BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//talks.cam.ac.uk//v3//EN
BEGIN:VTIMEZONE
TZID:Europe/London
BEGIN:DAYLIGHT
TZOFFSETFROM:+0000
TZOFFSETTO:+0100
TZNAME:BST
DTSTART:19700329T010000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0100
TZOFFSETTO:+0000
TZNAME:GMT
DTSTART:19701025T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
CATEGORIES:Applied and Computational Analysis
SUMMARY:Efficient Computation through Tuned Approximation 
 - David Keyes (KAUST)
DTSTART;TZID=Europe/London:20240516T160000
DTEND;TZID=Europe/London:20240516T170000
UID:TALK215842AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/215842
DESCRIPTION:Numerical software is being reconstructed to provi
 de opportunities to tune dynamically the accuracy 
 of computation to the requirements of the applicat
 ion\, resulting in savings of memory\, time\, and 
 energy. Floating point computation in science and 
 engineering has a history of “oversolving” relativ
 e to requirements or worthiness for many models. S
 o often are real datatypes defaulted to double pre
 cision that GPUs did not gain wide acceptance in s
 imulation environments until they provided in hard
 ware operations not required in their original dom
 ain of graphics. However\, driven by performance o
 r energy incentives\, much of computational scienc
 e is now reverting to employ lower precision arith
 metic where possible. Many matrix operations consi
 dered at a blockwise level allow for lower precisi
 on and\, in addition\, many blocks can be approxim
 ated with low rank near equivalents. This leads to
  smaller memory footprint\, which implies higher r
 esidency on memory hierarchies\, leading in turn t
 o less time and energy spent on data copying\, whi
 ch may even dwarf the savings from fewer and cheap
 er flops. We provide examples from several applica
 tion domains\, including a look at campaigns in ge
 ospatial statistics and seismic processing that ea
 rned Gordon Bell Prize finalist status in\, resp.\
 , 2022 and 2023.
LOCATION:Centre for Mathematical Sciences\, MR14
CONTACT:Hamza Fawzi
END:VEVENT
END:VCALENDAR