Must Read

10 of the world's fastest supercomputers

The cream of the Top500 supercomputer list

By Tim Greene, Executive Editor, IDG News Service | PT

Network World [slideshow] - Top 10 Supercomputers 2018 [slide-01]

10 of the world's fastest supercomputers

The semi-annual Top500 ranking of the world’s fastest supercomputers is in for fall 2018 with China claiming 227 of the 500 spots on the list, although it managed to take just two places in the top 10. The United states took five of the top 10, including first and second place. New to the Top500 rankings at number 205 is Astra, an HPE-built machine at Sandia National Laboratories that is the first powered by ARM chips to make the list. The top 10 highlighted in this slideshow demonstrate what might become available in corporate data centers.

IBM Sequoia is a petascale Blue Gene/Q supercomputer

10. Sequoia

This U.S. Department of Energy computer was the world’s most powerful in 2012 and has managed to cling to the top 10 ever since. This past summer Sequoia ranked number 8, but it dropped to number 10 in the latest list. It is an IBM BlueGene/Q system that can deliver 17.17 petaflops using 1,572,864 cores and peaks out at 20.13 petaflops.

Titan supercomputer at Oak Ridge National Laboratory [ORNL]

9. Titan

Once the most powerful supercomputer in the U.S., the Cray XK7 Titan installed at the U.S. Department of Energy Oak Ridge National Laboratory now ranks at number 9 on the international list, sliding down two places from number 7 in June. It achieved 17.6 petaflops using NVIDIA K20x GPU accelerators.

General view of the SuperMUC-NG, LRZ / Leibniz Supercomputing Centre, Garching bei München, Germany

8. SuperMUC-NG

This new-to-the-list supercomputer, SuperMUC-NG was built for the Leibniz Supercomputing Centre in Germany by Lenovo, comprised of Intel Xeon Scalable Skyklake processors. It is the most powerful supercomputer powered solely by x86 processors – more than more than 311,040 cores – and can deliver HPL performance of 19.5 petaflops.

AI Bridging Cloud Infrastructure supercomputer

7. AI Bridging Cloud Infrastructure (ABCI)

Installed at the National Institute of Advanced Industrial Science and Technology in Japan, AI Bridging Cloud Infrastructure (ABCI) is an energy-efficient machine built by Fujitsu using Primergy CX2550 servers equipped with Xeon Gold processors and Nvidia Tesla V100 GPUs. It’s capable of 19.9 petaflops and boasts energy efficiency of 12.05 gigaflops/watt.

The Trinity supercomputer at the Los Alamos National Laboratory [LANL]

6. Trinity

Trinity, a Cray XC40 system, improved its performance up to 20.2 petaflops from 14.14 petaflops, and also boosted its ranking one slot from number 9 last spring to number 6 now. It’s the only top 10 system to use Intel Xeon Phi processors and is housed at the U.S. Department of Energy’s Los Alamos National Laboratory. Its power efficiency is 3.678 gigaflops/watt.

The Cray XC30 \'Piz Daint\' system at the Swiss National Supercomputing Centre

5. Piz Daint

Piz Daint continued its steady ascent up the Top500 list, reaching number 5 this fall after listing number 6 earlier this year. It started its climb back in 2012 when it jumped into the Top500 at number 114. Stationed at the Swiss National Supercomputing Centre in Lugano, Switzerland, it is the most powerful supercomputer in Europe at 21.2 petaflops. It’s powered by Intel Xeon processors and NVIDIA Tesla P100 GPUs.

4. Tianhe-2A

This system at the National Super Computer Center in Guangzhou, China, maintained its ranking of number 4 from the last Top500 list. It is powered by Intel Xeon E5-2692v2 and Matrix-2000 processors with a core count near 5 million. Its maximum performance is 61.44 petaflops. Its power efficiency is 3.325 gigaflops per watt.

Sunway TaihuLight supercomputer, National Supercomputing Center in Wuxi Jiangsu, China

3. Sunway TaihuLight

Sunway TaihuLight spent two years at the number 1 spot, but has been dropping, ranking number 2 in June and now number 3. Installed at China’s National Supercomputing Center in Wuxi, it’s HPL performance of 93.0 petaflops. It’s notable for not using any accelerator chips, relying instead on 40,960 Sunway 26010 processors, each with 260 cores. Its power efficiency is 6.051 gigaflops/watt.

sierra supercomputer lawrence livermore national laboratory

2. Sierra

The U.S. Department of Energy’s Oak Lawrence Livermore National Laboratory houses Sierra, which is scrapping its way up from number 3 in the last ranking to number 2 by improving its performance from 71.6 petaflops to 94.6 petaflops. It has 1,572,480 cores supplied by IBM Power9 processors and boosted by Nvidia Volta GV100 accelerators, which add another 1,382,400 cores to the mix.

1. Summit

After being named the fastest supercomputer in its debut earlier this year, Summit didn’t rest on its laurels. It increased its performance from 122.3 petaflops to 143.5 petaflops to remain in first place. It was built for the U.S. Department of Energy’s Oak Ridge National Laboratory with 2,282,544 IBM Power9 cores and 2,090,880 Nvidia Volta GV100 cores. Summit has a theoretical peak performance of 187.66 petaflops and maintains a power efficiency of 11.324 gigaflops/watt.