Teleconference 2017-12-06

From VTKM
Jump to navigation Jump to search

Attendees: Ken Moreland (SNL), Matt Letter (SNL), Ollie Lo (LANL), Matt Larsen (LLNL), Aaron Knoll (NVIDIA), Tom Otahal (SNL), Robert Maynard (Kitware), Sujin Philip (Kitware), Berk Geveci (Kitware), Hank Childs (UO)

Ollie has been looking at connected components implementation. Trying to run a test program on some of the algorithm.

Matt Larsen has been working on fixing background/foreground colors in the rendering library. Also working on log-scaling for 1D, non-OpenGL plots. Possibility of getting a student this summer to ray cast on higher-order meshes.

Tom Otahal is getting close to openning a merge request for CPU radix sort.

Rob has updated the dispatcher to use C++11 variadic arguments instead of FunctionInterface. Also looking at data model converters between VTK and VTK-m.

Sujin found an issue with virtual coordinate system. Ran into an issue with the VTKM_EXEC_CONSTANT declaration. If compiled with CUDA, the data were not available on the host. Removed the constant and instead now declare the arrays as static const in functions.

Hank's team has been working on getting spatial structures working. It works but performance needs to be improved. Also working on particle advection issues. The Oregon folks are going to put down advection and focus on search structures and clip.