BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Speeding up HEP experiments' software with a library of fast and a
utovectorisable mathematical functions
DTSTART;VALUE=DATE-TIME:20131015T154700Z
DTEND;VALUE=DATE-TIME:20131015T160700Z
DTSTAMP;VALUE=DATE-TIME:20210115T161741Z
UID:indico-contribution-1512757@indico.cern.ch
DESCRIPTION:Speakers: Danilo Piparo (CERN)\nDuring the first four years of
data taking at the Large Hadron Collider (LHC)\, the simulation and recon
struction programs of the experiments proved to be extremely resource cons
uming. In particular\, for complex event simulation and reconstruction app
lications\, the impact of evaluating elementary functions on the runtime i
s sizeable (up to one fourth of the total)\, with an obvious effect on the
power consumption of the hardware dedicated to their execution. This situ
ation clearly needs improvement\, especially considering the even more dem
anding data taking scenarios after this first LHC long shut down. A possib
le solution to this issue is the VDT (VectorisD maTh) mathematical library
. VDT provides the most common mathematical functions used in HEP in an op
en source product. The functions' implementations are fast\, can be inline
d\, provide an approximate accuracy and usable in vectorised loops. Their
implementation is portable across platforms: x86 and ARM processors\, Xeon
Phi coprocessors and GPGPUS. In this contribution\, we describe the featu
res of the VDT mathematical library\, showing significant speedups with re
spect to the Libm library and comparable accuracies. Moreover\, taking as
examples simulation and reconstruction workflows ran in production by the
experiments\, we show the benefits of the usage of VDT in terms of runtime
reduction and stability of physics output.\n\nhttps://indico.cern.ch/even
t/214784/contributions/1512757/
LOCATION:Amsterdam\, Beurs van Berlage Effectenbeurszaal
URL:https://indico.cern.ch/event/214784/contributions/1512757/
END:VEVENT
END:VCALENDAR