BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Europe/Stockholm
X-LIC-LOCATION:Europe/Stockholm
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20210916T132452Z
LOCATION:Michel Mayor
DTSTART;TZID=Europe/Stockholm:20210708T110000
DTEND;TZID=Europe/Stockholm:20210708T113000
UID:submissions.pasc-conference.org_PASC21_sess157_msa321@linklings.com
SUMMARY:Performance Optimization and Productivity of Alya towards Exascale
DESCRIPTION:Minisymposium\n\nPerformance Optimization and Productivity of 
 Alya towards Exascale\n\nBorrell, Garcia-Gasulla, Mira, Houzeaux\n\nAlya&n
 bsp;is the high-performance computational mechanics code developed at the 
 Computer Applications in Science and Engineering (CASE) department from th
 e Barcelona Supercomputing Center. A highly multidisciplinary team develop
 s alya to solve multi-physics problems from several application areas. Par
 ticularly Alya is part of the software hub of six European Centers of Exce
 llence for HPC Applications. Namely: CoEC from the combustion simulation d
 omain, EoCoE from the energy sector, Excellerat from the engineering domai
 n, ChEESE from Solid Earth simulations, ComBioMed focused on biomedical ap
 plications, and the Raise CoE focused on AI- and simulation-based engineer
 ing. Alya was designed for massively parallel supercomputers from scratch.
  Its parallelization embraces four computer hierarchy levels: 1) A domain-
 decomposition approach with MPI is used for distributed memory paralleliza
 tion. 2) At the node level, both loop and task parallelisms are considered
 . Dynamic load balance techniques have also been introduced both at the no
 de and system levels. 3) At the CPU level, most computing-intensive kernel
 s are optimized for vectorization. 4) Finally, accelerators like GPU are a
 lso exploited through OpenACC pragmas or with CUDA, to further enhance the
  code's performance on heterogeneous computers. In this presentation, we w
 ill describe the evolution of the parallelization and optimization strateg
 ies adopted in the kernel of Alya, with the support of the Performance Opt
 imization and Productivity (POP) European Center of Excellence.\n\nDomain:
  Chemistry and Materials, Climate and Weather, Physics, Solid Earth Dynami
 cs, Life Sciences, Engineering
END:VEVENT
END:VCALENDAR