Notes from October 20th, 2010 meeting - CDAT/cdat GitHub Wiki

  • Topic:
  • Parallelizing CDAT's Core
    • Attendants
    • Charles Doutriaux
    • Jim Ahrens
    • Tom Maxwell
    • Claudio Silva
    • Huy Vo
    • Peer-Timo Bremer
    • Valerio Pascucci
    • John Patchett
    • Bob Drach
    • Emanuele Santos
    • Berk Geveci
  • Summary
    • Some think we need to focus on use case 1 and 2 in the context of Paraview first and uses these to report quickly to sponsors and show great performance increase
    • Galen should be reporting soon with preliminary result on reading data in
    • There is a strong feeling that we need to think strong and hard on the design
    • seems that temporal/linear parallelization is the main thing we need to focus on for now
    • Should we come up with a vistrails/pythonic language/dialect ?
    • If we tweak the user scripting language, we need to come up with a translator
    • Need to keep the metadata around and pass them to Paraview
    • We should expect 50% of GUI-based users and 50% of scripting-based users
    • Everybody is going to look into more details at the CDAT specific use cases as well, see: Parallelizing CDAT Core Use Cases