BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:DataFlow SuperComputing for BigData - Veljko Milutinovic
DTSTART:20170928T140000Z
DTEND:20170928T150000Z
UID:TALK72573@talks.cam.ac.uk
CONTACT:Liang Wang
DESCRIPTION:This presentation analyses the essence of DataFlow SuperComput
 ing\, defines\nits advantages and sheds light on the related programming m
 odel. DataFlow\ncomputers\, compared to ControlFlow computers\, offer spee
 dups of 20 to 200\n(even 2000 for some applications)\, power reductions of
  about 20\, and size\nreductions of also about 20. However\, the programmi
 ng paradigm is\ndifferent\, and has to be mastered. The talk explains the 
 paradigm\, using\nMaxeler as an example\, and sheds light on the ongoing r
 esearch in the\nfield. Examples include: Cryptography and Security\, Tradi
 ng and Finances\,\nCreditDerivatives and numerous related BankingApplicati
 ons\,\nSignalProcessing\, GeoPhysics\, WeatherForecast\, OilGas\, DataEngi
 neering\,\nDataMining\, SmartGrid\, ScientificSimulations\, BrainResearch\
 , Genomics\,\netc.\n\nAlso\, a recent study from Tsinghua University in Ch
 ina is presented\, which\nreveals that\, for Shallow Water Weather Forecas
 t (a BigData problem)\, on\nthe 1U level\, the Maxeler DataFlow machine is
  14 times faster than the\nTianhe machine\, rated #1 on the Top 500 list (
 based on Linpack\, which is a\nsmalldata benchmark). Given enough time\, t
 he talk also gives a tutorial\nabout the programming in space\, which is t
 he programming paradigm used for\nthe Maxeler dataflow machines (establish
 ed in 2014 by Stanford\, Imperial\,\nTsinghua\, and the University of Toky
 o). The talk concludes with selected\nexamples and a tool overview (<a hre
 f="http://appgallery.maxeler.com">appgallery.maxeler.com</a> and\n<a href=
 "http://webIDE.maxeler.com">webIDE.maxeler.com</a>). A detailed tutorial o
 n programming in space will be\navailable after the talk. Related hands-on
  activities will be performed by\nremote login (<a href="http://maxeler.mi
 .sanu.ac.rs">maxeler.mi.sanu.ac.rs</a>). Since December 2016\, Maxeler is 
 also\navailable via Amazon AWS. In December 2016\, Hitachi of Japan announ
 ced its\npartnership with Maxeler (also available via Amazon AWS)\, statin
 g that\,\nfor their finance and cryptography applications\, Maxeler is ord
 ers of\nmagnitude faster than any other ControlFlow platform (i.e.\, CPU o
 r GPU).\n\n\nBio: Prof. Veljko Milutinovic (1951) received his PhD from th
 e University of\nBelgrade\, spent about a decade on various faculty positi
 ons in the USA\n(mostly at Purdue University)\, and was a co-designer of t
 he DARPAs first\nGaAs RISC microprocessor. Later\, for almost 3 decades\, 
 he taught and\nconducted research at the University of Belgrade\, in EE\, 
 BA\, MATH\, and\nPHYS /CHEM. Now he serves as the Chairman of the Board fo
 r the Maxeler\noperation in Belgrade\, Serbia. His research is mostly in d
 atamining algorithms and\ndataflow computing\, with the emphasis on mappin
 g of data analytics\nalgorithms onto fast energy efficient architectures. 
 For 7 of his books\,\nforewords were written by 7 different Nobel Laureate
 s with whom he\ncooperated on his past industry sponsored projects. He has
  over 40 IEEE\njournal papers\, over 40 papers in other SCI journals (4 in
  ACM journals)\,\nover 400 Thomson-Reuters citations\, and about 4000 Goog
 le Scholar\ncitations. Short courses on the subject he delivered so far in
  a number of\nuniversities worldwide: MIT\, Harvard\, Boston\, NEU\, Colum
 bia\, NYU\,\nPrinceton\, Temple\, Purdue\, IU\, UIUC\, Michigan\, EPFL\, E
 TH\, Karlsruhe\,\nHeidelberg\, etc. Also at the World Bank in Washington D
 C\, BNL\, IBM TJ\nWatson\, Yahoo\, etc.\n
LOCATION:FW26\, Computer Laboratory\, William Gates Building
END:VEVENT
END:VCALENDAR