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:20230831T095746Z
LOCATION:Davos
DTSTART;TZID=Europe/Stockholm:20230627T100900
DTEND;TZID=Europe/Stockholm:20230627T101000
UID:submissions.pasc-conference.org_PASC23_sess110_pos120@linklings.com
SUMMARY:P38 - Loki v0.1.1: A Source-To-Source Translation Tool for Numeric
 al Weather Prediction Codes and More
DESCRIPTION:Poster\n\nMichael Staneker, Ahmad Nawab, Balthasar Reuter, and
  Michael Lange (ECMWF)\n\nAll known or presumed candidates for exascale su
 percomputers will feature novel computing hardware or heterogeneous archit
 ectures, with GPUs currently being a cornerstone of this development. Usin
 g these machines efficiently with today's numerical weather prediction (NW
 P) codes requires adapting large code bases to new programming paradigms a
 nd applying architecture specific optimizations. Encoding all these optimi
 zations within a single code base is infeasible. Source-to-source translat
 ion offers the possibility to use existing code as-is and apply the necess
 ary transformations and hardware-specific optimizations. To that end, we p
 resent Loki, a Python tool purpose built for ECMWF’s Integrated Forecastin
 g System (IFS) that offers automatic source-to-source translation capabili
 ties based on compiler technology to target a broad range of programming p
 aradigms. Following the recent open-source release of version 0.1.1, Loki 
 is available on GitHub and ready for testing by the weather and climate co
 mmunity and beyond. It offers an API to encode custom transformations, all
 owing for expert-guided code translation. It supports multiple Fortran fro
 nt ends, and can output Fortran, C, Python and now also CUDA Fortran. In t
 his poster, we highlight Loki’s key features, and present a performance co
 mparison between auto-translated code and manually optimized variants.\n\n
 Session Chair: Jibonananda Sanyal (National Renewable Energy Laboratory)
END:VEVENT
END:VCALENDAR