1. The slightly mysterious Chinese one: Tianhe-1A
Tianhe-1 and Tianhe-1AActiveTianhe-1 Operational 29 October 2009, Tianhe-1A Operational 28 October 2010
Sponsors National University of Defense Technology
Operators National Supercomputing Center
Location National Supercomputing Center,Tianjin,People's Republic of China
Operating system. Linux[1]
Storage96 TB (98304 GB) for Tianhe-1,262TB for Tianhe-1A
SpeedTianhe-1: 563teraFLOPS(Rmax) 1,206.2teraFLOPS(Rpeak),Tianhe-1A: 2,566.0teraFLOPS(Rmax) 4,701.0teraFLOPS(Rpeak)
RankingTOP500: 2nd, June 2011 (Tianhe-1A)
Purpose Petroleum exploration, aircraft simulation
Sources. top500.org
Tianhe-I
Simplified Chinese天河一号
Traditional Chinese天河一號
Literal meaning"Milky Way No.1"
China's supercomputer is currently the world's fastest: it can run at a sustained 2.5 petaflops (a petaflop is a thousand trillion floating point operations per second) thanks to its 186,368 cores and 229,376GB of RAM.While the horsepower comes from off-the-shelf Intel and Nvidia chips, theNew York Times saysthat the Chinese machine's speed is down to its interconnect, the networking technology that connects the individual nodes of the computer together,which is twice as fast as the InfiniBand technology used in many other supercomputers.It's located in Shenzhen's National Supercomputing Centre, where it's used byuniversities and Chinese companies.
2. The one with a quarter of a million cores: Jaguar
Jaguar, a Cray XT5-HE supercomputer located at the USOak Ridge National Laboratory, has quite a few cores: TOP500says there are nearly a quarter of a million since its most recent upgrade.Jaguar's 224,162 cores come courtesy of a whole bunch of six-core Opteron chips, and its performance is a hefty 1.76 petaflops. Oak Ridge says it's the world's fastest supercomputer for unclassified research.
3. The other slightly mysterious Chinese one: Dawning Nebulae
Nebulae (computer)
Active 2011
Location. National Supercomputing Center (Shenzhen)
Architecture. TC3600 Blade system.
Power. 2,580.00 kW
Operating systemLinux[1]
Speed. 2,984.3 TFlops/s （Rmax）
1,271.0 TFlops/s （Rpeak）
Ranking. TOP500: 2, June 2010
Sources. Top500
Active 2011
Location. National Supercomputing Center (Shenzhen)
Architecture. TC3600 Blade system.
Power. 2,580.00 kW
Operating systemLinux[1]
Speed. 2,984.3 TFlops/s （Rmax）
1,271.0 TFlops/s （Rpeak）
Ranking. TOP500: 2, June 2010
Sources. Top500
When it launched in early 2010 the Chinese Dawning Nebulae supercomputer was the world's fastest, with performance of 1.27 petaflops, but it's already in third place thanks to Jaguar and China's own newer, faster Tianhe-1A. Like its sibling Nebulae is in the National SupercomputingCentre in Shenzhen.
4. The one with the rubbish name: TSUBAME 2.0
Tsubameis asupercomputerthat operates at the GSIC Center at theTokyo Institute of Technologyin Japan. It has a peak of 2,288Tflopsand in June 2011 ranked 5th in the world.It was developed at the Tokyo Institute of Technology in collaboration withNECandHP, and has 1,400 nodes using bothHPProliantandNVIDIA Teslaprocessors.
Tokyo's TSUBAME 2.0 offers similar performance to Jaguar - it peaks at 2.3 petaflops, with sustained performance of around 1.4 petaflops - but it's one-quarter of the size and uses one-quarter of the power thanks to its heavy reliance on Nvidia Fermi GPUs as well as Intel CPUs.According to project lead Professor Satoshi Matsuoka, TSUBAME will really shine in climate and weather forecasting, biomolecular modelling and tsunami simulations.
5. The planet-saver: Hopper
Hopper is working on the big stuff: climatechange, clean energy, astrophysics, particle physics... its home, the US Department of Energy'sNational Energy Research Scientific Computing Center, offers its services to more than 3,000 researchers in the fields of climate research, chemistry, new material development and other crucial fields.
6. The French one: Tera-100
The first petaflop-scale supercomputer to bedesigned and built in Europeis pretty fast: "its capacity to transfer information isequivalent to a million people watching high-definition films simultaneously", the press release says.Built around Intel Xeon 7500 processors, the successor to 2005's Tera 10 is 20 times faster and seven times more energy efficient as its predecessor. It's another nuclear one: Tera-100's mission is to help guarantee the reliability of Europe's nukes
The result of a collaborative program between Bull and the CEA which began in 2008, Tera 100 is the first petaflops-scale* supercomputer ever designed and developed in Europe. Its theoretical maximum power of 1.25 Petaflops means it ranks among the three most powerful supercomputers in the world. Tera 100 is destined for the French nuclear weapons simulationprogram, aimed at guaranteeing the reliability of nuclear deterrent weapons.Tera 100 was powered up on 26 May 2010, just a few weeks after its installation in March 2010. Tera 100 consists of 4,300 bullx™ S Series servers, the model announced on the market by Bull in April 2010. It features 140,000 Intel® Xeon® 7500 processing cores, 300TB of central memory and a total storage capacity of over 20PB. Its 500GB/sec throughput to the global file system is a world record for a system of this type.Tera 100 offers exceptional processing capacity. By way of comparison, it can effectively carry out more operations in a single second than the world’s population would be capable of performing in 48 hours if each person completed one operation a second, day and night. Its capacity to transfer information is equivalent to a million people watching high-definition films simultaneously and its storage capacity corresponds to over 25 billion books.“Tera 100 being powered up represents a significant industrial success,” commented Jean Gonnord, Computer Simulation and IT Project Director at the CEA. “It highlights both the CEA’s and Bull’s expertise in developing ultra high-performance technologies, to the highest level worldwide, and it fully validates the industrial and research partnership that the CEA and Bull have succeeded in developing: a partnership whose outputs will immediately benefit the whole European scientific and industrial community.”“We are extremely proud of this successful achievement in petaflops-scale systems,” confirmed Philippe Miltin, Vice-President of Bull’s Products and Systems Division. “These kinds of technologies are not only fundamentally important for applications such as those at the CEA, but also for the design of the new generation of ‘computing power plants’ and massive Cloud Computing infrastructures; which is why expertise in petaflops technologies is a major asset for France, and Europe as a whole.”“Representing the biggest system ever designed around Intel® Xeon® processors, Tera 100 demonstrates the appropriateness of using Intel processors for High-Performance Computing, in termsof cost, power consumption and processing power. We are very proud to be involved in this major project, alongside the CEA and Bull,” commented Kirk Skaugen, Vice-President and Group Data Center General Manager, Intel.
7.Theformerchampion:Roadrunner
Supercomputing is a fast-moving field, andRoadrunneris the proof: in 2008 it was the first supercomputer to crack the petaflop barrier for sustained performance, but its 1.04 petaflop speed means it fell to seventh place in just two years.Built by IBM for the US Department of Energy, it was designed to work out whether the US's nuclear weapons would remain safe as they age - although like most supercomputers it's also available toindustry, with car and aerospace industries paying for a go.
8. The answer to life, the universe and everything: Kraken
Can your computing project be handled by a machine with 511 cores? Then don't bother coming to Kraken: it's best suited to jobs that use "at least 512 cores". It's got plenty to spare: theNational Institute for Computational Sciences reportsthat the Cray supercomputer has 112,895 compute cores spread across 9,408 nodes.Itspurpose? To help "solve the world's greatest scientific challenges, such as understanding the fundamentals of matterand unlocking the secrets to the origin of our universe".
9. The ultimate DVD ripper: JUGENE
Germany's supercomputer was designed for low power consumption as well as highperformance, and it's been involved in some interesting projects - including tryingto work out how DVDs work. According toScientific Computing, it's improving our understanding of "the processes involved in writing and erasing a DVD", which should lead to storage media that works better, lasts longer and provides higher capacity.
JUGENE(Jülich Blue Gene) was asupercomputerbuiltbyIBMforForschungszentrum Jülichin Germany. It was based on theBlue Gene/Pand succeeded theJUBLbased on an earlier design. It was at the introduction the second fastest computer in the world,and the month before its decommissioning in July 2012 it was still at the 25th position in theTOP500list. The computer was owned by the "Jülich Supercomputing Centre" (JSC) and theGauss Centre for Supercomputing.
10. The one keeping nukes saf:Cielo
Nuclear science and supercomputers are a match made in heaven: the former can use the latter to test things without blowing anybody up or irradiating them for generations.Cielois used for "classified operations" by the US National Nuclear Security Administration, and it's getting a big upgrade this year: its 6,704 computing nodes will be upped to 9,000, and its memory will go from 221.5TB to around 300TB.
Hi. I’m Designer of Blog Magic. I’m CEO/Founder of HackerTheme. I’m Creative Art Director, Web Designer, UI/UX Designer, Interaction Designer, Industrial Designer, Web Developer, Business Enthusiast, StartUp Enthusiast, Speaker, Writer and Photographer. Inspired to make things looks better.
0 comments:
Post a Comment