China’s Milkyway 2 has ranked as the fastest high-performance computing system in the world for a fifth consecutive time on the bi-annual Top500 list of the most powerful supercomputers in the world.
Also known as Tianhe-2, it remained at 33.86 petaflops (quadrillions of calculations per second), which was almost double that of system that took second spot — the US Department of Energy’s Titan supercomputer.
There has been little change among the 10 fastest supercomputers on the list in recent years.
The only change in the June 2015 edition was in the seventh spot — the Shaheen II Cray XC40 system installed at King Abdullah University of Science and Technology. The Saudi Arabian system achieved 5.536 petaflops to become the highest-ranked Middle East system in the history of the Top500 list and first from the region to break into the top 10.
The number of supercomputers on the list using accelerators, such as Intel’s Xeon Phi chips or Nvidia’s GPUs, grew to 88 systems. Phi powered Milkyway 2 and Stampede, the Texas Advanced Computing Center’s system at the University of Texas at Austin that came in second. Nvidia GPUs were used in Titan and Piz Daint, the system at the Swiss National Supercomputing Center in the sixth spot.
More than 86 percent of Top500 systems are using Intel processors.
The Linpack benchmark used to rank the Top500 list has given way in recent years to the High Performance Conjugate Gradients (HPCG) benchmark, which hopes to make a more relevant metric for evaluating HPC systems.
A list of fastest supercomputers according to the HPCG benchmark had its third edition was released this week as well. It is mostly a mix of similar Top500 systems and also has the Milkyway 2 system in the top.