The first supercomputer, Atlas, dates back to the early ‘60s. It was installed in Manchester University, and is far less powerful than ordinary desktop computers in our homes today. Looking around the world, it’s quite obvious that technology has taken a big leap since the hazy ‘60s. Every five years, in fact, the supercomputers of today become obsolete, thanks to rapidly advancing technology in information and computer development and research.
Today’s supercomputers are measured in PetaFLOPS, a processing speed equal to a million billion, or a thousand trillion, floating point operations per second. These machines are made to help scientists and meteorologists forecast global warming and weather; to stimulate brain activity or the effects of global warming; to advance nuclear technology and security, and much more throughout the world.
According to a senior scientist at the Beijing Computing Center, only a small fraction of the total capacity of these monolithic supercomputer centers are being used today. He claims that, “The supercomputer bubble is worse than a real estate bubble... [because] a computer, no matter how fast it is today, will become garbage in five years.”
Despite that claim, nations are still investing billions of dollars a year in the race to advance technology and become the top players in the tech and supercomputer world. Rather than rate the top ten supercomputers by speed (because that fluctuates so much), below are the world's ten most impressive supercomputers are ranked by their eye watering price tags.
10 IBM Roadrunner (US) - $130 million
The Roadrunner was built by IBM for the Los Alamos National Laboratory in New Mexico, USA. It became operational in 2008, and was designed for a peak performance of 1.7 petaFLOPS. On May 25, 2008, it achieved 1.026 PFLOPS, becoming the world’s first TOP500 Linpack sustained 1.0 petaflops system. It eventually reached a top performance of 1.456 PFLOPS in November of the same year, retaining its top spot on the TOP500 list.
According to the Supermicro Green500 list, in 2008, Roadrunner was the fourth-most energy-efficient supercomputer in the world. The supercomputer was decommissioned on March 31, 2013, and replaced with a smaller, more energy efficient supercomputer called Cielo. The purpose of Roadrunner was highly classified: to model the decay of the US nuclear arsenal.
9 Vulcan BlueGene/Q (US) - $100 million
Vulcan is a 24-rack supercomputer system that was created by IBM for the DoE and is stationed at the Lawrence Livermore National Laboratory in Livermore, California. It has a 5 PetaFLOPS peak, and is currently the ninth fastest supercomputer in the world, according to Top500.org. The BlueGene/Q is the third generation of IBM projects (after BlueGene/L and BlueGene/P) that aims at creating supercomputers that can reach operating speeds in the PFLOPS range, with low power consumption.
Vulcan became usable in 2013, and Lawrence Livermore National Laboratory for research in biology, plasma physics, climate science, molecular systems, solid and fluid engineering, and other complex subjects of study. It is also used in support of DoE and National Nuclear Security Administration (NNSA) missions.
8 SuperMUC (Germany) - $111 million
SuperMUC is currently the 14th fastest supercomputer in the world. It was formerly the 10th fastest in 2013, but with the speed at which technology advances, it was soon surpassed. Nonetheless, it is the second-fastest supercomputer in Germany (behind current #8, JUQUEEN). SuperMUC is operated by the Leibniz Supercomputing Centre (LRZ) at the Bavarian Academy of Sciences. It's housed near Munich.
The system was created by IBM, operates on Linux, contains over 19,000 Intel and Westmere-EX processors, and has a peak performance of a little over 3 PFLOPS. The system is noted for its new form of cooling that IBM developed, called Aquasar, which uses hot water to cool the processors. The design cuts the cooling electricity usage by 40%.
SuperMUC is used by European researches in a number of fields, including medicine, astrophysics, quantum chromodynamics, computational fluid dynamics, life sciences, computational chemistry, genome analysis, and earth quake simulations.
7 Trinity (US) - $174 million
Despite what it is being used for, you might expect the Trinity supercomputer to be even more expensive. But with newer, stronger technology comes a paralleled reduction of costs associated with creating newer, more powerful supercomputers. The US government offered supercomputer manufacturers Cray a $174 million contract to build this Cray XC supercomputer, along with a Cray Sonexion storage system for the National Nuclear Security Administration (NNSA). Trinity will be used to keep America’s nuclear arsenal secure, safe, and effective.
Trinity will be a joint effort between the Sandia National Laboratory and Los Alamos National Laboratory, as part of the NNSA’s Advanced Simulation and Computing Program. It’s unclear when Trinity will be up and running, and if the supercomputer will run the nation’s nuclear arsenal in the event that the US should need to deploy nukes in combat situations. Whatever the case, let’s hope that Trinity is in the right hands.
6 Sequoia BlueGene/Q (US) - $250 million
The petascale BlueGene/Q supercomputer Sequoia was developed by IBM, again for the NNSA, as part of the Advanced Simulation and Computing Program. It was deployed in June 2012 at the Lawrence Livermore National Laboratory, where it immediately became the world’s fastest supercomputer, according to TOP500.org. It currently sits in the number three spot, with a theoretical peak of 20 PFLOPS, or 20 trillion calculations per second.
Sequoia was the first supercomputer to cross 10 petaFLOPS of sustained performance, and some record-breaking science applications have been run on the system. For instance, the Cardioid code - a project that models the electrophysiology of the human heart - achieved nearly 12 PFLOPS with a real-time simulation. Other purposes of the computer are to study astronomy, energy, human genome, climate change, and of course nuclear weapons.
5 ASC Purple and BlueGene/L (US) - $290 million
These two supercomputers came as a pack. The two computers were announced by the DoE in 2002 to be contracted out to IBM for $290 million. They were installed in 2005 in the Lawrence Livermore Lab, and were decommissioned in 2010. At the time, the ASC Purple was ranked 66th on the TOP500 supercomputers list. The BlueGene/L was an older generation and inferior model to the BlueGene/Q, which system currently has four different supercomputers on TOP500’s top 10 list.
The ASCI Purple at the Lawrence Livermore Lab was built as stage five of the US Department of Energy and the NNSA’s Advanced Simulation and Computing Program, built to simulate and replace live WMD testing. The BlueGene/L focused on important scientific areas, such as predicting global climate change, and studying the interaction between atmospheric density and pollution.
At the time, at a press conference, the DoE announced that these two systems would have 1.5 times more processing power than all other 500 machines on the 2002 TOP500 list combined.
4 Sierra and Summit (US) - $325 million
Nvidia and IBM will soon help America to reclaim its top position in supercomputer speeds, tech breakthroughs, scientific research, and economic and national security. Built using IBM Power Servers and Nvidia Tesla GPU accelerators, the two supercomputers dubbed Sierra and Summit will be installed in 2017.
Currently, China’s Tianhe-2 supercomputer is the world’s fastest, able to deliver 55 PFLOPS of power, which is twice more than the second on the list. The upcoming Sierra system will have no problem running at over 100 PFLOPS, while Summit will have processing capabilities of as much as 300 PFLOPS.
Sierra’s purpose at the Lawrence Livermore National Laboratory will be to ensure safety and effectiveness of (you guessed it) the nation’s nuclear program. Meanwhile, Summit will the Oak Ridge National Laboratory’s aging Titan supercomputer, meant for scientific applications around the world.
3 Tianhe-2 (China) - $390 million
As mentioned above, China’s Tianhe-2 (translated to “Milky Way-2” in English) is the world’s fastest current supercomputer. Tianhe-2 was developed by a team of 1,300 scientists and engineers, and it is located in National Supercomputer Center in Guangzhou. Since 2013 it has consecutively ranked #1 on TOP500's list of fastest supercomputers. Tianhe-2 was sponsored by the 863 High Technology Program, and was initiated by the Chinese government and the government of Guangzhou province.
It was built by China’s National University of Defense Technology (NUDT) after the US government rejected Intel’s application for an export license for the CPUs and coprocessor boards. This move by the US was a blow to Intel and their suppliers, and a drag to US information technology development, while also being a boost for China’s own processor-development industry.
Tianhe-2 is able to perform 33,860 trillion calculations per second. One hour of these calculations by the supercomputer is the equivalent of 1,000 years of difficult sums by 1.3 billion people. As unfathomable as that number is, the future only grows brighter for information technology. Tianhe-2 is used for simulation, analysis, and government security applications.
2 Earth Simulator (Japan) - $500 million
The Earth Simulator (quite the ominous name) was developed by the Japanese government way back in 1997. The project cost 60 billion yen, or roughly $500 million in today’s economy. It was developed as a highly parallel vector supercomputer system, used to run global climate models, and to evaluate the effects of global warming and problems in solid earth geophysics.
The Earth Simulator (ES) was completed in 2002, developed for Japan’s Aerospace Exploration Agency, the Japan Atomic Energy Research Institute, and the Japan Marine Science and Technology Center.
ES was the fastest supercomputer in the world from 2002 to 2004. As you can expect from the rapid growth of technology since then, it doesn’t hold a candle to the speed of modern supercomputers, but it was big news in the early 21st century. The Earth Simulator System has several features to help protect the computer from earthquakes (rubber supports on a seismic isolation system) and lightning (a high-voltage, shielded nest that hangs over the building).
1 Fujitsu K (Japan) - $1.2 billion
Despite having the two most expensive supercomputers in the world, Japan’s heralded technology has been lacking in the supercomputer department as of late. Still, the K computer, named for the Japanese word “kei,” and meaning 10 quadrillion, is the fourth fastest supercomputer in the world, with a theoretical peak speed of 11 PFLOPS. The system cost 140 billion yen, or $1.2 billion to create.
In 2011, TOP500 ranked K the world’s fastest supercomputer, and in November 2011 the system became the first computer to top 10 PFLOPS officially. In 2012, K was superceded by IBM’s Sequoia as the world’s fastest supercomputer. The K computer, located at the RIKEN Advanced Institute for Computational Science, is 60 times faster than the Earth Simulator. K costs $10 million a year to operate, using 9.89 MW of power, or the equivalent of almost 10,000 suburban homes, or one million linked desktop computers.