Teleconference 2018-11-28

Attendees: Ken Moreland (SNL), Nick Davis (SNL), Taylor Berger (SNL), Ollie (LANL), Abhishek Yenpure (UO), Matt Larsen (LLNL), Dave Pugmire (ORNL), Robert Maynard (Kitware), Berk Geveci (Kitware), Haucheng Liu (Kitware), Hank Childs (UO), Allie Vacanti (Kitware)

ECP notes and announcements:

• End of Q1 coming up fast!
  • At the beginning of December will start collecting progress for month/quarter.
  • Want to have all of the milestones completed by December 19 at the latest.
• Annual meeting is soon (January 14-18)
  • Everyone going must register by December 14 at https://ecpannualmeeting.com/
  • Dave P. is working on our poster
  • We are part of a tutorial proposal
• Mike Heroux is looking for "User Stories"
  • Is there anything that ECP can change that would help us?
  • Should we request something about computer time?
    • Small amount (50K hr?) of discretionary time for debugging and performance testing?
      • Agreed, we should ask
    • Set up nightly build/test?
      • Berk and Rob will submit - Ken sends link
  • Ask to fund everyone directly instead of through subcontracts?
    • Subcontractors would like this
  • Anything else about administration?
• A new "Capability Assessment Report" revision is coming up
  • I don't expect anyone other than myself to have to respond, but I will let you know.

ECP deliverable updates:

• [MS-18/09] Better Dynamic Types Impl
  • Rob: It is moving forward
  • ArrayHandleVirtual is in a merge request
• [MS-19/01] VTK-m Release 2
  • Done
• [MS-19/02] Productionize zfp compression
  • Matt and Dave has been in discussion for hand off of zfp code
  • Mark Kim, who has worked on it, will be working on it
• [MS-19/03] Clip
  • Abhishek working hard to getting the Maredith algorithm and its tables in the existing implementation

Where should the VTK-m User's Guide be posted?

• For now, just store it on the Wiki until we make the jump to a full website.
• Allie: It would be nice to convert to html as well

Consolidating worklet tests

• MR 1432 merged
• Why are the executables named Worklet_list_of_devices? Why not follow the UnitTests_vtkm_worklet_testing convention?
  • Suggest changing this
• Will the rendering tests be converted as well?
  • The should be. Hauchang will work on that next.
• Should there be a simplified way to specify devices on the command line?
  • Ken suggested having a vtkm::Initialize method that would consolidate all initialize methods (such as this and InitLogging).
  • It should be optional to call that
  • Allie maybe volunteers

LANL in debugging hell for hypergridtrees for a few days, but things are looking better now.

Brent has been working on verdict (cell metrics stuff).

UO has a new student: Steven Walten.

Matt has been looking into integrating VTK-m in SW4.

Matt needs expanded capabilities of querying meshes based on the results of locator queries. For example, it is not straightforward to, once a cell containing a point is found, to interpolate within the mesh to that point.

Dave has some progress on ghost cells. Some progress on particle tracing.

Rob has been working on virtuals. Also released VTK-m 1.3.