The latest edition of the
Top500 supercomputers was released on 19 June. The United States and China continue to lead the pack in terms of total number of systems, with 169 and 160 respectively, though the US is no longer represented in the top three - the first time since 1996.
This is due to the rise of the Swiss Piz Daint supercomputer, which is now in third place. Nevertheless, the US hosts five of the top ten systems.
Other finding of the June report include the growth in overall performance of the Top500, which has risen to 749 petaflops - a jump of 32 percent from a year ago. Meanwhile Intel has continued its dominance, with its Xeon or Xeon Phi chips found in 444 of the 500 supercomputers.
As for system vendors, HPE has the most top 500 systems, with 144 - 25 of which were built by SGI, which HPE acquired in 2016. It is followed by Lenovo, with 88, and Cray, with 57. Cray continues to lead with regards to overall processing power, claiming 21.4 percent of the list's total performance.
1. Sunway TaihuLight
© Jack Dongarra: Report on the Sunway TaihuLight System, June 2016
© Flickr/Sam Churchill
Bolstering the Chinese claim to supercomputing superpower is another monolith that has retained its previous high rank. Also known as Milky-Way-2, it was
knocked off the top spot last year by the Sunway TaihuLight, which boasts three times the speed of its predecessor. The supercomputer was built by China's Nation University of Defense Technology and was the number one system for three consecutive years before falling into second place.
Top speed: 33.9 petaflops Total cores: 3,120,000
3. Piz Daint
© Swiss National Supercomputing Centre (CSCS)
The new number three - and Europe's fastest supercomputer - was named after an Alpine mountain less than 80 miles from its Swiss National Computing Center home.
Occupying eight place in last November's rankings, Piz Daint was upgraded with additional NVIDIA P100 Tesla GPUs. This resulted in a significant performance improvement, from 9.8 petaflops to 19.6 petaflops.
Top speed: 19.6 petaflops Total cores: 361,760
© Wikipedia/US Department of Energy
The first American entry on the list has more than earned its imposing name. The Cray construction has contributed to research breakthroughs at the Oak Ridge Leadership Computing Facility (OLCF) that have improved nuclear power plant safety and performance, boosted drug development, and improved the understanding of climate change.
Top speed: 17.6 petaflops Total cores: 560,640
© Wikimedia Commons/Lawrence Livermore National Laboratory
The IBM construction was named the world’s fastest supercomputer in June 2012, but has slowly slipped down the list. It’s primarily used for nuclear weapons simulations.
Top speed: 17.2 petaflops Total cores: 1,572,864
Cori was named after American biochemist Gerty Cori, the first woman to win a Nobel Prize in physiology or medicine. The Cray creation is installed at Berkeley Lab's National Energy Research Scientific Computing Center (NERSC).
Top speed: 14 petaflops Total cores: 622,336
Japan’s highest entry in the TOP500 is powered by the same Intel “Knights Landing” Xeon Phi 7250 processor as the Cori computer that pipped it into the top five. Built by Fujitsu, it’s run jointly by the University of Tokyo and the University of Tsukuba.
Top speed: 13.6 petaflops Total cores: 556,104
8. K computer
© RIKEN Advanced Institute for Computational Science
The second consecutive Fujitsu-manufactured Japanese entrant on our list reached number one in its 2011 prime. It’s used in a range of fields including meteorology, disaster prevention and medicine.
Top speed: 10.5 petaflops Total cores: 705,024
© Wikipedia/Argonne National Laboratory
According to manufacturer IBM, if every person in the United States performed one calculation every second, it would take them almost a year to do as many calculations as Mira can in just one second. The machine was initially deployed to work on sixteen research projects selected by the Department of Energy.
Top speed: 8.6 petaflops Total cores: 786,432
© Los Alamos National Laboratory
Another Cray system, Trinity runs at the
Los Alamos National Laboratory.
Top speed: 8.1 petaflops Total cores: 301,056