Teleconference 2019-10-02

From VTKM
Revision as of 17:59, 2 October 2019 by Kmorel (talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search

Attendees: Ken Moreland (SNL), Matt Larsen (LLNL), Nick Davis (SNL), Robert Maynard (Kitware), Allie Vacanti (Kitware), David Pugmire (ORNL), Hank Childs (UO), Abhishek Yenpure (UO), Sujin Philip (Kitware)

ECP Updates

  • ECP review was last week
    • Overall went well
    • Mike Heroux latched on overlap between VTK-m and Kokkos
  • Capability Assessment Report (CAR) also coming up
    • Maybe the document Jim is asking us to fill out this week?

ECP task updates

Good job everyone with the FY19 deliverables!

ECP/VTK-m Project Management

Rountable notes

Matt noticed some performance abnormality. Matt wrote a filter to check to see if the ghost cells could be removed from a structured data set and remain structured. However, the algorithm that was identifying whether this was possible was running very slow. Matt made some changes to speed it up.

There are some issues with AMR data with lots of blocks. What are our plans for dealing with performance issues with lots of small partitions.

Hank has been getting some bad performance with particle advection on unstructured grids. This seems to be related to time to build locators.

Rob working on image connectivity algorithms. Also fixed invoker to support both mask and scatter.

Sujin looking into dashboard issues. Will be changing the VTK-c name to something else and other cleanup.

No meeting next week.