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:20190719T085744Z
LOCATION:HG D 1.2
DTSTART;TZID=Europe/Stockholm:20190613T121500
DTEND;TZID=Europe/Stockholm:20190613T124500
UID:submissions.pasc-conference.org_PASC19_sess167_msa130@linklings.com
SUMMARY:Performance of the New Generation Unstructured Mesh Global Ocean M
 odel (FESOM2) on Modern Supercomputers. Challenges and Bottlenecks
DESCRIPTION:Minisymposium\nClimate and Weather\n\nPerformance of the New G
 eneration Unstructured Mesh Global Ocean Model (FESOM2) on Modern Supercom
 puters. Challenges and Bottlenecks\n\nKoldunov, Aizinger, Rakowsky, Scholz
 , Sidorenko...\n\nHigh computational power is clearly needed in studies re
 lying on the modeling of physical processes in the ocean. Since the first 
 internal Rossby radius is decreasing in high latitudes down to several kil
 ometers, very fine meshes are necessary to simulate the emerging eddy dyna
 mics. We will introduce Finite-volumE Sea ice-Ocean circulation Model, Ver
 sion 2.0 (FESOM2), formulated on unstructured meshes, that make it possibl
 e to utilize more flexible meshes with variable resolution. We discuss its
  performance in the global eddy-resolving simulations and regional simulat
 ions down to 1 km scale. We will show the results of the analysis of the m
 ain computational kernels with a special focus on bottlenecks in parallel 
 scalability. Several model enhancements, improving this scalability for la
 rge numbers of processes, are described and tested. Model grids at differe
 nt resolutions are used on four HPC systems with differing computation and
  communication hardware to demonstrate the model's scalability and through
 put. Furthermore, strategies for improvements in parallel performance are 
 presented and assessed. We show that in terms of throughput FESOM2.0 is on
  par with the state-of-the-art structured ocean models and in a realistic 
 eddy resolving configuration (1/10 degree resolution) can produce about 16
  years per day on 14 000 cores.
END:VEVENT
END:VCALENDAR