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:20250822T115804Z
LOCATION:Campussaal - Plenary Room
DTSTART;TZID=Europe/Stockholm:20250617T103000
DTEND;TZID=Europe/Stockholm:20250617T110000
UID:submissions.pasc-conference.org_PASC25_sess150_pos154@linklings.com
SUMMARY:P27 - Integrating the ICON4Py Python-Based Dynamical Core into ICO
 N
DESCRIPTION:Mauro Bianco (ETH Zurich / CSCS), Magdalena Luz (ETH Zurich), 
 Christoph Muller and Daniel Hupp (MeteoSwiss), Anurag Dipankar (ETH Zurich
 ), Edoardo Paone (ETH Zurich / CSCS), Xavier Lapillonne (MeteoSwiss), Nico
 letta Farabullini (ETH Zurich), Enrique Gonzales Pareder and Hannes Vogt (
 ETH Zurich / CSCS), Ong Chia Rui (ETH Zurich), Till Ehrengruber (ETH Zuric
 h / CSCS), Yilu Chen (ETH Zurich), and Philip Muller and Christos Kotsalos
  (ETH Zurich / CSCS)\n\nThe integration of Python-based high-performance c
 omputing into legacy Fortran climate models offers new opportunities for f
 lexibility and efficiency. This poster presents the integration, in the Fo
 rtran ICON implementation of the dynamical core implemented in Python, as 
 part of ICON4py, a still in-progress Python-based implementation of the IC
 ON climate model. ICON4Py leverages GT4Py for optimizing the numerical com
 putations. For improved interoperability, the Fortran application invokes 
 the Python interpreter, which in turns executes the ICON4py dynamical core
 . The performance of ICON4py is enhanced through the Data Centric (DaCe) b
 ackend of GT4Py, developed by SPCL at ETH Zurich. DaCe applies dataflow tr
 ansformations to optimize memory locality and parallel execution, improvin
 g efficiency across CPU and GPU architectures. This approach provides comp
 utational performance while maintaining the code maintainable. As we plan 
 to use this implementation in a production-setting, the poster will addres
 s the verification and testing process. Performance evaluations demonstrat
 e competitive or improved runtime efficiency compared to the legacy system
 . By integrating Python’s expressiveness with GT4Py and DaCe’s performance
  optimizations, this poster showcases a viable pathway for modernizing cli
 mate models. Our approach exemplifies how  scientific software can evolve 
 to leverage emerging computing paradigms while maintaining accuracy and pe
 rformance.\n\nSession Chair: David Moxey (King's College London)\n\n
END:VEVENT
END:VCALENDAR