Teleconference 2016-02-24

Attendees: Ken Moreland (SNL), Hank Childs (ORNL), Dave Pugmire (ORNL), James Kress (UO), Chris Sewell (LANL), Matt Larsen (LLNL), Tom Fogal (NVIDIA), Robert Maynard (Kitware), Berk Geveci (Kitware)

Ken has been working on the VTK-m User's Guide. Worklet and execution environment documentation is finished. The new version is uploaded on the wiki. There is some question about how much it should be integrated with the textbook. Also, the book should probably be published by Kitware with an open license in the same way as ParaView and similar books.

Base rendering support for structured grids is built. OS mesa renderer for a surface is also written. Matt is working on adding ray tracing. There is a test that produces an image. Working on an unstructured grid renderer. Jeremy is working on annotations (fonts). The work is in Matt's rendering branch. It has not been merged in yet because there is a build issue when OS Mesa is not available. There needs to be some work to support probably X host and EGL. There needs to be some more CMake configuration to manage this. Rob and Dave will talk offline.

Chris is putting together the GTC talk. We should have the IO, filter, and rendering packages in good enough shape to present that there.

The Legion/VTK-m integration was worked on during a visit between Kitware and LANL.

Filter policies are looking very good. There are some minor issues with the execution signatures, but those should be resolved soon. There are some things we are not totally happy with the design, but it is a good first pass.

Matt working on renderers and Strawman work. Looking into atomic operations in VTK-m. Support 32-bit atomics, signed and unsigned. But not working on Windows. Waiting for a requisition for an Windows machine.

VTK filter implementations for threshold and contour as a first pass. For all structured data and all unstructured grids that we support and polydata of a single cell type.

Working on a GTC implementation for PyFR integration.

Legion is looking at the VTK-m filter design to see if that helps with integration.