Attendees: Ken Moreland (SNL), Matt Larsen (LLNL), Mark Kim (ORNL), Dave Pugmire (ORNL), Rob Maynard (Kitware), Berk Geveci (Kitware), Utkarsh Ayachit (Kitware), Sujin Philip (Kitware), Tom Otahal (SNL), Ollie Lo (LANL), James Kress (ORNL)
Matt still refactoring raytracing to support points. Working now on scaling points based on scalar field. Next up is r-z meshes. There has been some work by Kitware and ORNL with r-z meshes, so there should be some sharing that happens.
Matt is interested in the connected components that Ollie is working on. Currently sorting out some merge conflicts.
Dave implemented dot product and cross product worklets for predefined function stuff.
Dave is looking at particle advection code and doing it in MPI parallel.
Rob has been working on predefined function stuff. Converting constant fields to color arrays. Working on more generalized color-table support.
Rob has been working on the new CMake re-write. Part of this was reenable several warnings during CUDA compile and fixing the warnings.
Rob added some fixes for the Intel compiler.
Berk has been working with some fusion folks. The fusion data uses an r-z mesh, and that has required new forms of cell sets to handle that data. Planning a new coordinate system to work in the r, z, theta space. Next steps: VTK-m resampling methods to do things like volume rendering.
Sujin made some changes to make execution whole arrays always use the signature tags.
The virtual coordinates is mostly working. Just one test failure on windows.
Utkarsh is working on integrating DIY into VTK-m so that you can do hybrid parallelism. DIY is being included as a third party library. (Linking external library will also be supported.)
Tom is getting ready to open a merge request for the parallel radix sort for TBB. Also doing performance testing of parallel radix sort on different processes and figuring out where the thresholds of using serial algorithms are.