Teleconference 2018-08-22
Attendees: Ken Moreland (SNL), Matt Letter (SNL), Abhishek Yenpure (UO/SNL), Rob Maynard (Kitware), Sujin Philip (Kitware), Allie Vecanti (Kitware), Berk Geveci (Kitware), Matt Larsen (LLNL), Ollie Lo (LANL), Dave Pugmire (ORNL)
Matt Letter is working on a readme for the user's guide.
Abhishek is doing performance tests on merge points algorithm and comparing against VTK.
Rob has been working on the runtime device adapter work. Trying to modify try execute to favor a particular device (such as where data already are) as opposed to the generally preferred.
Sujin working on enabling separable compilation on CUDA. Only one test is failing. Not sure why.
ORNL plans to get the rendering milestone tied up next week.
Will start implementing ghost cells soon. Dave proposed putting ghost cells on the cell set itself, but the general consensus of the group was it would be better to make ghost cell indication a field.
One example in VTK-m needs VTK. It is an odd example that does not seem to do anything with VTK-m. It would be nice to get rid of it. It seems wrong to have VTK-m build refer to VTK (especially because VTK refers to VTK-m).