The new edition of the ranking of the 500 highest performing computers has been published in the world, this being the 54th edition. Besides that toAt the same time, a new version of the alternative qualificationto Graph 500 for cluster systems, designed to evaluate the performance of supercomputer platforms related to the simulation of physical processes and the tasks of processing large data sets inherent in such systems. The Green500 rating is no longer issued separately and is combined with the Top500.
Regarding the new edition of the TOP 500 we may find that the top ten leaders have not changed since, first, in the ranking, IBM positions itself with the Summit cluster at Oak Ridge National Laboratory (USA). The cluster runs Red Hat Enterprise Linux and has 2,4 million cores processor (using 9-core IBM Power22 3.07C 22GHz CPUs and NVIDIA Tesla V100 accelerators), providing 148 petaflops performance.
While for the second place is occupied by the American cluster Sierra, installed by IBM at Livermore National Laboratory based on a platform similar to the Summit and demonstrating performance at the level of 94 petaflops (approximately 1,5 million cores).
Third, there is the Chinese Sunway TaihuLight cluster, which operates in the national supercomputer center of China, which includes more than 10 million cores and shows a throughput of 93 petaflops. Despite the close performance, the Sierra cluster consumes half the power of the Sunway TaihuLight.
Fourth is the Chinese Tianhe-2A cluster, which includes nearly 5 million cores and demonstrates the performance of 61 petaflops.
Frontera Cluster, manufactured by Dell for the Texas Computer Center, ranks fifth in the ranking. The cluster runs CentOS Linux 7 and includes more than 448 Xeon Platinum 8280 28C 2.7GHz-based cores. The total RAM size is 1.5 PB, and the performance reaches 23 petaflops, which is 6 times less than the leader in the rating.
Regarding the distribution by number of supercomputers in different countries, in this new edition, these are distributed as follows:
- China: 228 (219 - six months ago). In total, they generate 31.9% of all productivity (half a year ago - 29.9%).
- United States: 117 (116). Total productivity is estimated at 37.8% (half a year ago - 38.4%).
- Japan: 29 (continues with the same number as six months ago)
- France: 18 (19).
- Germany: 16 (14).
- Netherlands: 15 (13).
- Ireland: 14 (13).
- Great Britain: 11 (18).
- Canadá 9 (8).
- Italy: 5 (continues with the same number as six months ago).
- Singapore 4 (5).
- Australia, South Korea, Saudi Arabia, Brazil, Russia: 3.
The general distribution of the number of supercomputers in different parts of the world is as follows: 274 supercomputers are in Asia (267 six months ago), 129 in America (127) and 94 in Europe (98), 3 in Oceania.
In the classification of the operating systems used In supercomputers, Linux continues to dominate in the TOP 500.
Distribution on Linux distributions:
- 49.6% do not detail the distribution,
- 26.4% use CentOS.
- 6.8% Cray Linux.
- 4.8% RHEL.
- 3% SUSE.
- 2% Ubuntu.
- 0.4% Scientific Linux.
The minimum performance threshold to enter the Top500 for 6 months increased from 1022 to 1142 teraflops (Last year only 272 groups outperformed petaflops, 138 two years ago, 94 94 three years ago).
The total throughput of all systems in the classification for the year increased from 1.559 to 1.650 exaflops (three years ago there were 566 petaflops). The system that closes the current rating was 397º in the previous issuance and 311º in the previous year.
The main CPUs are Intel CPUs: 94% (half a year ago it was 95.6%), IBM Power: 2.8% (2.6%) in second place, AMD 0.6% (0.4%) in third place, 0.6% in fourth SPARC64 (0,8%).
35.6% (half a year ago 33.2%) of all used processors have 20 cores, 13.8% (16.8%) - 16 cores, 11.2% (11.2%) - 12 cores, 11% (11.2%) - 18 cores, 7.8% (7%) - 14 cores.
144 of 500 sets (half a year ago - 133) they also use accelerators or coprocessorsWhile 135 systems use NVIDIA chips (half a year ago there were 125), 5 use Intel Xeon Phi (5), 1 PEZY (1), 1 hybrid solutions are used (it was 1), in 1 - Matrix-2000 (1), in 1 AMD Vega GPU (ago six months, no AMD accelerators were used).
Among cluster manufacturers, Lenovo ranked first: 34.8% (34.6% a year ago), Sugon 14.2% (12.6%) ranked second, Inspur fell to third place - 13.2% (14.2%), Hewlett- ranked fourth Packard - 7% (8%) and 7% (7.8%), followed by Atos - 4.6%, IBM 2.6 (2.4%), Fujitsu 2.6% (2.6%), Penguin Computing - 2.2% (1.8%), Dell EMC 2.2% (3%), Huawei 2% (1.4%), NVIDIA 1.2%. Five years ago, the distribution among manufacturers was as follows: Hewlett-Packard 36%, IBM 35%, Cray 10.2%, and SGI 3.8%.