Teleconference 2016-05-25

Revision as of 18:11, 25 May 2016 by Kmorel (talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search

Attendees: Ken Moreland (SNL), Alok Hota (SNL), Matt Larsen (LLNL), Eric Bruger (LLNL), Jeremy Meredith (ORNL), Chris Sewell (LANL), Rob Maynard (Kitware), Berk Geveci (Kitware)

The dejagore dashboard is having the CUDA memory issue again. Apparently upgrading the card did not solve the problem. Rob is going to disable all other dashboards to see if we still have the problem.

The recent labeling contributions need a set of bitmap fonts. To handle this, there is an array in the source that defines the bitmap. There are two issues with introducing this. The first is that these fonts have their own license. Mixing that with our own code with our own license might not be a good idea. The second issue is that if the array is in a header, then you replicate the array in every object file that includes the header.

One solution suggested was to create a TPL directory in VTK-m. You could declare a different license under that directory. Also, we could build to its own library, thus circumnavigating the bloating issue.

Rob working on VTK/VTK-m integration. Marching cubes and level of detail working. Working on threshold. A bit more work because the output is a permutation dataset. Image data and unstructured grids are working. Polydata only supports homogeneous types.

Chris is working with Hamish Carr. On track for an LDAV submission for a parallel contour tree submission.

VTK-m/Legion integration continues. Ran into some issues when increasing the size of grids. This might be an issue with virtual mapping in Legion.

LANL is bringing in Samuel, Hank Childs's student. Will probably be looking into wavelet compression in VTK-m.

Jeremy's annotation is ready to go modulo the issue with font arrays.

Matt getting non-mesa rendering. Also working on annotations that do not rely on OpenGL.

We have been talking about a doing another VTK-m code sprint for a while now, but the planning has stalled. This week we resumed discussion of it. The first question is where we are going to have the meeting (and then when it will be). There were several suggestions proposed. Berk has volunteered to take the first logistical step to determine where the meeting will be. Here is a list of locations proposed:

  • Clifton Park, NY
  • Los Alamos, NM
  • Albuquerque, NM
  • Argonne, IL
  • Eugene, OR
  • Portland, OR
  • Livermore, CA