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:20190719T085743Z
LOCATION:HG D 1.1
DTSTART;TZID=Europe/Stockholm:20190612T133000
DTEND;TZID=Europe/Stockholm:20190612T140000
UID:submissions.pasc-conference.org_PASC19_sess132_msa192@linklings.com
SUMMARY:Towards Targeting Multiple HPC Architectures with IFS Physics
DESCRIPTION:Minisymposium\nClimate and Weather\n\nTowards Targeting Multip
 le HPC Architectures with IFS Physics\n\nLange, Marsden\n\nThe increasing 
 complexity and diversity of HPC architectures demands a sustainable strate
 gy to migrate existing operational weather prediction codes to a flexible 
 infrastructure that is able to cope with multiple programming models and v
 ersatile hardware characteristics. Modern software engineering techniques 
 can be used to achieve this synergy between performance and flexibility, e
 ven within the constraints of operational codes, through the use of librar
 ies, automated code generation and domain-specific languages (DSL). In thi
 s talk we present parts of the technological roadmap for the ECWMF IFS tow
 ards multi-architecture capabilities with particular focus on physical par
 ameterizations. We highlight some of the software technologies used to fac
 ilitate an approach based on "evolution" via source-to-source translation 
 rather than the "revolution" of a full rewrite. We show preliminary perfor
 mance results for a standalone cloud microphysics parameterization on mult
 iple HPC architectures, including GPU and FPGA, and demonstrate a software
  toolchain to generalize the approach to a wide range of physical paramete
 rizations before highlighting the importance of flexible data layout manag
 ement for weather prediction software.
END:VEVENT
END:VCALENDAR

