History of the Supercomputers Vincent Keller (Vincent.Keller@epfl.ch) CADMOS October 23, 2013 Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 1 / 50
Summary Introduction 1 1930 - 1965 : The first computers 2 1965 - 1996 : The “real” supercomputers 3 1996 - Today : Beowulf clusters 4 2000’s : The return of the specialized machines 5 A very special supercomputer 6 Conclusion 7 Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 2 / 50
Introduction Outline Introduction 1 1930 - 1965 : The first computers 2 1965 - 1996 : The “real” supercomputers 3 1996 - Today : Beowulf clusters 4 2000’s : The return of the specialized machines 5 A very special supercomputer 6 Conclusion 7 Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 3 / 50
Introduction From “computer” and “supercomputer” definitions... Faster than human being Electronic machine Digital processing of the information Stored program Execution of instructions Huge computational power Small cycle times Huge resources Parallel operations Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 4 / 50
1930 - 1965 : The first computers Outline Introduction 1 1930 - 1965 : The first computers 2 1965 - 1996 : The “real” supercomputers 3 1996 - Today : Beowulf clusters 4 2000’s : The return of the specialized machines 5 A very special supercomputer 6 Conclusion 7 Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 5 / 50
1930 - 1965 : The first computers 1937 : FASTER THAN HUMAN BEING Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 6 / 50
1930 - 1965 : The first computers Konrad Zuse, Z1 - Z4 1937 Z1 : mecanical machine http://www.horst-zuse.homepage.t-online.de/z1.html Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 7 / 50
1930 - 1965 : The first computers Konrad Zuse, Z1 - Z4 1941 : ELECTRO-MECANICAL MACHINE (relays + mecanics) Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 8 / 50
1930 - 1965 : The first computers Konrad Zuse, Z1 - Z4 1941 Z3 : electro-mechanical http://www.horst-zuse.homepage.t-online.de/z3.html Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 9 / 50
1930 - 1965 : The first computers Harvard MARK 1 1944 FULL ELECTRONIC MACHINE (vacuum tubes) DIGITAL PROCESSING (Military classified machines) Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 10 / 50
1930 - 1965 : The first computers Harvard MARK 1 1944 Harvard MARK 1 (IBM ASCC) : ballistics and warship design http://www.computermuseum.li Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 11 / 50
1930 - 1965 : The first computers Colossus 1942 ENIGMA, Lorentz SZ/42 http://www.cryptomuseum.com Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 12 / 50
1930 - 1965 : The first computers Colossus 1944 Colossus : decypher Nazis codes Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 13 / 50
1930 - 1965 : The first computers ENIAC 1946 ENIAC, full electronic, first Turing-complete http://www.fi.edu/learn/case-files/eckertmauchly/ENIAC_Image_2.jpg Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 14 / 50
1930 - 1965 : The first computers Manchester MARK 1 1948 : STORED PROGRAMS Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 15 / 50
1930 - 1965 : The first computers Manchester MARK 1 1948 Manchester MARK 1 : stored programs Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 16 / 50
1930 - 1965 : The first computers IBM 701 1954 : FIRST COMMERCIAL SUCCESS Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 17 / 50
1930 - 1965 : The first computers IBM 701 1954 IBM 701 : First commercial success http://www.columbia.edu Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 18 / 50
1930 - 1965 : The first computers Experimental Transistor Computer 1955 : FROM VACUUM TUBES TO TRANSISTORS Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 19 / 50
1930 - 1965 : The first computers Experimental Transistor Computer 1953 University of Manchester Experimental Transistor Computer Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 20 / 50
1965 - 1996 : The “real” supercomputers Outline Introduction 1 1930 - 1965 : The first computers 2 1965 - 1996 : The “real” supercomputers 3 1996 - Today : Beowulf clusters 4 2000’s : The return of the specialized machines 5 A very special supercomputer 6 Conclusion 7 Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 21 / 50
1965 - 1996 : The “real” supercomputers CDC 6600 1965 : PARALLEL OPERATIONS Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 22 / 50
1965 - 1996 : The “real” supercomputers CDC 6600 1965 CDC 6600 : First commercial supercomputer http://www.computer-history.info Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 23 / 50
1965 - 1996 : The “real” supercomputers Cray 1 1976 : VECTOR ARCHITECTURE Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 24 / 50
1965 - 1996 : The “real” supercomputers Cray 1 1976 Cray 1 : The revolution Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 25 / 50 http://www.cbi.umn.edu
1965 - 1996 : The “real” supercomputers Cray 1 Cray 1 at EPFL http://www.epfl.ch Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 26 / 50
1965 - 1996 : The “real” supercomputers Cray X-MP 1982 : PARALLEL VECTOR PROCESSORS Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 27 / 50
1965 - 1996 : The “real” supercomputers Cray X-MP 1982 Cray X-MP : Parallel vector CPUs http://www.craysupercomputers.com Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 28 / 50
1965 - 1996 : The “real” supercomputers Cray X-MP Cray X-MP at EPFL (originally from CERN) http://www.epfl.ch Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 29 / 50
1965 - 1996 : The “real” supercomputers Cray 2 1985 Cray 2 : Very large memory (2 GB) http://www.craysupercomputers.com Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 30 / 50
1965 - 1996 : The “real” supercomputers Cray 2 Cray 2 at l’EPFL http://www.epfl.ch Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 31 / 50
1965 - 1996 : The “real” supercomputers Thinking Machines CM-2 and CM-5 1987 : MASSIVELY PARALLEL MACHINES Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 32 / 50
1965 - 1996 : The “real” supercomputers Thinking Machines CM-2 and CM-5 1987 CM-2 : a visionnary idea http://www.new-npac.org Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 33 / 50
1965 - 1996 : The “real” supercomputers Thinking Machines CM-2 and CM-5 1993 : BEGINNING OF TOP500 Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 34 / 50
1965 - 1996 : The “real” supercomputers Thinking Machines CM-2 and CM-5 1993 CM-5 : Very first #1 TOP500 Thinking Machines brochure Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 35 / 50
1965 - 1996 : The “real” supercomputers The (temporary) end mid nineties END OF PROPRIETARY DESIGN. OPEN-SOURCE (real) BIRTH Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 36 / 50
1996 - Today : Beowulf clusters Outline Introduction 1 1930 - 1965 : The first computers 2 1965 - 1996 : The “real” supercomputers 3 1996 - Today : Beowulf clusters 4 2000’s : The return of the specialized machines 5 A very special supercomputer 6 Conclusion 7 Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 37 / 50
1996 - Today : Beowulf clusters NASA Beowulf 1994 NASA Goddard : Thomas Sterling starts waste recovery http://www.nasa.gov Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 38 / 50
1996 - Today : Beowulf clusters EPFL : Bellatrix mid 2000’s : MULTI-CORES NODES Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 39 / 50
1996 - Today : Beowulf clusters EPFL : Bellatrix 2012 Bellatrix : A standard cluster http://hpc.epfl.ch Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 40 / 50
2000’s : The return of the specialized machines Outline Introduction 1 1930 - 1965 : The first computers 2 1965 - 1996 : The “real” supercomputers 3 1996 - Today : Beowulf clusters 4 2000’s : The return of the specialized machines 5 A very special supercomputer 6 Conclusion 7 Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 41 / 50
2000’s : The return of the specialized machines IBM BG/L 2005 : MASSIVELY PARALLEL, MULTI-CORES NODES Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 42 / 50
2000’s : The return of the specialized machines IBM BG/L 2005 IBM BlueGene/L : 16’384 nodes http://hpc.epfl.ch Vincent Keller (CADMOS) CADMOS Retreat, Leysin October 23, 2013 43 / 50
Recommend
More recommend