Log In | Users | Register
COSMO Web
Edit | Attach | New | Raw | Diff | Print | | Tools
You are here: ICON » Older_meetings » ICONmeet20180117

ICON meeting January 17th, 2018

Participants:

Bernhard Enz (BE); Colombe Siegenthaler (CS); Doris Folini (DF); Guy de Morsier (GDM); Katherine Osterried (KO); Philippe Marti (PM); William Sawyer (WS); Xavier Lapillonne (XL)

Update of each participant:

* Guy de Morsier (GDM):
  • GDM created successfully IFS data for ICON through Mars on ECMWF.
  • GDM successfully compiled ICON on Kesch.
  • GDM is now struggling with ICON grid generation.
  • GDM will begin with LAM simulation as soon as the grid is generated.
  • GDM will be able to report on the COSMO/ICON dynamical core comparison experiment in September.

* Bernhard Enz (BE):

  • BE is looking for a solution to replace nudging since there is for the moment no nudging facility in ICON.
  • BE is thinking of using the LAM mode. However, his domain might be too large.

* Philippe Marti (PM):

  • PM is struggling with cray compiler options/versions on Piz Daint.
  • The gpu version is not OK with cray 8.6.1 -> need 8.5.5. DWD uses cray 8.4, MPI uses intel,gcc and pgi. Cray is needed since they are supporting OpenACC.

* William Sawyer (WS):

  • WS never got amip test case to run (CPU version): get seg fault. The problem seems to be in jsbach. Cray 8.5.5 compiler.
  • MSC student beginning 1st of February for 4 months to work on turbulence (vdiff).

* Colombe Siegenthaler (CS):

  • CS some benchmarking with amip run (cpu version). So has an amip run compiling and running with cray. CS uses "module load cdt/16.11" and cray 8.5.5.

* Xavier Lapillonne, Doris Folini, Katherine Osterried : nothing to report

Summary of the discussions:

* Is it not clear if/when cray will stop support OpenAcc standards. WS is in contact with cray people, hopefully he will have an answer end of February.

* PGI will not support quad precision library, which is required by JSBACH. A work around is to use double precision instead. Luis Kornblueh said it is absolutely not possible.

* Concerning the testing platform: We will probably end up using both Jenkins and buildbot. Jenkins for local tests, and builtbot to get the local development back into the MPI repository.

* Discussion about the way C2SM will handle hosting of ICON code : it will be in Github, similar to what we offer for COSMO. The diversity of the users will be higher that for COSMO since there will be in addition the global modellers. KO and CS will begin to write a draft of the possible workflow. XL will be their Meteoswiss contact.

Next ICON related meetings people are attending:

* ICCARUS 2018, 26.02.2018 - 02.03.2018, https://www.dwd.de/SharedDocs/termine/DE/forschung/termin_iccarus_2018.html : GDM, KO

* ICON Developers Meeting, 06-08 February 2018 : CS, KO, WS, XL

* ICON Infrastructure Partnership Meeting, 08-09 February 2018 : WS, Carlos Osuna

Task points:

* CS gives the details on her amip runs on Piz Daint.

* WS volunteers for mediate Jenkins at CSCS.

* As soon as an existing version runs with cray 8.6.1, everyone should use this version.

Next meeting:

March 2018 spacer
This site is managed by the Center for Climate Systems Modeling (C2SM).
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors. Ideas, requests, problems? Send feedback!