Teleconference 2016-12-14

Attendees: Ken Moreland (SNL), Matt Larsen (LLNL), Mark Kim (ORNL), Tom Otahal (SNL), Dave Pugmire (ORNL), James Kress (ORNL), Dongliang Chu (ORNL), Oli Lo (LANL), Rob Maynard (Kitware)

Matt has changed all the ray casting code to follow the filtering model such that it uses a try execute to try different architectures.

Have an unstructued-based sample renderer. It uses a 3D rasterization technique.

Working on CSG ray casting and integrating it into the miniapp.

Getting Dongliang up and running with VTK and VTK-m. Going to be comparing implementations in the two products with Oxbow.

Starting to coordinate with PSI SciDAC project. Now that 1D rendering in place, can convert all of their code to VTK-m. James will be working on this.

Dave has a need for a VTK-m reader of ADIOS data for Xlotl data. The question is whether this reader should be added to the VTK-m library. We did not come to a conclusion on this discussion. On one hand, we want to make it easy to read in data. On the other hand, a general I/O library is a bit out of scope for the project and could become a maintenance headache. We need to have more discussion on this, particularly with Kitware, who has to support this indefinitely.

We need to add an external data mechanism to VTK-m for testing purposes. It will follow similar mechanisms for VTK and ParaView.

Oli is working on floating on floating point atomics and getting TBB tests not actually spawning threads.

Rob has been working on the point gradient filter. It has just been merged into VTK-m.

Rob has also been working on reducing the amount and size of vtables. Still working on that.