Attendees: Ken Moreland (SNL), Rob Maynard (Kitware), Matt Larsen (LLNL), Chris Sewell (LANL), Tom Fogal (NVIDIA), Jeremy Meredith (ORNL), Dave Pugmire (ORNL)
Matt brought up an issue that the current vtkm::Min and Max are slow because they are calling fmin and fmax, which are not getting inlined. We should always use vtkm::Min/Max with the possible exception of compiling for a CUDA device (where these functions are special to the compiler).
Progress on the todo list for VTK-m V1.0:
- Settling of the core interface - Rob working on interface for finding what device adapters are available at runtime. Probably will move device adapter traits outside of internal.
- Standardization of "filters" - Rob is going to work on this. The current suggested approach is that filters compile to libraries and there will be a mechanism to select some alternate form of standard storage and other lists to recompile for different layouts.
- What is the distinction between multipass worklets and filters?
- Rob has a proposal for filter interfaces in issue #17
- Support worklet operations that are not one-to-one output - Done
- Basic file I/O - Done
- Basic rendering - Not started yet. It will be a bit of pressure to be ready by January
- Clean up doxygen - Need to get documentation syncing nightly.
- User's Guide - Trying to catch up.
- Green dashboard
- Still have problems with some CUDA dashboards
- Some working CUDA dashboards have a timer failure
- There have been some interop failures. - Rob thinks they are fixed.
- Some further output issues on Windows (see delve dashboards e.g. https://open.cdash.org/viewTest.php?onlyfailed&buildid=4123995)
- Some timeout issues on renar (e.g. https://open.cdash.org/viewTest.php?onlyfailed&buildid=4123644). Could be caused by JIT of CUDA kernels or might be because multiple things are running.
- Prototype VTK/ParaView/VisIt integration - There was some integration for the SC demo, but that might not be very close to how we integrate in general. Still need to work on that.
- (Technically not part of the release, but should indicate we have viable components.)
Rob has been working on adding vectorization to VTK-m. This is a work in progress and we do not have official results, yet. More later.
Pat Fasil merged some streamline code on structured grids. Hank Childs' team is interested in expanded this to unstructured grids.