-
Notifications
You must be signed in to change notification settings - Fork 66
Telecon: 2022 03 09
Cyrus Harrison edited this page Mar 9, 2022
·
4 revisions
Discussion Topics:
-
Crusher / Spock porting
- Update on Progress
- Plan: Particle Advection with ARM Wind Data. (Either in situ or via replay)
- Talk about VTK-m filters lib in VTK-h, object libs, etc.
- Kokkos super hack is blocking features (rendering? vtk-m filter libs)
- This is a complex order of operations related to setting up Kokkos issue
- Rendering: Rocm goes into a opt spiral for rendering lib. (Takes a long long time)
-
Ghost Data (VTK-m will provide a way to accept the name of the ghosts array)
-
Cell arrays with many variables in VTK-m (Energy Groups)
- Best path: Try lots of fields vs array var
Still Pending:
- VTK-m CUDA Arch vs Lib sizes?
- we can explore making a path to directly pass
sm_ZZ
(we have a ticket for this in VTK-m)
- we can explore making a path to directly pass
- Cell Metrics (getting volume of each cell)
- 2 VTK-m filters - Cell Metrics and Mesh Quality (similar, we should check Cell Metrics, we have looked at Mesh Quality)
- We might want a fast path in VTK-m for cell metrics for structured grid
- Cell Metrics is simpler than Mesh Quality