Teleconference 2017-09-06

Attendees: Ken Moreland (SNL), James Kress (ORNL), Manish Mathai (UO/LLNL), Abhishek Yenpure (UO/ORNL), Matt Larsen (LLNL), Mark Kim (ORNL), Matt Letter (SNL), Dave Pugmire (ORNL), Rob Maynard (Kitware), Sujin Philip (Kitware), Allie Vacanti (Kitware), Berk Geveci (Kitware), Ollie Lo (LANL)

James is working on SC/ECP demo. Mostly XGC readers.

Mark is continuing on texture visualization stuff.

Dave P. has also been working on the demo.

Dave's team has been working on integrating particle advection with DIY. This might be work that migrates to VTK-h.

Manish has been working on getting annotations to work in rendering. Now has 2D line drawing working. Currently tackling texture rendering.

Abhishek thinks VTK-m is "awesome," so it's official.

Abhishek has been working on the particle advection and will be working on performance studies. Also plans to work on re-scheduling particles to help with load balancing.

Pat was creating a merge request for her cosmology work. Struggling a bit with the gitlab process.

Matt Larsen has been fixing problems with the triangulator that have caused some raytracing artifacts. Also added a fast path for external faces in structured meshes.

Larsen to a pass on improving BVH structures. Trying to make it more general to use as locators and to identify many locations at once.

Rob has been wrapping up the last of the gradient work. Added some fast paths for permutations of structured cells. Also adding fast paths for Marching Cubes.

Allie looking at vertex clustering. It is working but not getting the performance we want. There is a question of why we are getting fewer triangles with the VTK-m algorithm vs. the VTK algorithm.

Sujin has been working on virtual calls. Implementing a small experiment to figure out how to call across libraries in CUDA.

VTK-m will be a git submodule to VTK very soon.