The United States is on high of the supercomputing world within the Top500 rating of probably the most highly effective programs. The Frontier system from Oak Ridge National Laboratory (ORNL) working on AMD EPYC CPUs took first place from final yr’s champ, Japan’s ARM A64X Fugaku system. It’s nonetheless within the integration and testing course of on the ORNL in Tennessee, however will finally be operated by the US Air Force and US Department of Energy.
Frontier, powered by Hewlett Packard Enterprise’s (HPE) Cray EX platform, was the highest machine by a large margin, too. It’s the primary (identified) true exascale system, hitting a peak 1.1 exaflops on the Linmark benchmark. Fugaku, in the meantime, managed lower than half that at 442 petaflops, which was nonetheless sufficient to maintain it in first place for the earlier two years.
Frontier was additionally probably the most environment friendly supercomputer, too. Running at simply 52.23 gigaflops per watt, it beat out Japan’s MN-3 system to seize first place on the Green500 listing. “The indisputable fact that the world’s quickest machine can be probably the most vitality environment friendly is simply merely wonderful,” ORNL lab director Thomas Zacharia stated at a press convention.
Other machines within the TOP10 embody one other HPE Cray EX system set up at EuroHPC in Finland (151.9 petaflops), the IBM-built Summit system utilizing 22-core Power( CPUs and NVIDIA Tesla V100 GPUs (148.8 petaflops) and Lawrence Livermore’s Sierra, a smaller-scale model of Summit that hit 94.6 Pflop/s.
China held two top-ten spots with its Sunway TaihuLight from the National Research Center of Parallel Computer Engineering & Technology (NRCPC) and Tianhe-2A constructed by China’s National University of Defense Technology (NUDT). However, China is rumored to have already got at least two exascale programs (in response to the Linmark benchmark) on new Sunway Oceanlite and Tianhe-3 programs. Due to the present state of semiconductor politics, nevertheless, China is reportedly not revealing any new benchmarks or essential advances.