Agenda Notes Monday March 9th 2015 3:00 PM

MPAS-Developer Telecon:

Date: 9 March 2015
Time: 3:00pm MST

Call-in number: 1-866-740-1260
Access Code: 4978161

---

Agenda:

• Readable dimensions -- next steps
• Testing core
• Shared infrastructure pull request guidelines
• Misc. framework fixes resulting from single-precision 3-km benchmarks
• Uniformly decomposed dimensions

Notes from last telecon related to items from above:

• Possibly create a framework timeline workflow
  • It would be nice to describe the time required to get something into framework
  • Could use a testing core to test framework addition and operators
  • Could create a test suite for each core to verify their core is functional and not broken
  • Should discuss at low level telecon
  • Will have a document to describe the process for getting things added to framework
• Testing core
  • Should have a PR by next high level telecon to create / add the testing core.
  • Will have some set of tests for some framework (might not include all tests to begin with, but should be extensible).

---

Notes:

Todd joined

• Init atmosphere core:
  • Has three steps
    • Interpolation of terrain data (called static processing step) (Does not involve vertical dimension)
    • Generate a vertical grid
    • Take atmospheric analysis data, and interpolate to 3D atmospheric grid
  • Problem in static processing step
    • After static processing, nVertLevels is defined as soemthing like 1
  • Fixes:
    • Override definition of variable. i.e. do not let it be read in
    • Allow packages on members of a stream (i.e. do not write a field, even if it exists if the package is off)
• Should merge the bare-bones core without adding testing to it
• Working on framework time frame document:
  • Distinguish thresholds on developments to determine how much time is spent on each development.
    • Incremental updates take less than the full limit?
    • Try to determine upper bound on full process.
• Framework fixes from ATM Single-precision 3-km benchmarks
  • Changes to fix issues with single-precision
  • Changes to fix large processor count issues
  • Changes to fix large number of columns
• Uniformly decomopsed dimensions have been submitted as a PR.
  • Feel free to comment on the design document in the PR