Difference between revisions of "Teleconference 2023-11-08"

From VTKM
Jump to navigation Jump to search
(Created page with "Attendees: Ken Moreland (ORNL), Berk Geveci (Kitware) == ECP Updates == * KPP-3 ** All KPP-3's are in * Highlights ** Based on KPP-3 work ** VisIt? ** A second WDMApp? ** Po...")
 
 
Line 1: Line 1:
−
Attendees: Ken Moreland (ORNL), Berk Geveci (Kitware)
+
Attendees: Ken Moreland (ORNL), Berk Geveci (Kitware), Roxana Bujack (LANL), Jay Wang (ORNL), Tushar Athawale (ORNL), Dave Pugmire (ORNL), Vicente Bolea (Kitware), Abhishek Yenpure (Kitware), Jeff Amstutz (Nvidia), Mark Bolstad (SNL)
  
 
== ECP Updates ==
 
== ECP Updates ==
Line 27: Line 27:
 
No meetings next two weeks.
 
No meetings next two weeks.
  
−
Progress being made on getting VTK-m working with ParaView on Aurora. There were some linking issues with SYCL. Abhi has a branch that allows the VTK build to work with SYCL. Silvio now needs to get a special ParaView build that uses this branch of VTK. https://gitlab.kitware.com/vtk/vtk/-/merge_requests/10592
+
We should ask AMD for some hardware that Kitware can install in their testing cluster. Ask if they have consumer-level hardware that supports ROCm. This would be easier for Kitware to buy and regardless easier to set up as it would not require extra server mounting hardware.
−
 
 
−
There are issues with the install of ParaView on Frontier. You cannot just load it directly. You need to first load multiple dependent modules. There is a script to do that correctly.
 
−
 
 
−
<pre>
 
−
source /lustre/orion/csc444/world-shared/spack/modules/module-use-gcc.sh
 
−
module load paraview
 
−
</pre>
 
−
 
 
−
The Todd Gamblin through the software stewardship work is creating a "High Performance Software Foundation" umbrella organization under the Linux Foundation for HPC code. Ken and Berk have be pursuing this. There have been some issues that popped up over who has authority to start this and similarities of trademark to VTK.
 
−
 
 
−
We have a read-the-docs site for VTK-m documentation up and running. Right now the documentation is minimal, but more is forthcoming.
 
−
 
 
−
The latest RC for the release is out. We are waiting for Cyrus to confirm that it works well for Ascent.
 
−
 
 
−
Jeff's VTK-m GUI is now available open-source on GitHub: https://GitHub.com/NVIDIA/VTK-m-graph
 
−
 
 
−
Intel is doing a live demo on Sunspot at SC. It would be interesting if VTK-m was part of that. Currently ParaView is running on Sunspot. We could potentially run VTK-m under ParaView if that could be made working. There are some current linking issues with VTK-m/VTK, but Abhi thinks he can get it working within the next month. Another option would be to use Jeff's VTK-m connector demonstration to show something running and rendering with ANARI.
 

Latest revision as of 13:47, 8 November 2023

Attendees: Ken Moreland (ORNL), Berk Geveci (Kitware), Roxana Bujack (LANL), Jay Wang (ORNL), Tushar Athawale (ORNL), Dave Pugmire (ORNL), Vicente Bolea (Kitware), Abhishek Yenpure (Kitware), Jeff Amstutz (Nvidia), Mark Bolstad (SNL)

ECP Updates

  • KPP-3
    • All KPP-3's are in
  • Highlights
    • Based on KPP-3 work
    • VisIt?
    • A second WDMApp?
    • Possible WarpX (get particle advection into Ascent?)
    • Rendering work with Manish
  • Confluence and JIRA going away December 31
  • Zoom going away May 31

Porting Activities

ECP task updates

ECP/VTK-m Project Management

Roundtable

No meetings next two weeks.

We should ask AMD for some hardware that Kitware can install in their testing cluster. Ask if they have consumer-level hardware that supports ROCm. This would be easier for Kitware to buy and regardless easier to set up as it would not require extra server mounting hardware.

Retrieved from "https://m-old.vtk.org/index.php?title=Teleconference_2023-11-08&oldid=1440"