Attendees: Ken Moreland (SNL), Dave Pugmire (ORNL), Berk Geveci (Kitware), Rob Maynard (Kitware), Chris Sewell (LANL), Sam Li (LANL), Matt Larsen (LLNL), Mark Kim (ORNL)
Dave has been working on getting 1D data sets working. Have some Xolotl code to do 1D plots. Want to apply that in a new fission project.
Mark Kim is working on getting GLEW working with OSMesa. GLEW supports OSMesa in 2.0 if compiled correctly. We are going to make a check to ensure that GLEW is 2.0 or higher to hopefully check for this in most cases.
Rob has been working on boost replacement.
There has been some issues on dejagore and moving to CUDA 8. There are some issues with try_compile. Integrating VTK-m with CUDA enabled into VTK.
Berk has been working on integrating VTK with VTK-m enabled in the Sensei framework. There is now an example with it working with one of their miniapps.
Chris is ready for folks to look at the streaming MR so that it can be merged in before too long. There might be some comments on getting the work indices correct. There is also a scan that works with streaming data.
Ollie Lo will start working on the progress in the upcoming fiscal year.
Sam has been optimizing the 2D wavelet compression. Also comparing with other implementations. In terms of speedup, there is similar performance. In terms of accuracy, the VTK-m algorithm is higher because they can take advantage of global coefficients. Although there are not apple-to-apple comparisons on GPU, the performance looks pretty good.
Matt fixed some issues with reading in VTK legacy files that contain string fields. Matt is also participating in an SC demo involving volume rendering enabled with VTK-m.
Implementation on Contour Tree algorithm in VTK-m is complete. There is currently an application to release that at open source. When that is complete, it will be contributed to VTK-m.