Supercomputer#The TOP500 list

{{Main|History of supercomputing}}

File:IBM 7030 Stretch circuit board.jpg]]

File:CDC 6600.jc.jpg

File:Cray-1-deutsches-museum.jpg preserved at the Deutsches Museum ]]

In 1960, UNIVAC built the Livermore Atomic Research Computer (LARC), today considered among the first supercomputers, for the US Navy Research and Development Center. It still used high-speed drum memory, rather than the newly emerging disk drive technology.{{cite book |title=Computers: The Life Story of a Technology |author1=Eric G. Swedin |author2=David L. Ferro |publisher = JHU Press|isbn= 9780801887741 |year=2007| page=57}} Also, among the first supercomputers was the IBM 7030 Stretch. The IBM 7030 was built by IBM for the Los Alamos National Laboratory, which then in 1955 had requested a computer 100 times faster than any existing computer. The IBM 7030 used transistors, magnetic core memory, pipelined instructions, prefetched data through a memory controller and included pioneering random access disk drives. The IBM 7030 was completed in 1961 and despite not meeting the challenge of a hundredfold increase in performance, it was purchased by the Los Alamos National Laboratory. Customers in England and France also bought the computer, and it became the basis for the IBM 7950 Harvest, a supercomputer built for cryptanalysis.{{cite book |title=Computers: The Life Story of a Technology |author1=Eric G. Swedin |author2=David L. Ferro |publisher = JHU Press|isbn= 9780801887741 |year=2007| page=56}}

The third pioneering supercomputer project in the early 1960s was the Atlas at the University of Manchester, built by a team led by Tom Kilburn. He designed the Atlas to have memory space for up to a million words of 48 bits, but because magnetic storage with such a capacity was unaffordable, the actual core memory of the Atlas was only 16,000 words, with a drum providing memory for a further 96,000 words. The Atlas Supervisor swapped data in the form of pages between the magnetic core and the drum. The Atlas operating system also introduced time-sharing to supercomputing, so that more than one program could be executed on the supercomputer at any one time.{{cite book |title=Computers: The Life Story of a Technology |author1=Eric G. Swedin |author2=David L. Ferro |publisher = JHU Press|isbn= 9780801887741 |year=2007| page=58}} Atlas was a joint venture between Ferranti and Manchester University and was designed to operate at processing speeds approaching one microsecond per instruction, about one million instructions per second.{{citation |title=The Atlas |url=http://www.computer50.org/kgill/atlas/atlas.html |publisher=University of Manchester |access-date=21 September 2010 |url-status=dead |archive-url=https://web.archive.org/web/20120728105352/http://www.computer50.org/kgill/atlas/atlas.html |archive-date=28 July 2012 }}

The CDC 6600, designed by Seymour Cray, was finished in 1964 and marked the transition from germanium to silicon transistors. Silicon transistors could run more quickly and the overheating problem was solved by introducing refrigeration to the supercomputer design.The Supermen, Charles Murray, Wiley & Sons, 1997. Thus, the CDC6600 became the fastest computer in the world. Given that the 6600 outperformed all the other contemporary computers by about 10 times, it was dubbed a supercomputer and defined the supercomputing market, when one hundred computers were sold at $8 million each.{{cite book|author=Paul E. Ceruzzi|title=A History of Modern Computing|url=https://archive.org/details/historyofmodernc00ceru_0|url-access=registration|year=2003|publisher=MIT Press|isbn=978-0-262-53203-7|page=[https://archive.org/details/historyofmodernc00ceru_0/page/161 161]}}{{cite book |title=History of computing in education |publisher=Springer Science & Business Media |author1=John Impagliazzo |author2=John A. N. Lee |year=2004 |isbn= 978-1-4020-8135-4 |page=[https://archive.org/details/springer_10.1007-b98985/page/n179 172] |url=https://archive.org/details/springer_10.1007-b98985}}{{cite book|author1=Andrew R. L. Cayton|author2=Richard Sisson|author3=Chris Zacher|title=The American Midwest: An Interpretive Encyclopedia|url=https://books.google.com/books?id=n3Xn7jMx1RYC&pg=PA1489|year=2006|publisher=Indiana University Press|isbn=978-0-253-00349-2|page=1489}}

Cray left CDC in 1972 to form his own company, Cray Research.{{cite book |title=Wisconsin Biographical Dictionary |first=Caryn |last= Hannan |year=2008 |isbn=978-1-878592-63-7 |pages= 83–84 |url=https://books.google.com/books?id=V08bjkJeXkAC&pg=PA83|publisher=State History Publications}} Four years after leaving CDC, Cray delivered the 80 MHz Cray-1 in 1976, which became one of the most successful supercomputers in history.Readings in computer architecture by Mark Donald Hill, Norman Paul Jouppi, Gurindar Sohi 1999 {{ISBN|978-1-55860-539-8}} page 41-48Milestones in computer science and information technology by Edwin D. Reilly 2003 {{ISBN|1-57356-521-0}} page 65 The Cray-2 was released in 1985. It had eight central processing units (CPUs), liquid cooling and the electronics coolant liquid Fluorinert was pumped through the supercomputer architecture. It reached 1.9 gigaFLOPS, making it the first supercomputer to break the gigaflop barrier.Due to Soviet propaganda, it can be read sometimes that the Soviet supercomputer M13 was the first to reach the gigaflops barrier. Actually, the M13 building began in 1984, but it was not operational before 1986. [https://www.computer-museum.ru/english/galglory_en/Rogachev.php Rogachev Yury Vasilievich, Russian Virtual Computer Museum]

{{Main|Supercomputer architecture|Parallel computer hardware}}

File:BlueGeneL cabinet.jpg/L, showing the stacked blades, each holding many processors]]

The only computer to seriously challenge the Cray-1's performance in the 1970s was the ILLIAC IV. This machine was the first realized example of a true massively parallel computer, in which many processors worked together to solve different parts of a single larger problem. In contrast with the vector systems, which were designed to run a single stream of data as quickly as possible, in this concept, the computer instead feeds separate parts of the data to entirely different processors and then recombines the results. The ILLIAC's design was finalized in 1966 with 256 processors and offer speed up to 1 GFLOPS, compared to the 1970s Cray-1's peak of 250 MFLOPS. However, development problems led to only 64 processors being built, and the system could never operate more quickly than about 200 MFLOPS while being much larger and more complex than the Cray. Another problem was that writing software for the system was difficult, and getting peak performance from it was a matter of serious effort.

But the partial success of the ILLIAC IV was widely seen as pointing the way to the future of supercomputing. Cray argued against this, famously quipping that "If you were plowing a field, which would you rather use? Two strong oxen or 1024 chickens?"{{cite web|url=https://www.brainyquote.com/quotes/seymour_cray_103779|title=Seymour Cray Quotes|website=BrainyQuote}} But by the early 1980s, several teams were working on parallel designs with thousands of processors, notably the Connection Machine (CM) that developed from research at MIT. The CM-1 used as many as 65,536 simplified custom microprocessors connected together in a network to share data. Several updated versions followed; the CM-5 supercomputer is a massively parallel processing computer capable of many billions of arithmetic operations per second.{{cite web|title=ComputerGK.com : Supercomputers |url=http://www.computergk.com/computers/supercomputers/ |date=3 October 2014 |author=Steve Nelson}}

In 1982, Osaka University's LINKS-1 Computer Graphics System used a massively parallel processing architecture, with 514 microprocessors, including 257 Zilog Z8001 control processors and 257 iAPX 86/20 floating-point processors. It was mainly used for rendering realistic 3D computer graphics.{{cite web|url=http://museum.ipsj.or.jp/en/computer/other/0013.html|title=LINKS-1 Computer Graphics System-Computer Museum|website=museum.ipsj.or.jp}} Fujitsu's VPP500 from 1992 is unusual since, to achieve higher speeds, its processors used GaAs, a material normally reserved for microwave applications due to its toxicity.{{Cite web | url=https://www.fujitsu.com/global/about/corporate/history/products/computer/supercomputer/vpp500.html |title = VPP500 (1992) - Fujitsu Global}} Fujitsu's Numerical Wind Tunnel supercomputer used 166 vector processors to gain the top spot in 1994 with a peak speed of 1.7 gigaFLOPS (GFLOPS) per processor.{{cite web|url=http://www.netlib.org/benchmark/top500/reports/report94/main.html |title=TOP500 Annual Report 1994 |publisher=Netlib.org |date=1 October 1996 |access-date=9 June 2012}}{{Cite conference

|author1=N. Hirose |author2=M. Fukuda

|title=Proceedings High Performance Computing on the Information Superhighway. HPC Asia '97

|name-list-style=amp

|year=1997

|chapter=Numerical Wind Tunnel (NWT) and CFD Research at National Aerospace Laboratory

|pages=99–103

|conference=Proceedings of HPC-Asia '97

|publisher=IEEE Computer SocietyPages

|doi=10.1109/HPC.1997.592130

|isbn=0-8186-7901-8

}} The Hitachi SR2201 obtained a peak performance of 600 GFLOPS in 1996 by using 2048 processors connected via a fast three-dimensional crossbar network.H. Fujii, Y. Yasuda, H. Akashi, Y. Inagami, M. Koga, O. Ishihara, M. Syazwan, H. Wada, T. Sumimoto, [http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.86.5625&rep=rep1&type=pdf Architecture and performance of the Hitachi SR2201 massively parallel processor system], Proceedings of 11th International Parallel Processing Symposium, April 1997, pages 233–241.Y. Iwasaki, The CP-PACS project, Nuclear Physics B: Proceedings Supplements, Volume 60, Issues 1–2, January 1998, pages 246–254.A.J. van der Steen, [https://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.79.7986&rep=rep1&type=pdf Overview of recent supercomputers], Publication of the NCF, Stichting Nationale Computer Faciliteiten, the Netherlands, January 1997. The Intel Paragon could have 1000 to 4000 Intel i860 processors in various configurations and was ranked the fastest in the world in 1993. The Paragon was a MIMD machine which connected processors via a high speed two-dimensional mesh, allowing processes to execute on separate nodes, communicating via the Message Passing Interface.Scalable input/output: achieving system balance by Daniel A. Reed 2003 {{ISBN|978-0-262-68142-1}} page 182

Software development remained a problem, but the CM series sparked off considerable research into this issue. Similar designs using custom hardware were made by many companies, including the Evans & Sutherland ES-1, MasPar, nCUBE, Intel iPSC and the Goodyear MPP. But by the mid-1990s, general-purpose CPU performance had improved so much in that a supercomputer could be built using them as the individual processing units, instead of using custom chips. By the turn of the 21st century, designs featuring tens of thousands of commodity CPUs were the norm, with later machines adding graphic units to the mix.

In 1998, David Bader developed the first Linux supercomputer using commodity parts.{{cite web| url= https://www.computer.org/press-room/2021-news/david-bader-to-receive-2021-ieee-cs-sidney-fernbach-award | title=David Bader Selected to Receive the 2021 IEEE Computer Society Sidney Fernbach Award|publisher=IEEE Computer Society|date=September 22, 2021 |accessdate= 2023-10-12}} While at the University of New Mexico, Bader sought to build a supercomputer running Linux using consumer off-the-shelf parts and a high-speed low-latency interconnection network. The prototype utilized an Alta Technologies "AltaCluster" of eight dual, 333 MHz, Intel Pentium II computers running a modified Linux kernel. Bader ported a significant amount of software to provide Linux support for necessary components as well as code from members of the National Computational Science Alliance (NCSA) to ensure interoperability, as none of it had been run on Linux previously.{{cite journal|last=Bader|first=David A.|journal=IEEE Annals of the History of Computing|title=Linux and Supercomputing: How My Passion for Building COTS Systems Led to an HPC Revolution|date=2021|volume=43|issue=3|pages=73–80|doi=10.1109/MAHC.2021.3101415|s2cid=237318907 |doi-access=free}} Using the successful prototype design, he led the development of "RoadRunner," the first Linux supercomputer for open use by the national science and engineering community via the National Science Foundation's National Technology Grid. RoadRunner was put into production use in April 1999. At the time of its deployment, it was considered one of the 100 fastest supercomputers in the world.{{cite news|last=Fleck|first=John|title=UNM to crank up $400,000 supercomputer today|newspaper=Albuquerque Journal|date=April 8, 1999|page=D1}} Though Linux-based clusters using consumer-grade parts, such as Beowulf, existed prior to the development of Bader's prototype and RoadRunner, they lacked the scalability, bandwidth, and parallel computing capabilities to be considered "true" supercomputers.

File:Processor families in TOP500 supercomputers.svg]]

File:2x2x2torus.svg used by systems such as Blue Gene, Cray XT3, etc.]]

Systems with a massive number of processors generally take one of two paths. In the grid computing approach, the processing power of many computers, organized as distributed, diverse administrative domains, is opportunistically used whenever a computer is available.{{cite book |title=Grid computing: experiment management, tool integration, and scientific workflows |url=https://archive.org/details/gridcomputingexp00prod |url-access=limited |first1=Radu |last1=Prodan |first2=Thomas |last2=Fahringer |year=2007 |isbn=978-3-540-69261-4 |pages=[https://archive.org/details/gridcomputingexp00prod/page/n17 1]–4 |publisher=Springer }} In another approach, many processors are used in proximity to each other, e.g. in a computer cluster. In such a centralized massively parallel system the speed and flexibility of the {{vanchor|interconnect}} becomes very important and modern supercomputers have used various approaches ranging from enhanced Infiniband systems to three-dimensional torus interconnects.Knight, Will: "[https://www.newscientist.com/article/dn12145-ibm-creates-worlds-most-powerful-computer/ IBM creates world's most powerful computer]", NewScientist.com news service, June 2007{{cite web |author=N. R. Agida|year=2005 |title=Blue Gene/L Torus Interconnection Network {{pipe}} IBM Journal of Research and Development | volume= 45, No 2/3 March–May 2005 |page= 265 |url=http://www.cc.gatech.edu/classes/AY2008/cs8803hpc_spring/papers/bgLtorusnetwork.pdf |work=Torus Interconnection Network|display-authors=etal|archive-url=https://web.archive.org/web/20110815102821/http://www.cc.gatech.edu/classes/AY2008/cs8803hpc_spring/papers/bgLtorusnetwork.pdf|archive-date=15 August 2011}} The use of multi-core processors combined with centralization is an emerging direction, e.g. as in the Cyclops64 system.{{Cite book | chapter-url=https://link.springer.com/content/pdf/10.1007/11577188_18.pdf |archive-url=https://ghostarchive.org/archive/20221009/https://link.springer.com/content/pdf/10.1007/11577188_18.pdf |archive-date=2022-10-09 |url-status=live |doi = 10.1007/11577188_18|isbn = 978-3-540-29810-6|chapter = Performance Modelling and Optimization of Memory Access on Cellular Computer Architecture Cyclops64|title = Network and Parallel Computing|series = Lecture Notes in Computer Science|year = 2005|last1 = Niu|first1 = Yanwei|last2 = Hu|first2 = Ziang|last3 = Barner|first3 = Kenneth|author3-link = Kenneth Barner|last4 = Gao|first4 = Guang R.|volume = 3779|pages = 132–143}}Analysis and performance results of computing betweenness centrality on IBM Cyclops64 by Guangming Tan, Vugranam C. Sreedhar and Guang R. Gao The Journal of Supercomputing Volume 56, Number 1, 1–24 September 2011

As the price, performance and energy efficiency of general-purpose graphics processing units (GPGPUs) have improved, a number of petaFLOPS supercomputers such as Tianhe-I and Nebulae have started to rely on them.{{cite web |last=Prickett |first=Timothy |title=Top 500 supers – The Dawning of the GPUs |publisher=Theregister.co.uk |date=31 May 2010 |url=https://www.theregister.co.uk/2010/05/31/top_500_supers_jun2010/ }} However, other systems such as the K computer continue to use conventional processors such as SPARC-based designs and the overall applicability of GPGPUs in general-purpose high-performance computing applications has been the subject of debate, in that while a GPGPU may be tuned to score well on specific benchmarks, its overall applicability to everyday algorithms may be limited unless significant effort is spent to tune the application to it.{{cite book |chapter=Considering GPGPU for HPC Centers: Is It Worth the Effort? |author1=Hans Hacker|author2=Carsten Trinitis|author3=Josef Weidendorfer|author4=Matthias Brehm|title=Facing the Multicore-Challenge: Aspects of New Paradigms and Technologies in Parallel Computing|editor1=Rainer Keller|editor2=David Kramer|editor3=Jan-Philipp Weiss |year=2010 |isbn= 978-3-642-16232-9 |pages= 118–121 |chapter-url=https://books.google.com/books?id=-luqXPiew_UC&pg=PA118|publisher=Springer Science & Business Media}} However, GPUs are gaining ground, and in 2012 the Jaguar supercomputer was transformed into Titan by retrofitting CPUs with GPUs.{{cite web |title=Cray's Titan Supercomputer for ORNL Could Be World's Fastest |author=Damon Poeter |publisher=Pcmag.com |date=11 October 2011 |url=https://www.pcmag.com/article2/0,2817,2394515,00.asp}}{{cite web |title=GPUs Will Morph ORNL's Jaguar into 20-Petaflop Titan |first= Michael |last=Feldman |publisher=Hpcwire.com |date=11 October 2011 |url=http://www.hpcwire.com/hpcwire/2011-10-11/gpus_will_morph_ornl_s_jaguar_into_20-petaflop_titan.html}}{{cite web |title=Oak Ridge changes Jaguar's spots from CPUs to GPUs |author= Timothy Prickett Morgan |publisher=Theregister.co.uk |date= 11 October 2011 |url=https://www.theregister.co.uk/2011/10/11/oak_ridge_cray_nvidia_titan/}}

High-performance computers have an expected life cycle of about three years before requiring an upgrade.[http://www.netl.doe.gov/File%20Library/Research/onsite%20research/R-D190-2014Nov.pdf "The NETL SuperComputer"] {{Webarchive|url=https://web.archive.org/web/20150904034017/http://www.netl.doe.gov/File%20Library/Research/onsite%20research/R-D190-2014Nov.pdf |date=4 September 2015 }}.

page 2. The Gyoukou supercomputer is unique in that it uses both a massively parallel design and liquid immersion cooling.

A number of special-purpose systems have been designed, dedicated to a single problem. This allows the use of specially programmed FPGA chips or even custom ASICs, allowing better price/performance ratios by sacrificing generality. Examples of special-purpose supercomputers include Belle,Condon, J.H. and K.Thompson, "[https://www.sciencedirect.com/science/article/pii/B9780080268989500073 Belle Chess Hardware]", In Advances in Computer Chess 3 (ed.M.R.B.Clarke), Pergamon Press, 1982. Deep Blue,{{Cite book

|last=Hsu|first=Feng-hsiung|author-link=Feng-hsiung Hsu

|year=2002

|title=Behind Deep Blue: Building the Computer that Defeated the World Chess Champion

|publisher=Princeton University Press

|isbn=978-0-691-09065-8}} and HydraC. Donninger, U. Lorenz. [https://doi.org/10.1007%2F978-3-540-30117-2_101 The Chess Monster Hydra.] Proc. of 14th International Conference on Field-Programmable Logic and Applications (FPL), 2004, Antwerp – Belgium, LNCS 3203, pp. 927 – 932 for playing chess, Gravity Pipe for astrophysics,J Makino and M. Taiji, Scientific Simulations with Special Purpose Computers: The GRAPE Systems, Wiley. 1998. MDGRAPE-3 for protein structure prediction and molecular dynamics,RIKEN press release, [http://www.riken.jp/engn/r-world/info/release/press/2006/060619/index.html Completion of a one-petaFLOPS computer system for simulation of molecular dynamics] {{Webarchive|url=https://web.archive.org/web/20121202053547/http://www.riken.jp/engn/r-world/info/release/press/2006/060619/index.html |date=2 December 2012 }} and Deep Crack for breaking the DES cipher.{{cite book|author=Electronic Frontier Foundation|url=https://archive.org/details/crackingdes00elec|title=Cracking DES – Secrets of Encryption Research, Wiretap Politics & Chip Design|publisher=Oreilly & Associates Inc|year=1998|isbn=978-1-56592-520-5}}

{{See also|Computer cooling|Green500}}

File:Summit (supercomputer).jpg supercomputer was as of November 2018 the fastest supercomputer in the world.{{cite news|url=https://www.nytimes.com/2018/06/08/technology/supercomputer-china-us.html|title=Move Over, China: U.S. Is Again Home to World's Speediest Supercomputer|last=Lohr|first=Steve|date=8 June 2018|newspaper=New York Times|access-date=19 July 2018}} With a measured power efficiency of 14.668 GFlops/watt it is also the third most energy efficient in the world.{{cite web|url=https://www.top500.org/green500/lists/2018/11/|title=Green500 List - November 2018|website=TOP500|language=en|access-date=19 July 2018}}]]

Throughout the decades, the management of heat density has remained a key issue for most centralized supercomputers.{{cite journal |title=The TianHe-1A Supercomputer: Its Hardware and Software |author= Xue-June Yang |author2=Xiang-Ke Liao |display-authors=etal |journal=Journal of Computer Science and Technology | volume= 26 |issue= 3 |pages= 344–351 |doi= 10.1007/s02011-011-1137-8 |year= 2011 |s2cid= 1389468 }}The Supermen: Story of Seymour Cray and the Technical Wizards Behind the Supercomputer by Charles J. Murray 1997, {{ISBN|0-471-04885-2}}, pages 133–135Parallel Computational Fluid Dyynamics; Recent Advances and Future Directions edited by Rupak Biswas 2010 {{ISBN|1-60595-022-X}} page 401 The large amount of heat generated by a system may also have other effects, e.g. reducing the lifetime of other system components.Supercomputing Research Advances by Yongge Huáng 2008, {{ISBN|1-60456-186-6}}, pages 313–314 There have been diverse approaches to heat management, from pumping Fluorinert through the system, to a hybrid liquid-air cooling system or air cooling with normal air conditioning temperatures.Parallel computing for real-time signal processing and control by M. O. Tokhi, Mohammad Alamgir Hossain 2003, {{ISBN|978-1-85233-599-1}}, pages 201–202 A typical supercomputer consumes large amounts of electrical power, almost all of which is converted into heat, requiring cooling. For example, Tianhe-1A consumes 4.04 megawatts (MW) of electricity.{{cite press release

| url=http://pressroom.nvidia.com/easyir/customrel.do?easyirid=A0D622CE9F579F09&version=live&prid=678988&releasejsp=release_157

| title=NVIDIA Tesla GPUs Power World's Fastest Supercomputer

| publisher=Nvidia

| date=29 October 2010

| access-date=21 February 2011

| archive-date=2 March 2014

| archive-url=https://web.archive.org/web/20140302031237/http://pressroom.nvidia.com/easyir/customrel.do?easyirid=A0D622CE9F579F09&version=live&prid=678988&releasejsp=release_157

| url-status=dead

}} The cost to power and cool the system can be significant, e.g. 4 MW at $0.10/kWh is $400 an hour or about $3.5 million per year.

File:IBM HS20 blade server.jpg blade ]]

Heat management is a major issue in complex electronic devices and affects powerful computer systems in various ways.{{cite web |title=Better Computing Through CPU Cooling |first= Alexander A. |last=Balandin |publisher=IEEE |date= October 2009 |url=https://spectrum.ieee.org/semiconductors/materials/better-computing-through-cpu-cooling/0 |archive-url=https://archive.today/20120714070104/http://spectrum.ieee.org/semiconductors/materials/better-computing-through-cpu-cooling/0 |url-status=dead |archive-date=14 July 2012 }} The thermal design power and CPU power dissipation issues in supercomputing surpass those of traditional computer cooling technologies. The supercomputing awards for green computing reflect this issue.{{cite web | url = http://www.green500.org/ | title = The Green 500 | publisher = Green500.org | access-date = 14 August 2011 | archive-date = 26 August 2016 | archive-url = https://web.archive.org/web/20160826075608/http://www.green500.org/ | url-status = dead }}{{cite web | url = http://www.itnews.com.au/News/65619,green-500-list-ranks-supercomputers.aspx | work = iTnews Australia | title = Green 500 list ranks supercomputers | url-status = dead | archive-url = https://web.archive.org/web/20081022193316/http://www.itnews.com.au/News/65619,green-500-list-ranks-supercomputers.aspx | archive-date = 22 October 2008 | df = dmy-all }}{{cite journal |author=Wu-chun Feng |year=2003 |title=Making a Case for Efficient Supercomputing {{pipe}} ACM Queue Magazine, Volume 1 Issue 7, 10 January 2003 doi 10.1145/957717.957772 |journal=Queue |volume=1 |issue=7 |pages=54 |doi=10.1145/957717.957772 |s2cid=11283177 |doi-access=free }}

The packing of thousands of processors together inevitably generates significant amounts of heat density that need to be dealt with. The Cray-2 was liquid cooled, and used a Fluorinert "cooling waterfall" which was forced through the modules under pressure. However, the submerged liquid cooling approach was not practical for the multi-cabinet systems based on off-the-shelf processors, and in System X a special cooling system that combined air conditioning with liquid cooling was developed in conjunction with the Liebert company.Computational science – ICCS 2005: 5th international conference edited by Vaidy S. Sunderam 2005, {{ISBN|3-540-26043-9}}, pages 60–67

In the Blue Gene system, IBM deliberately used low power processors to deal with heat density.{{cite web

|title=IBM uncloaks 20 petaflops BlueGene/Q super

|website=The Register

|date=22 November 2010

|url=https://www.theregister.co.uk/2010/11/22/ibm_blue_gene_q_super/

|access-date=25 November 2010

}} The IBM Power 775, released in 2011, has closely packed elements that require water cooling.{{cite web|last=Prickett |first=Timothy |url=https://www.theregister.co.uk/2011/07/15/power_775_super_pricing/ |title=The Register: IBM 'Blue Waters' super node washes ashore in August |publisher=Theregister.co.uk |date=15 July 2011 |access-date=9 June 2012}} The IBM Aquasar system uses hot water cooling to achieve energy efficiency, the water being used to heat buildings as well.{{cite web |url=https://www-03.ibm.com/press/us/en/pressrelease/32049.wss |title=IBM Hot Water-Cooled Supercomputer Goes Live at ETH Zurich |website=IBM News room |date=2 July 2010 |url-status=dead |archive-url=https://web.archive.org/web/20110110032000/https://www-03.ibm.com/press/us/en/pressrelease/32049.wss |archive-date=10 January 2011 |access-date=16 March 2020 }}{{cite web |author=Martin LaMonica |url=http://news.cnet.com/8301-11128_3-20004543-54.html |title=CNet 10 May 2010 |publisher=News.cnet.com |date=10 May 2010 |access-date=9 June 2012 |archive-date=1 November 2013 |archive-url=https://web.archive.org/web/20131101060256/http://news.cnet.com/8301-11128_3-20004543-54.html |url-status=dead }}

The energy efficiency of computer systems is generally measured in terms of "FLOPS per watt". In 2008, Roadrunner by IBM operated at 376 MFLOPS/W.{{cite news | url = http://www.cnn.com/2008/TECH/06/09/fastest.computer.ap/index.html | work = CNN | title = Government unveils world's fastest computer |quote= performing 376 million calculations for every watt of electricity used. |archive-url = https://web.archive.org/web/20080610155646/http://www.cnn.com/2008/TECH/06/09/fastest.computer.ap/index.html |archive-date = 10 June 2008}}{{cite web|url = http://www.hpcwire.com/topic/processors/IBM_Roadrunner_Takes_the_Gold_in_the_Petaflop_Race.html|title = IBM Roadrunner Takes the Gold in the Petaflop Race|url-status = live|archive-url = https://web.archive.org/web/20081217131938/http://www.hpcwire.com/topic/processors/IBM_Roadrunner_Takes_the_Gold_in_the_Petaflop_Race.html|archive-date = 17 December 2008|access-date=16 March 2020|df = dmy-all}}{{cbignore|bot=InternetArchiveBot}} In November 2010, the Blue Gene/Q reached 1,684 MFLOPS/W{{cite web| url=http://www.serverwatch.com/hreviews/article.php/3913536/Top500-Supercomputing-List-Reveals-Computing-Trends.htm| title = Top500 Supercomputing List Reveals Computing Trends| date = 20 July 2010|quote=IBM... BlueGene/Q system .. setting a record in power efficiency with a value of 1,680 MFLOPS/W, more than twice that of the next best system.}}{{cite web| url=http://www.datacenterknowledge.com/archives/2010/11/18/ibm-system-clear-winner-in-green-500/|title = IBM Research A Clear Winner in Green 500|date = 18 November 2010}} and in June 2011 the top two spots on the Green 500 list were occupied by Blue Gene machines in New York (one achieving 2097 MFLOPS/W) with the DEGIMA cluster in Nagasaki placing third with 1375 MFLOPS/W.{{cite web |url=http://www.green500.org/lists/2011/06/top/list.php |archive-url=https://web.archive.org/web/20110703094255/http://www.green500.org/lists/2011/06/top/list.php |url-status=live |archive-date=3 July 2011 |title=Green 500 list |publisher=Green500.org |access-date=16 March 2020 }}{{cbignore|bot=InternetArchiveBot}}

Because copper wires can transfer energy into a supercomputer with much higher power densities than forced air or circulating refrigerants can remove waste heat,

Saed G. Younis.

[http://hdl.handle.net/1721.1/7058 "Asymptotically Zero Energy Computing Using Split-Level Charge Recovery Logic"].

1994.

page 14.

the ability of the cooling systems to remove waste heat is a limiting factor.

[http://www.pnl.gov/computing/resources/esdc/1_Cooling.stm "Hot Topic – the Problem of Cooling Supercomputers"] {{webarchive|url=https://web.archive.org/web/20150118222233/http://www.pnl.gov/computing/resources/esdc/1_Cooling.stm |date=18 January 2015 }}.

Anand Lal Shimpi.

[http://www.anandtech.com/show/6421/inside-the-titan-supercomputer-299k-amd-x86-cores-and-186k-nvidia-gpu-cores "Inside the Titan Supercomputer: 299K AMD x86 Cores and 18.6K NVIDIA GPUs"].

2012.

{{As of|2015}}, many existing supercomputers have more infrastructure capacity than the actual peak demand of the machine{{snd}} designers generally conservatively design the power and cooling infrastructure to handle more than the theoretical peak electrical power consumed by the supercomputer. Designs for future supercomputers are power-limited{{snd}} the thermal design power of the supercomputer as a whole, the amount that the power and cooling infrastructure can handle, is somewhat more than the expected normal power consumption, but less than the theoretical peak power consumption of the electronic hardware.

Curtis Storlie; Joe Sexton; Scott Pakin; Michael Lang; Brian Reich; William Rust.

[https://arxiv.org/abs/1412.5247 "Modeling and Predicting Power Consumption of High-Performance Computing Jobs"].

2014.

{{Main|Supercomputer operating systems}}

Since the end of the 20th century, supercomputer operating systems have undergone major transformations, based on the changes in supercomputer architecture.Encyclopedia of Parallel Computing by David Padua 2011 {{ISBN|0-387-09765-1}} pages 426–429 While early operating systems were custom tailored to each supercomputer to gain speed, the trend has been to move away from in-house operating systems to the adaptation of generic software such as Linux.Knowing machines: essays on technical change by Donald MacKenzie 1998 {{ISBN|0-262-63188-1}} page 149-151

Since modern massively parallel supercomputers typically separate computations from other services by using multiple types of nodes, they usually run different operating systems on different nodes, e.g. using a small and efficient lightweight kernel such as CNK or CNL on compute nodes, but a larger system such as a full Linux distribution on server and I/O nodes.Euro-Par 2004 Parallel Processing: 10th International Euro-Par Conference 2004, by Marco Danelutto, Marco Vanneschi and Domenico Laforenza, {{ISBN|3-540-22924-8}}, page 835Euro-Par 2006 Parallel Processing: 12th International Euro-Par Conference, 2006, by Wolfgang E. Nagel, Wolfgang V. Walter and Wolfgang Lehner {{ISBN|3-540-37783-2}} page[https://web.archive.org/web/20190801201606/https://pdfs.semanticscholar.org/2aeb/c9b51047d5b79462f47d89f30f0f90389280.pdf An Evaluation of the Oak Ridge National Laboratory Cray XT3] by Sadaf R. Alam etal International Journal of High Performance Computing Applications February 2008 vol. 22 no. 1 52–80

While in a traditional multi-user computer system job scheduling is, in effect, a tasking problem for processing and peripheral resources, in a massively parallel system, the job management system needs to manage the allocation of both computational and communication resources, as well as gracefully deal with inevitable hardware failures when tens of thousands of processors are present.Open Job Management Architecture for the Blue Gene/L Supercomputer by Yariv Aridor et al. in Job scheduling strategies for parallel processing by Dror G. Feitelson 2005 {{ISBN|978-3-540-31024-2}} pages 95–101

Although most modern supercomputers use Linux-based operating systems, each manufacturer has its own specific Linux distribution, and no industry standard exists, partly due to the fact that the differences in hardware architectures require changes to optimize the operating system to each hardware design.{{cite web |url=http://www.top500.org/overtime/list/32/os |title=Top500 OS chart |publisher=Top500.org |access-date=31 October 2010 |url-status=dead |archive-url=https://web.archive.org/web/20120305234455/http://www.top500.org/overtime/list/32/os |archive-date=5 March 2012 }}

{{Main|Message passing in computer clusters}}

{{See also|Parallel computing|Parallel programming model}}

File:Wide-angle view of the ALMA correlator.jpg correlator{{cite news|title=Wide-angle view of the ALMA correlator|url=http://www.eso.org/public/images/eso1253a/|access-date=13 February 2013|newspaper=ESO Press Release}}]]

The parallel architectures of supercomputers often dictate the use of special programming techniques to exploit their speed. Software tools for distributed processing include standard APIs such as MPI{{cite book |first=Frank |last=Nielsen | title=Introduction to HPC with MPI for Data Science | year=2016

| publisher=Springer |isbn=978-3-319-21903-5 |pages=185–221}} and PVM, VTL, and open source software such as Beowulf.

In the most common scenario, environments such as PVM and MPI for loosely connected clusters and OpenMP for tightly coordinated shared memory machines are used. Significant effort is required to optimize an algorithm for the interconnect characteristics of the machine it will be run on; the aim is to prevent any of the CPUs from wasting time waiting on data from other nodes. GPGPUs have hundreds of processor cores and are programmed using programming models such as CUDA or OpenCL.

Moreover, it is quite difficult to debug and test parallel programs. Special techniques need to be used for testing and debugging such applications.

{{Main|Grid computing}}

File:ArchitectureCloudLinksSameSite.png system connecting many personal computers over the internet]]

Opportunistic supercomputing is a form of networked grid computing whereby a "super virtual computer" of many loosely coupled volunteer computing machines performs very large computing tasks. Grid computing has been applied to a number of large-scale embarrassingly parallel problems that require supercomputing performance scales. However, basic grid and cloud computing approaches that rely on volunteer computing cannot handle traditional supercomputing tasks such as fluid dynamic simulations.{{Cite web|url=https://www.academia.edu/3991932|title = Chapter 03 Software and System Management|last1 = Rahat|first1 = Nazmul}}

The fastest grid computing system is the volunteer computing project Folding@home (F@h). {{as of|2020|4}}, F@h reported 2.5 exaFLOPS of x86 processing power. Of this, over 100 PFLOPS are contributed by clients running on various GPUs, and the rest from various CPU systems.{{cite web|url=https://stats.foldingathome.org/os|title=Client Statistics by OS|publisher=Stanford University|author=Pande lab|work=Folding@home|access-date=10 April 2020}}

The Berkeley Open Infrastructure for Network Computing (BOINC) platform hosts a number of volunteer computing projects. {{As of|2017|02}}, BOINC recorded a processing power of over 166 petaFLOPS through over 762 thousand active Computers (Hosts) on the network.{{Cite web |url=http://www.boincstats.com/stats/project_graph.php?pr=bo |website=BOINCstats |title=BOINC Combined |publisher=BOINC |access-date=30 October 2016 |postscript=Note this link will give current statistics, not those on the date last accessed. |url-status=dead |archive-url=https://web.archive.org/web/20100919090657/http://boincstats.com/stats/project_graph.php?pr=bo |archive-date=19 September 2010 }}

{{As of|2016|10}}, Great Internet Mersenne Prime Search's (GIMPS) distributed Mersenne Prime search achieved about 0.313 PFLOPS through over 1.3 million computers.{{cite web |url=http://www.mersenne.org/primenet |title=Internet PrimeNet Server Distributed Computing Technology for the Great Internet Mersenne Prime Search |work=GIMPS |access-date=6 June 2011 }} The PrimeNet server has supported GIMPS's grid computing approach, one of the earliest volunteer computing projects, since 1997.

{{Main|Quasi-opportunistic supercomputing}}

Quasi-opportunistic supercomputing is a form of distributed computing whereby the "super virtual computer" of many networked geographically disperse computers performs computing tasks that demand huge processing power.{{cite web|last1=Kravtsov|first1=Valentin |last2=Carmeli |first2=David |last3=Dubitzky |first3=Werner |last4=Orda |first4=Ariel |last5=Schuster |first5=Assaf |author-link5=Assaf Schuster |last6=Yoshpa |first6=Benny|title=Quasi-opportunistic supercomputing in grids, hot topic paper (2007)|url=http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.135.8993|work=IEEE International Symposium on High Performance Distributed Computing|publisher=IEEE|citeseerx=10.1.1.135.8993 |access-date=4 August 2011}} Quasi-opportunistic supercomputing aims to provide a higher quality of service than opportunistic grid computing by achieving more control over the assignment of tasks to distributed resources and the use of intelligence about the availability and reliability of individual systems within the supercomputing network. However, quasi-opportunistic distributed execution of demanding parallel computing software in grids should be achieved through the implementation of grid-wise allocation agreements, co-allocation subsystems, communication topology-aware allocation mechanisms, fault tolerant message passing libraries and data pre-conditioning.

Cloud computing with its recent and rapid expansions and development have grabbed the attention of high-performance computing (HPC) users and developers in recent years. Cloud computing attempts to provide HPC-as-a-service exactly like other forms of services available in the cloud such as software as a service, platform as a service, and infrastructure as a service. HPC users may benefit from the cloud in different angles such as scalability, resources being on-demand, fast, and inexpensive. On the other hand, moving HPC applications have a set of challenges too. Good examples of such challenges are virtualization overhead in the cloud, multi-tenancy of resources, and network latency issues. Much research is currently being done to overcome these challenges and make HPC in the cloud a more realistic possibility.{{Cite book|last1=Jamalian|first1=S.|last2=Rajaei|first2=H.|title=2015 IEEE International Conference on Cloud Engineering |chapter=ASETS: A SDN Empowered Task Scheduling System for HPCaaS on the Cloud |date=1 March 2015|pages=329–334|doi=10.1109/IC2E.2015.56|isbn=978-1-4799-8218-9|s2cid=10974077|url=https://zenodo.org/record/890225}}{{Cite book|last1=Jamalian|first1=S.|last2=Rajaei|first2=H.|title=2015 IEEE 8th International Conference on Cloud Computing |chapter=Data-Intensive HPC Tasks Scheduling with SDN to Enable HPC-as-a-Service |date=1 June 2015|pages=596–603|doi=10.1109/CLOUD.2015.85|isbn=978-1-4673-7287-9|s2cid=10141367|url=https://zenodo.org/record/890223}}{{Cite book|last1=Gupta|first1=A.|last2=Milojicic|first2=D.|title=2011 Sixth Open Cirrus Summit |chapter=Evaluation of HPC Applications on Cloud |date=1 October 2011|pages=22–26|doi=10.1109/OCS.2011.10|isbn=978-0-7695-4650-6|citeseerx=10.1.1.294.3936|s2cid=9405724}}{{Cite book|last1=Kim|first1=H.|last2=el-Khamra|first2=Y.|last3=Jha|first3=S.|last4=Parashar|first4=M.|title=2009 Fifth IEEE International Conference on e-Science |chapter=An Autonomic Approach to Integrated HPC Grid and Cloud Usage |date=1 December 2009|pages=366–373|doi=10.1109/e-Science.2009.58|isbn=978-1-4244-5340-5|citeseerx=10.1.1.455.7000|s2cid=11502126}}

In 2016, Penguin Computing, Parallel Works, R-HPC, Amazon Web Services, Univa, Silicon Graphics International, Rescale, Sabalcore, and Gomput started to offer HPC cloud computing. The Penguin On Demand (POD) cloud is a bare-metal compute model to execute code, but each user is given virtualized login node. POD computing nodes are connected via non-virtualized 10 Gbit/s Ethernet or QDR InfiniBand networks. User connectivity to the POD data center ranges from 50 Mbit/s to 1 Gbit/s.{{cite web|last1=Eadline|first1=Douglas|title=Moving HPC to the Cloud|url=http://www.admin-magazine.com/HPC/Articles/Moving-HPC-to-the-Cloud|website=Admin Magazine|access-date=30 March 2019}} Citing Amazon's EC2 Elastic Compute Cloud, Penguin Computing argues that virtualization of compute nodes is not suitable for HPC. Penguin Computing has also criticized that HPC clouds may have allocated computing nodes to customers that are far apart, causing latency that impairs performance for some HPC applications.{{cite web|last1=Niccolai|first1=James|title=Penguin Puts High-performance Computing in the Cloud|url=http://www.pcworld.com/article/170045/article.html|website=PCWorld|publisher=IDG Consumer & SMB|access-date=6 June 2016|date=11 August 2009}}

Supercomputers generally aim for the maximum in capability computing rather than capacity computing. Capability computing is typically thought of as using the maximum computing power to solve a single large problem in the shortest amount of time. Often a capability system is able to solve a problem of a size or complexity that no other computer can, e.g. a very complex weather simulation application.

Capacity computing, in contrast, is typically thought of as using efficient cost-effective computing power to solve a few somewhat large problems or many small problems.The Potential Impact of High-End Capability Computing on Four Illustrative Fields of Science and Engineering by Committee on the Potential Impact of High-End Computing on Illustrative Fields of Science and Engineering and National Research Council (28 October 2008) {{ISBN|0-309-12485-9}} page 9 Architectures that lend themselves to supporting many users for routine everyday tasks may have a lot of capacity but are not typically considered supercomputers, given that they do not solve a single very complex problem.

{{See also|LINPACK benchmarks|Grid computing#Fastest virtual supercomputers}}

File:Supercomputing-rmax-graph2.svg speed over 60 years]]

In general, the speed of supercomputers is measured and benchmarked in FLOPS (floating-point operations per second), and not in terms of MIPS (million instructions per second), as is the case with general-purpose computers.{{cite book |title=Performance Evaluation, Prediction and Visualization of Parallel Systems |author= Xingfu Wu |year=1999 |isbn= 978-0-7923-8462-5 |pages= 114–117 |url=https://books.google.com/books?id=IJZt5H6R8OIC&pg=PA116|publisher=Springer Science & Business Media}} These measurements are commonly used with an SI prefix such as tera-, combined into the shorthand TFLOPS (10¹² FLOPS, pronounced teraflops), or peta-, combined into the shorthand PFLOPS (10¹⁵ FLOPS, pronounced petaflops.) Petascale supercomputers can process one quadrillion (10¹⁵) (1000 trillion) FLOPS. Exascale is computing performance in the exaFLOPS (EFLOPS) range. An EFLOPS is one quintillion (10¹⁸) FLOPS (one million TFLOPS). However, The performance of a supercomputer can be severely impacted by fluctuation brought on by elements like system load, network traffic, and concurrent processes, as mentioned by Brehm and Bruhwiler (2015).Brehm, M. and Bruhwiler,

D. L. (2015) 'Performance Characteristics of the Plasma Wakefield Acceleration Driven by Proton Bunches'. Journal of Physics: Conference Series

No single number can reflect the overall performance of a computer system, yet the goal of the Linpack benchmark is to approximate how fast the computer solves numerical problems and it is widely used in the industry.{{Citation

|last1 = Dongarra

|first1 = Jack J.

|last2 = Luszczek

|first2 = Piotr

|last3 = Petitet

|first3 = Antoine

|title = The LINPACK Benchmark: past, present and future

|year = 2003

|journal = Concurrency and Computation: Practice and Experience

|volume = 15

|issue = 9

|pages = 803–820

|url = http://www.netlib.org/utk/people/JackDongarra/PAPERS/hplpaper.pdf

|doi = 10.1002/cpe.728

|s2cid = 1900724

}} The FLOPS measurement is either quoted based on the theoretical floating point performance of a processor (derived from manufacturer's processor specifications and shown as "Rpeak" in the TOP500 lists), which is generally unachievable when running real workloads, or the achievable throughput, derived from the LINPACK benchmarks and shown as "Rmax" in the TOP500 list.{{cite web|title=Understanding measures of supercomputer performance and storage system capacity|url=https://kb.iu.edu/d/apeq#measure-flops|publisher=Indiana University|access-date=3 December 2017}} The LINPACK benchmark typically performs LU decomposition of a large matrix.{{cite web|title=Frequently Asked Questions|url=https://www.top500.org/resources/frequently-asked-questions/|website=TOP500.org|access-date=3 December 2017}} The LINPACK performance gives some indication of performance for some real-world problems, but does not necessarily match the processing requirements of many other supercomputer workloads, which for example may require more memory bandwidth, or may require better integer computing performance, or may need a high performance I/O system to achieve high levels of performance.

{{Main|TOP500}}

{{Further|List of fastest computers|History of supercomputing}}

File:Top20supercomputers.png

Since 1993, the fastest supercomputers have been ranked on the TOP500 list according to their LINPACK benchmark results. The list does not claim to be unbiased or definitive, but it is a widely cited current definition of the "fastest" supercomputer available at any given time.

This is a list of the computers which appeared at the top of the TOP500 list since June 1993,{{cite web |author= |url=https://top500.org/lists/top500/ |title=Top500 lists |publisher=Top500.org |access-date=3 August 2024}} and the "Peak speed" is given as the "Rmax" rating. In 2018, Lenovo became the world's largest provider for the TOP500 supercomputers with 117 units produced.{{cite news |url=https://www.businesswire.com/news/home/20180625005341/en/ |work=Business Wire |title=Lenovo Attains Status as Largest Global Provider of TOP500 Supercomputers |date=25 June 2018}}

Legend:{{cite web |title=TOP500 DESCRIPTION |url=https://www.top500.org/project/top500_description/ |website=www.top500.org |access-date=23 June 2020 |archive-date=23 June 2020 |archive-url=https://web.archive.org/web/20200623154459/https://www.top500.org/project/top500_description/ |url-status=live }}

• Rank{{snd}}Position within the TOP500 ranking. In the TOP500 list table, the computers are ordered first by their Rmax value. In the case of equal performances (Rmax value) for different computers, the order is by Rpeak. For sites that have the same computer, the order is by memory size and then alphabetically.
• Rmax{{snd}}The highest score measured using the LINPACK benchmarks suite. This is the number that is used to rank the computers. Measured in quadrillions of 64-bit floating point operations per second, i.e., petaFLOPS.{{cite web |title=FREQUENTLY ASKED QUESTIONS |url=https://www.top500.org/resources/frequently-asked-questions/ |website=www.top500.org |access-date=23 June 2020 |archive-date=3 April 2021 |archive-url=https://web.archive.org/web/20210403083231/https://www.top500.org/resources/frequently-asked-questions/ |url-status=live }}
• Rpeak{{snd}}This is the theoretical peak performance of the system. Computed in petaFLOPS.
• Name{{snd}}Some supercomputers are unique, at least in their location, and are thus named by their owner.
• Model{{snd}}The computing platform as it is marketed.
• Processor{{snd}}The instruction set architecture or processor microarchitecture, alongside GPU and accelerators when available.
• Interconnect{{snd}}The interconnect between computing nodes. InfiniBand is most used (38%) by performance share, while Gigabit Ethernet is most used (54%) by number of computers.
• Manufacturer{{snd}}The manufacturer of the platform and hardware.
• Site{{snd}}The name of the facility operating the supercomputer.
• Country{{snd}}The country in which the computer is located.
• Year{{snd}}The year of installation or last major update.
• Operating system{{snd}}The operating system that the computer uses.

{{Prose|section|date=January 2020}}

{{expand section|date=January 2020}}

The stages of supercomputer application are summarized in the following table:

The IBM Blue Gene/P computer has been used to simulate a number of artificial neurons equivalent to approximately one percent of a human cerebral cortex, containing 1.6 billion neurons with approximately 9 trillion connections. The same research group also succeeded in using a supercomputer to simulate a number of artificial neurons equivalent to the entirety of a rat's brain.Kaku, Michio. Physics of the Future (New York: Doubleday, 2011), 65.

Modern weather forecasting relies on supercomputers. The National Oceanic and Atmospheric Administration uses supercomputers to crunch hundreds of millions of observations to help make weather forecasts more accurate.{{cite web |url=http://news.nationalgeographic.com/news/2005/08/0829_050829_supercomputer.html |archive-url=https://web.archive.org/web/20050905005850/http://news.nationalgeographic.com/news/2005/08/0829_050829_supercomputer.html |url-status=dead |archive-date=5 September 2005 |title=Faster Supercomputers Aiding Weather Forecasts |publisher=News.nationalgeographic.com |date=28 October 2010 |access-date=8 July 2011}}

In 2011, the challenges and difficulties in pushing the envelope in supercomputing were underscored by IBM's abandonment of the Blue Waters petascale project.{{Cite journal |url=http://search.ebscohost.com/login.aspx?direct=true&db=bwh&AN=8OGE.2B33479B.C267DC93&site=ehost-live |title=IBM Drops 'Blue Waters' Supercomputer Project |date=9 August 2011 |journal=International Business Times |access-date=14 December 2018}} {{subscription required|via=[https://www.ebsco.com EBSCO]}}

The Advanced Simulation and Computing Program currently uses supercomputers to maintain and simulate the United States nuclear stockpile.{{cite web|url=https://nnsa.energy.gov/aboutus/ourprograms/defenseprograms/futurescienceandtechnologyprograms/asc/supercomputers|title=Supercomputers|work=U.S. Department of Energy|access-date=7 March 2017|archive-date=7 March 2017|archive-url=https://web.archive.org/web/20170307210251/https://nnsa.energy.gov/aboutus/ourprograms/defenseprograms/futurescienceandtechnologyprograms/asc/supercomputers|url-status=dead}}

In early 2020, COVID-19 was front and center in the world. Supercomputers used different simulations to find compounds that could potentially stop the spread. These computers run for tens of hours using multiple paralleled running CPU's to model different processes.{{Cite web|title=Supercomputer Simulations Help Advance Electrochemical Reaction Research|url=https://ucsdnews.ucsd.edu/pressrelease/supercomputer-simulations-help-advance-electrochemical-reaction-research|website=ucsdnews.ucsd.edu|access-date=12 May 2020}}{{Cite web|title=IBM's Summit—The Supercomputer Fighting Coronavirus|url=http://emag.medicalexpo.com/summit-the-supercomputer-fighting-coronavirus/|date=16 April 2020|website=MedicalExpo e-Magazine|language=en-GB|access-date=12 May 2020}}{{Cite web|title=OSTP Funding Supercomputer Research to Combat COVID-19 – MeriTalk|url=https://www.meritalk.com/articles/ostp-funding-supercomputer-research-to-combat-covid-19/|language=en-US|access-date=12 May 2020}}

File:Taiwania 3.jpg. It was launched in 2020 and has a capacity of about two to three PetaFLOPS.]]

File:Supercomputer Share Top500 November2015.png

In the 2010s, China, the United States, the European Union, and others competed to be the first to create a 1 exaFLOP (10¹⁸ or one quintillion FLOPS) supercomputer.{{Cite news|url=https://www.nextbigfuture.com/2018/02/eu-1-2-supercomputer-project-to-several-10-100-petaflop-computers-by-2020-and-exaflop-by-2022.html|title=EU $1.2 supercomputer project to several 10-100 PetaFLOP computers by 2020 and exaFLOP by 2022 {{!}} NextBigFuture.com|date=4 February 2018|work=NextBigFuture.com|access-date=21 May 2018|language=en-US}} Erik P. DeBenedictis of Sandia National Laboratories has theorized that a zettaFLOPS (10²¹ or one sextillion FLOPS) computer is required to accomplish full weather modeling, which could cover a two-week time span accurately.{{cite web |last1=DeBenedictis |first1=Erik P. |title=The Path To Extreme Computing |url=http://www.zettaflops.org/PES/0-Organization-DeBenedictis.pdf |website=Zettaflops |publisher=Sandia National Laboratories |access-date=9 September 2020 |archive-url=https://web.archive.org/web/20070803175503/http://www.zettaflops.org/PES/0-Organization-DeBenedictis.pdf |archive-date=3 August 2007 |date=2004 |url-status=dead}}{{cite magazine|last1=Cohen|first1=Reuven|title=Global Bitcoin Computing Power Now 256 Times Faster Than Top 500 Supercomputers, Combined!|url=https://www.forbes.com/sites/reuvencohen/2013/11/28/global-bitcoin-computing-power-now-256-times-faster-than-top-500-supercomputers-combined/#660eb2ff6e5e|magazine=Forbes|date=28 November 2013|access-date=1 December 2017}}{{cite book |chapter=Reversible logic for supercomputing |title=Proceedings of the 2nd conference on Computing frontiers |last=DeBenedictis |first=Erik P. |year=2005 |isbn=978-1-59593-019-4 |pages=391–402 |publisher=ACM Press |chapter-url=http://portal.acm.org/citation.cfm?id=1062325 }} Such systems might be built around 2030.{{cite news |title=IDF: Intel says Moore's Law holds until 2029 |url=http://www.h-online.com/newsticker/news/item/IDF-Intel-says-Moore-s-Law-holds-until-2029-734779.html |work=Heise Online |date=4 April 2008 |url-status=dead |archive-url=https://web.archive.org/web/20131208075357/http://www.h-online.com/newsticker/news/item/IDF-Intel-says-Moore-s-Law-holds-until-2029-734779.html |archive-date=8 December 2013 }}

Many Monte Carlo simulations use the same algorithm to process a randomly generated data set; particularly, integro-differential equations describing physical transport processes, the random paths, collisions, and energy and momentum depositions of neutrons, photons, ions, electrons, etc. {{Anchor|dimension2016-01-29}}The next step for microprocessors may be into the third dimension; and specializing to Monte Carlo, the many layers could be identical, simplifying the design and manufacture process.{{cite book|last=Solem|first=J. C.|title=Monte-Carlo Methods and Applications in Neutronics, Photonics and Statistical Physics |chapter=MECA: A multiprocessor concept specialized to Monte Carlo |year=1985|publisher=Proceedings of the Joint los Alamos National Laboratory – Commissariat à l'Energie Atomique Meeting Held at Cadarache Castle, Provence, France 22–26 April 1985; Monte-Carlo Methods and Applications in Neutronics, Photonics and Statistical Physics, Alcouffe, R.; Dautray, R.; Forster, A.; Forster, G.; Mercier, B.; Eds. (Springer Verlag, Berlin) |volume=240|pages=184–195|doi=10.1007/BFb0049047|series=Lecture Notes in Physics|bibcode=1985LNP...240..184S|osti=5689714 |isbn=978-3-540-16070-0|chapter-url=https://digital.library.unt.edu/ark:/67531/metadc1089522/}}

The cost of operating high performance supercomputers has risen, mainly due to increasing power consumption. In the mid-1990s a top 10 supercomputer required in the range of 100 kilowatts, in 2010 the top 10 supercomputers required between 1 and 2 megawatts.{{cite book |title=Recent Advances in the Message Passing Interface: 18th European MPI Users' Group Meeting, EuroMPI 2011, Santorini, Greece, September 18-21, 2011. Proceedings |author1=Yiannis Cotronis |author2=Anthony Danalis |author3=Dimitris Nikolopoulos |author4=Jack Dongarra |publisher= Springer Science & Business Media|isbn=9783642244483 |year=2011 }} A 2010 study commissioned by DARPA identified power consumption as the most pervasive challenge in achieving Exascale computing.{{cite book |title=Energy-Efficient High Performance Computing: Measurement and Tuning |url=https://archive.org/details/energyefficienth00iiij |url-access=limited |author1=James H. Laros III |author2=Kevin Pedretti |author3=Suzanne M. Kelly |author4=Wei Shu |author5=Kurt Ferreira |author6=John Van Dyke |author7=Courtenay Vaughan |publisher= Springer Science & Business Media|isbn=9781447144922 |year=2012| page=[https://archive.org/details/energyefficienth00iiij/page/n9 1]}} At the time a megawatt per year in energy consumption cost about 1 million dollars. Supercomputing facilities were constructed to efficiently remove the increasing amount of heat produced by modern multi-core central processing units. Based on the energy consumption of the Green 500 list of supercomputers between 2007 and 2011, a supercomputer with 1 exaFLOPS in 2011 would have required nearly 500 megawatts. Operating systems were developed for existing hardware to conserve energy whenever possible.{{cite book |title=Energy-Efficient High Performance Computing: Measurement and Tuning |url=https://archive.org/details/energyefficienth00iiij |url-access=limited |author1=James H. Laros III |author2=Kevin Pedretti |author3=Suzanne M. Kelly |author4=Wei Shu |author5=Kurt Ferreira |author6=John Van Dyke |author7=Courtenay Vaughan |publisher= Springer Science & Business Media |isbn=9781447144922 |year=2012| page=[https://archive.org/details/energyefficienth00iiij/page/n12 2]}} CPU cores not in use during the execution of a parallelized application were put into low-power states, producing energy savings for some supercomputing applications.{{cite book |title=Energy-Efficient High Performance Computing: Measurement and Tuning |url=https://archive.org/details/energyefficienth00iiij |url-access=limited |author1=James H. Laros III |author2=Kevin Pedretti |author3=Suzanne M. Kelly |author4=Wei Shu |author5=Kurt Ferreira |author6=John Van Dyke |author7=Courtenay Vaughan |publisher= Springer Science & Business Media|isbn=9781447144922 |year=2012| page=[https://archive.org/details/energyefficienth00iiij/page/n13 3]}}

The increasing cost of operating supercomputers has been a driving factor in a trend toward bundling of resources through a distributed supercomputer infrastructure. National supercomputing centers first emerged in the US, followed by Germany and Japan. The European Union launched the Partnership for Advanced Computing in Europe (PRACE) with the aim of creating a persistent pan-European supercomputer infrastructure with services to support scientists across the European Union in porting, scaling and optimizing supercomputing applications. Iceland built the world's first zero-emission supercomputer. Located at the Thor Data Center in Reykjavík, Iceland, this supercomputer relies on completely renewable sources for its power rather than fossil fuels. The colder climate also reduces the need for active cooling, making it one of the greenest facilities in the world of computers.{{Cite web|url = http://www.intelfreepress.com/news/green-supercomputer-crunches-big-data-in-iceland/39/|title = Green Supercomputer Crunches Big Data in Iceland|date = 21 May 2015|access-date = 18 May 2015|website = intelfreepress.com|url-status = dead|archive-url = https://web.archive.org/web/20150520034755/http://www.intelfreepress.com/news/green-supercomputer-crunches-big-data-in-iceland/39/|archive-date = 20 May 2015|df = dmy-all}}

Funding supercomputer hardware also became increasingly difficult. In the mid-1990s a top 10 supercomputer cost about 10 million euros, while in 2010 the top 10 supercomputers required an investment of between 40 and 50 million euros. In the 2000s national governments put in place different strategies to fund supercomputers. In the UK the national government funded supercomputers entirely and high performance computing was put under the control of a national funding agency. Germany developed a mixed funding model, pooling local state funding and federal funding.

{{Main|AI takeover}}

Examples of supercomputers in fiction include HAL 9000, Multivac, The Machine Stops, GLaDOS, The Evitable Conflict, Vulcan's Hammer, Colossus, WOPR, AM, and Deep Thought. A supercomputer from Thinking Machines was mentioned as the supercomputer used to sequence the DNA extracted from preserved parasites in the Jurassic Park series.

{{cols|colwidth=21em}}

• ACM/IEEE Supercomputing Conference
• ACM SIGHPC
• High-performance computing
• High-performance technical computing
• Jungle computing
• Metacomputing
• Nvidia Tesla Personal Supercomputer
• Parallel computing
• Supercomputing in China
• Supercomputing in Europe
• Supercomputing in India
• Supercomputing in Japan
• SLURM
• Testing high-performance computing applications
• Ultra Network Technologies
• Quantum computing

{{colend}}

{{reflist}}

{{Commons category|Supercomputers}}

• McDonnell, Marshall T. (2013). [https://trace.tennessee.edu/utk_chembiopubs/93/ "Supercomputer Design: An Initial Effort to Capture the Environmental, Economic, and Societal Impacts"]. Chemical and Biomolecular Engineering Publications and Other Works.

{{Parallel computing}}

{{Computer sizes}}

{{Computer science}}

{{Portal bar|Electronics|Illinois|Linux|History of science|Spaceflight|Computer programming|United States}}

{{Authority control}}

Category:American inventions

Category:Cluster computing

Category:Concurrent computing

Category:Distributed computing architecture

Category:Parallel computing