Minutes of PRISM 2005 Community Meeting

(S. Valcke -07/12/2005)

The PRISM 2005 Community Meeting took place in Toulouse at CERFACS on November 16-17, gathering about 30 persons from different countries in Europe and one GFDL/ESMF representative (list of participants). The objectives were to inform the community about PRISM latest developments and evolution, to review the results of the PRISM User Survey that was realized in the European climate modelling community during the last few months, and help the PRISM Team developers gather the community's growing expectations into a work plan for the next 3 years.

The Community Meeting was divided into 5 sessions gathering presentations and discussions took place. The Community Meeting was followed by a PRISM Team meeting which organized the follow-up of the User Survey and Community Meeting. The present document contains links to the different presentations and highlights some discussions that took place during those two days; the minutes of the PRISM Team meeting that followed will be available shortly.

Community Meeting day 1 (Nov 16): Overview of current achievements and User Survey summary

  • Welcome by Jean-Claude Andre (CERFACS)
  • Session 1: “What is PRISM today?” (chair: Lois Steenman-Clark, NCAS/CGAM, Reading).

The objective of this session was to review PRISM current achievements and software tools.

  • Session 2: Experiences with PRISM and User Survey summary (chair: Reinhard Budich, MPI-M).

In the first part of the session, PRISM users presented their experience with the current PRISM tools and some of  their present expectations. This was followed by a presentation summarizing the results of the PRISM User Survey that was realized in the community over the past few months.

 

Community Meeting day 2 (Nov 17): Detailed analysis of User Survey for each PRISM tool and future evolution.

  • Session 3: Detailed analysis of User Survey and proposition of evolution for each PRISM tool (chair: Mick Carter, UK Met Office).

In this session, each PSI work group leader reviewed the part of the user survey specific to his/her tool and proposed a path for future developments. Additional discussion took place with the users during the meeting and some points discussed are summarized below. A presentation by Reinhard Budich on the PRISM User group and relations with the community (not initially planned) was also included. The complete User Survey results including replies by the work group leaders, and a summary of the User Survey including also plans for future developments will be available soon (see Document 1 and Document 2 in the PRISM team minutes).

  • Coupler and I/O (Sophie Valcke, CERFACS). Additional points discussed:
    • A full typical makefile should be provided with the sources to facilitate compilation of OASIS outside the SCE.
    • No effort should be spent to optimize OASIS3, but some efforts can be spent to find optimized way of using it and according user support should be provided (e.g. IPSL efforts to use many OASIS3 interpolation process in parallel)
    • CERFACS should seek numbers (CPU time,  elapse time) obtained by OASIS users and put “typical” numbers (with well- defined signification) on the web.
    • For OASIS3, the toymodel, as it is distributed as a demonstration example, should not use any binary auxiliary files (big endian vs little endian problem); this should be solved in the next official release.
    • The documentation should include precision about the version of auxiliary libraries (e.g. mpi, xmllib) used and working.
  • Standard Compiling Environment (Stephanie Legutke, MPI-M/M&D). Additional points discussed:
    • The interest in coding rule parser was mentioned.
    • The interest of a tool for automatic reorganization of a model sources into the standard PRISM directory was discussed.
    • The modularity of the SCE, especially between the assembling and the compiling parts needs to be clarified.
  • Standard Running Environment (Veronika Gayler, MPI-M/M&D). Additional points discussed:
    • Some users expressed doubts on the feasibility of designing and developing a standard running  environment, not too complex or too rigid.
    • Users insisted on improving the user initial experience with the SCE (and PRISM tools in general), maybe by providing a clearer step-by-step user guide and a demonstration example better documented (cf toymodels).
    • There is a need to clarify/implement more modularity so that “simple things stay simple”.
    • A map of the standard PRISM directory tree could be provided to help user understand it.
    • Some users complained that the SCE was “too close to DKRZ” environment. It should be investigated why those users made that remark.
    • In general for all tools, there is a strong need to accelerate the publication of documentation on the web site.
  • Graphical User Interface and Web Service System (Kristian Mogensen, ECMWF). Additional points discussed:
    • There is strong user interest for a “light“ GUI version including only  functionality to build OASIS4 XML files.
    • There is some interest from the users to access a prepIFS/SMS demonstration based on a real model (maybe NEMO, when it will be adapted at ECMWF?).
    • There is a need to clarify/support modularity between the different GUI parts (e.g. configuration vs monitoring).
    • The PrepIFS/OASIS4 parts may be more important for the users than SMS?
    • There is interest from the community to evaluate the tool and/or interact with those who will (e.g. in the framework of the GEMS project).
  • Data management, Diagnostics, and visualisation (Joerg Wegner, MPI-M/M&D)
    • Currently, the aim is to keep on developing low-end (LE) graphics with cdat/cdo but the need to have a wider and more formal review of cdat/cdo was expressed.
    • This work group could use the PRISM umbrella to engage with development of Grads, cdat and possibly other visualisation tools such as ferret, etc.
    • This workgroup should act as a technology watch group for post-processing and visualisation tools.
    • This workgroup will organize a workshop in January. The objectives of the workshop and its relation with PRISM objectives must be clarified and advertised.
  • User group and relations with the community (Reinhard Budich, MPI-M).
    • The main questions that were asked are:
      • How do we make the community to review PSI activities?
      • How do we increase buy-in?
    • The PSI work plan will be updated following the User Survey and the Community Meeting but other means are also needed to increase the buy-in; the following were mentioned:
      • newsletter
      • announcement posting
      • meetings (once a year is too much), maybe attached to other meetings?
      • special interest groups
      • request for comments
      • forums
    • There might be two different categories of users with different needs: small teams with reduced experience and manpower to address software infrastructure issues vs. large, experienced and well supported teams.
    • The community needs to be more involved to identify better what we want to do next (future funding).
    • It was proposed to involve more the  computing centres.
    • PSI should be more involved in data and model metadata definition. <>
  • Session 4: Other User Survey remarks and additional PRISM tools (chair: Sophie Valcke, CERFACS)
  • Session 5: Management aspects of PRISM (chair: Eric Guilyardi, IPSL, CNRS)

These two sessions were grouped into a general session of discussion on the User Survey results, on the role of the  PRISM Support Initiative, and on criteria to define tools that can be considered "PRISM tools" and which development can be considered PRISM effort. Those discussions lead to some conclusions on the next day during the PRISM Team meeting and will be summarized in a separate document produced by the PRISM Team in the coming weeks.

 

[ Top ]

Site Hits: