Teleconference 2017-08-02

Attendees: Ken Moreland (SNL), Alok Hota (SNL), Tom Otahal (SNL), James Kress (ORNL), Dave Pugmire (ORNL), Rob Maynard (Kitware), Sujin Philip (Kitware), Allison Vacanti (Kitware), Berk Geveci (Kitware), Matt Larsen (LLNL), Hank Childs (UO), Abhishek Yenpure (UO), David Camp (LBL), Brent Lesley (LBL/UO)

Intel HPC Connection is accepting submissions for presentations. It would be good to have some representation for VTK-m. Rob has said he would be interested and should not have any conflicts. Hank could help if necessary as well.

Alok working on loop tiling.

Tom is working on production improvements for external faces. He has implemented a fast path for structured data. Also working on passing cell data.

Berk brought up the need to skip over generating external faces for ghost cells. However, there is not yet any concept of ghost cells in external faces.

Dave and Abhishek have been working on the PIC particle advection.

James has been adding things to the rendering structure to make it more usable. Right now has a merge request for adding color to lines in 1D view.

Rob is essentially done with the gradient algorithm. However, for axis-aligned structured grids, we would like to look at neighboring points, which is not currently supported. Rob looking into an infrastructure for imaging algorithms that need point neighbor information.

Allie adding code to correctly use cudaMalloc vs. cudaManagedMalloc.

Sujin is working on some issues with connectivity information with explict cell sets cell to point when a point had no cells attached. Was also an issue where cell to point does not work at all with permutation cell set.

Matt's student, Manish, is implementing line drawing algorithms for the raycasting-based rendering views.

Matt is working on a rendering pipeline so that multiple renderings can be composed together.

Matt is having an issue with CDash where for some reason an object file is being corrupted in one of his tests.

Brent has been working on an image segmentation project this summer. This month there will be a week long hackathon to merge the shared memory and distributed memory parallel systems.