BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Dataflow Computing for Data-Intensive Applications
DTSTART;VALUE=DATE-TIME:20120329T090000Z
DTEND;VALUE=DATE-TIME:20120329T103000Z
DTSTAMP;VALUE=DATE-TIME:20130522T214128Z
UID:indico-event-181589@cern.ch
DESCRIPTION:\n	Maxeler Technologies provides Maximum Performance Computing
  (MPC) based on a dataflow model of computation. Combined with a multi-di
 sciplinary approach\, these dataflow solutions are in production\, reduci
 ng power consumption and data center space for a wide range of applicati
 ons. Typically\, our dataflow solutions utilize thousands of arithmetic 
 units on a chip to outperform top end microprocessors by 20x-40x in compu
 tations per cubic foot and computations per Watt\, and our analysis sugg
 ests that there is strong potential to do a lot better in future. Dataflo
 w is particularly suitable for highly demanding applications such as: fin
 ite element and finite difference PDE solvers on structured and unstructu
 red grids\, Monte Carlo methods\, multi-dimensional optimization problems
 \, real-time data processing\, and even sparse matrix solvers. At the co
 re of the approach\, we are building computers to match the problem\, ra
 ther than optimizing algorithms to standard microprocessors.\n\n	Given the
  success in Earth Sciences\, Quantitative Finance and Electronic Trading 
 and the availability of a user-friendly dataflow programming environment
 \, we believe that this novel approach to computing could potentially lea
 d to scientific breakthroughs in data-intensive branches of science.\n	I
 n spite of their spectacular speed\, FPGAs\, GPGPUs and also multicore arc
 hitectures obey to a restrictive computation paradigm that makes less attr
 active the numerical solution of many algebraic problems. Hence the real c
 hallenge for the developer is the reduction of a mathematical model to a s
 equence of computational tasks that perfectly fit the paradigm supported b
 y these extreme architectures. In a second part\, we illustrate this con
 cept for time imaging algorithms of general use in oil industry.\n	 \n\n	
 About the speakers\n\n	Dr. Oskar Mencer\, Maxeler Technologies\n\n	Prior t
 o founding Maxeler\, Oskar was Member of Technical Staff at the Computing
  Sciences Center at Bell Labs in Murray Hill\, leading the effort in "Str
 eam Computing". He joined Bell Labs after receiving a PhD from Stanford U
 niversity. Besides driving Maximum Performance Computing (MPC) at Maxeler
 \, Oskar is Consulting Professor in Geophysics at Stanford University and
  he is also affiliated with the Computing Department at Imperial College 
 London\, having received two Best Paper Awards\, an Imperial College Rese
 arch Excellence Award in 2007 and a Special Award from Com.sult in 2012 f
 or "revolutionising the world of computers".\n\n	Dr. Ernesto Bonomi\, CRS
 4\n\n	Director at CRS4 of three highly-motivated research groups (Imaging 
 and Numerical Geophysics\, Environmental Sciences\, and Process Engineerin
 g and Combustion) which include about 20 research staff\, developing and a
 pplying numerical simulation models and data analysis tools for earth expl
 oration and seismic imaging\, hydrology and territorial planning\, meteoro
 logy\, and clean combustion processes.\n\n	Head of the developers? team pr
 oviding innovative industrial acoustic imaging tools based on original mat
 hematical developments and on efficient data-driven parallel algorithms gi
 ving rise to the implementation of very-large-scale 3D numerical applicati
 ons used routinely in production on HPC platforms and\, more recently\, on
  acceleration hardware (FPGA and GPU).\n\nhttp://indico.cern.ch/conference
 Display.py?confId=181589
LOCATION:CERN Council Chamber
URL:http://indico.cern.ch/conferenceDisplay.py?confId=181589
END:VEVENT
END:VCALENDAR