Every year a new list of world’s most powerful supercomputers is published by TOP500. It’s a ranking that demonstrates the herculean capacity of these machines have and what are they used for. It is pretty interesting to witness this list every year; where every country on the planet is competing to be the part of the “Third Industrial Revolution (Read Internet)”, it is also a perfect time to understand what makes a supercomputer so super and what future holds for their uses.
What is a supercomputer?
A supercomputer is a computer with a high level of performance compared to a “general purpose computer”. The performance of a supercomputer is measured in “Floating Point operations per second” also known as FLOPS. Supercomputers can perform up to nearly a hundred quadrillions of FLOPS per second, measured in P(eta)FLOPS. In other words, a large server farm might be powering your Netflix subscription or your Google Search, but its computing power is focused on multiple independent tasks and not just a single one.
The top supercomputers are ranked based on P(eta)FLOPS that forms the basis of this list. So, let's find out the top 10 Supercomputers based on their computational power:
SUNWAY TAIHULIGHT (China) - The Sunway TaihuLight is a Chinese supercomputer which is ranked as number one as of March 2018 in the TOP500 list of the fastest supercomputers in the world. It has a LINPACK benchmark rating of 93 P(eta)FLOPS which is nearly three times as fast as the previous record holder Tianhe - 2. The Sunway TaihuLight is also ranked as the 16th most energy-efficient supercomputer, which makes it extremely desirable.
TIANHE - 2 - Tianhe - 2 is a supercomputer located in National Supercomputer Center in Guangzhou, China, powered with a 33.86 P(eta)FLOPS. It was developed by a team of 1300 scientists and engineers. It held the record of being the fastest supercomputer for June 2013, November 2013, June 2014, November 2014, June 2015 and November 2015 in TOP500 list. It was finally replaced by Sunway TaihuLight in June 2016.
PIZ DAINT - Piz Daint is a supercomputer named after the mountain Piz Daint in the Swiss Alps. Until the end of 2015, it was ranked 8th on the TOP500 list which is higher than any other supercomputer in Europe. Towards the end of 2016, the computing performance of Piz Daint was tripled to 25 P(eta)FLOPS which made it the third most powerful supercomputer in the world. Interestingly, the original Piz Daint system was installed in December 2012 which was later upgraded and combined with another system to build the new Piz Daint.
GYOUKOU - Gyoukou is a supercomputer deployed at Japan Agency for Marine-Earth Science and Technology Yokohama Institute for Earth Science. It was developed by ExaScaler and PEZY Computing. After an upgrade to the full scale using newer ZettaScaler - 2.2 system, Gyoukou achieved 19,135.8 TeraFLOPS on November 2017 taking a lead in the chart and standing at number 4 on the most powerful supercomputers on Earth. Gyoukou is highly energy efficient and ranks at number 5 on the November 2017 Green500 list.
TITAN - Built by Cray at Oak Ridge National Laboratory is an upgrade of Jaguar, a previous supercomputer at Oak Ridge. It uses GPUs in addition to conventional Central Processing Units (CPUs). Titan is the first such hybrid to perform over 10 P(eta)FLOPS which is available to researchers for a variety of science projects. Titan is going to be eclipsed in 2018 by Summit (by IBM) and will feature fewer nodes with much greater GPU capability per node.
IBM SEQUOIA - Sequoia is a supercomputer constructed by IBM for the National Nuclear Security Administration as a part of Advanced Simulation and Computing Program. Sequoia replaced the K Computer as the world’s fastest supercomputer in June 2012 with the performance of 17.17 P(eta)FLOPS. However, Sequoia had dropped to 6th position by November 2017 due to stiff competition from Tianhe - 2, Titan and others of the league. Sequoia completely runs on Linux with CNK running over 98000 nodes. It is equally interesting to know that Sequoia is very energy efficient that consumes 37% less than K computer.
TRINITY - This supercomputer is at Los Alamos National Laboratory in New Mexico, United States. It was originally built as the first Advanced Technology System for the National Nuclear Security Administration. As of November 2017, Trinity stands at 7th rank after an upgrade with Intel “Knights Landing” Xeon Phi processors. It debuted in November 2015 right at 6th place.
CRAY - This one is a massively parallel multiprocessor supercomputer manufactured by Cray that consists of Intel Haswell Xeon processors connected together by Cray’s proprietary “Aries” interconnect, stored in air-cooled cabinets. Cray is deployed in various countries including India, Finland, Australia, Poland, Germany, Saudi Arabia, Sweden, United States, United Kingdom and Switzerland.
OAK FOREST - PACS - Oakforest PACS is manufactured by Fujitsu at Kashiwa, Joint Center for Advanced High-Performance Computing Japan. Its LINPACK performance is 13554.6 TFLOPS and is has Linux as its core operating system. It is made up of 8208 computational nodes using Intel Xeon Phi high-performance processors with Knights Landing architecture.
K Computer - Another mammoth from Fujitsu’s clan of supercomputers, K Computer is installed at the Riken Advanced Institute for Computational Science campus in Kobe, Hyogo Prefecture, Japan. It is based on a distributed memory architecture with over 80000 computer nodes. K Computer became the first computer to top 10 P(eta)FLOPS in November 2011. At the moment K Computer’s LINPACK performance is 10.51 P(eta)FLOPS.
Who is winning the supercomputer war?
The race for computational speed has just started, The US, Japan, and China are leading the way for building the arsenal of supercomputers; however, Europe is scaling in a different way altogether. Europe plans to spend up to $102 billion to develop its own technology that aims to develop its own exascale machines (exaFLOPS) by 2022 - 23.
China still holds the top two spots in the list of fastest computers in the world, the US is preparing to go an extra mile to fight for the top spot. The Department of Energy awarded six companies a total of $258 million to further the research and development of world’s first exascale supercomputer.
So where are we heading?
Who is winning the supercomputer war might not be the major debate as we enter the third industrial revolution, however, the bigger question to answer would be will computers be centralized or distributed so that everyone becomes a part of sharing economy and prominent issues like Global Warming could be addressed with more innovative solutions through the computing power.
So, stay tuned and check out our upcoming blog on “Will computers be centralized or distributed?"