1) Current status of the FVCubed Integration (Will & Brian, Bill)

  • default values and resolution dependencies
    Bill claims default values can be used up to 0.25 degree, but
    it is not clear that they run with defaults on J/W test cases
  • bug and/or IBM compiler problem in vertical remapping routine
  • status of all test cases
    Test cases 1, 2 are working. Test case 5 runs with modification
    of n_split
  • problems with the advection test case
    Bizarre! Trying to run test case 3 on Cray XT5 to get a
    different perspective. The advection blob is still being
    truncated in the first time step.
  • order of the dynamics & physics, reversed ?
    Order is now reversed, in line with other dycores (physics first
    then dynamics). Brian will take a quick look to make sure.
  • solved problems in inidat (Brian)
    issues with PIO (ldof); fixed.
    Test: read in initial file, commented out dycore, wrote out history; identical
    latitudes, longitudes on initial file are not correct
  • status of restart I/O
    may have same issues as inidat with ldof function; Will will check
  • status of ESMF clock
    removed
  • are we in sync with NASA/GFDL, can we access the repository
    Will will contact Bill; Brian suggested he might be able to
    trickle these into the vendor branch.
  • versions & version control at NCAR
    cam3_6_57

2) Open issues (Will, Brian, Jim)

Has been updated; several points have been resolved

  • plans how to tackle them, priorities
  • physics coupling
  • time schedules

3) Performance or porting bottlenecks (Will, Pat?)

  • can we run on the Cray system at high resolutions (memory problem solved?)
    Basically solved (although we still use load_balance -1)
  • hybrid parallelization
    waiting to see
  • plans for performance analysis
    some initial results on Cray XT3 sent by Will.
    For proper benchmarking: one month run with monthly averaging

4) Plan for future model tests:

  • dry dycore tests
  • physcis coupling, aqua-planet
  • Mark's experience with NCAR's tri-grid coupling strategy

Latest AMIP runs with land and ocean; Jim reports that it is running.

5) Documentation on web page
https://wiki.ucar.edu/display/ccsm/Implement+FVCUBED+Dycore

  • No labels