflops/watt six months ago and 219 Mflops/watt one year ago).
  • Average power consumption of a TOP10 system is 4.56 MW (up from 4.3 MW six months ago) and average power efficiency is 464 Mflops/watt (unchanged).
  • Most energy efficient supercomputers are based on:
    • BlueGene/Q Prototype with 2,029 Mflops/watt (No. 17, 21, 65, 66)
    • A Chinese system called Mole-8.5 with nVidia accelerators with 919 Mflops/watt (No. 21)
    • Four Appro systems at Sandia, UCSD, LLNL, and LANL with up to 888 Mflops/watt (No. 62, 48, 15, and 44)
    • The TSUBAME 2.0 system at the GSIC Center in Tokyo with 865 Mflops/watt (No. 5)
    • The No. 1 Fujitsu K Computer at RIKEN with 830 Mflops/watt

Highlights from the TOP50:

  • The entry level into the TOP50 is at 205 Teraflop/s
  • The U.S. share of the TOP50 stayed at 46 percent up from 44 percent.
  • China, Germany, France, and Japan have 6, 5, 4, and 3 systems.
  • Cray leads the TOP50 with 30 percent of systems and 23 percent of performance.
  • IBM has 14 percent of systems and 11 percent of performance.
  • Fujitsu’s K computer system accounts for 27 percent of performance.
  • 60 percent of systems are installed at research labs and 28 percent at universities.
  • There is only a single system using Gigabit Ethernet in the TOP50.
  • The average concurrency level is 87,233 cores per system – up from 76,650 cores per system six months ago and 64,618 one year ago.

All changes are from June 2011 to November 2011.

About the TOP500 List

The TOP500 list is compiled by Hans Meuer of the University of Mannheim, Germany; Erich Strohmaier and Horst Simon of NERSC/Lawrence Berkeley National Laboratory; and Jack Dongarra of the University of Tennessee, Knoxville. For more information, visit www.top500.org