BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Neural simulation on diverse computational hardware - Dr Dan Goodm
 an\, Imperial
DTSTART:20150209T140000Z
DTEND:20150209T150000Z
UID:TALK57904@talks.cam.ac.uk
CONTACT:Prof Simon Moore
DESCRIPTION:Simulating the activity of the brain requires enormous computa
 tional resources: a recent simulation of a human brain scale neural networ
 k (around 10^11 neurons) for just one second of simulated time took 50 day
 s on a cluster of 27 machines. Considerable effort is being put into innov
 ative applications of diverse\, modern computational hardware to this prob
 lem: general purpose graphics processing units\; the SpiNNaker supercomput
 er specialised for simulating neural networks that will use 1M ARM cores\;
  field programmable gate arrays\; etc. Unfortunately\, in all these cases 
 the use of this cutting edge hardware comes at a very high cost of technic
 al expertise and the majority of neuroscientists are simply not equipped t
 o make use of them. I will discuss some of the challenges in making these 
 techniques accessible to a wider non-specialist user base. In particular\,
  I will describe two important problems that need to be solved to make thi
 s possible: automatic runtime code generation targeted at different hardwa
 re platforms\; and automatic expression rewriting to maximise accuracy wit
 h low precision numerics.\n
LOCATION:SS03\, Computer Laboratory\, William Gates Building
END:VEVENT
END:VCALENDAR