Difference between revisions of "Teleconference 2021-12-15"
Jump to navigation
Jump to search
(Created page with "Attendees: == ECP Updates == * ECP Project Reviews ** ECP/VTK-m scheduled for December 9, 1:15PT, 2:15MT, 3:15CT, 4:15ET ** PI's should be there * Crusher test and developm...") |
|||
| Line 1: | Line 1: | ||
| − | Attendees: | + | Attendees: Ken Moreland (ORNL), Berk Geveci (Kitware), Terry Turton (LANL), Abishek Yenpure (UO), Gunther Weber (LBNL), Nick Davis (SNL), Tushar Athawale (ORNL), Mark Bolstad (SNL), Vicente Bolea (Kitware) |
== ECP Updates == | == ECP Updates == | ||
| − | |||
| − | |||
| − | |||
* Crusher test and development system for Frontier | * Crusher test and development system for Frontier | ||
** Access will be dispensed like on Spock | ** Access will be dispensed like on Spock | ||
| Line 31: | Line 28: | ||
== Roundtable == | == Roundtable == | ||
| + | |||
| + | Gunther brings up an issue where he tried contour trees with large data sets that ran slowly on CUDA but crashed with HIP/Kokkos. This might be an issue of lack of proper unified memory on the HIP/Kokkos end. We are not sure. | ||
R&D100? | R&D100? | ||
Latest revision as of 12:34, 15 December 2021
Attendees: Ken Moreland (ORNL), Berk Geveci (Kitware), Terry Turton (LANL), Abishek Yenpure (UO), Gunther Weber (LBNL), Nick Davis (SNL), Tushar Athawale (ORNL), Mark Bolstad (SNL), Vicente Bolea (Kitware)
ECP Updates
- Crusher test and development system for Frontier
- Access will be dispensed like on Spock
- Information Slides
- Software sustainability town halls
- Let's think about possible highlights
- App engagement
- Success with porting (e.g. run on Aurora)
- Engagement with other ST teams (e.g. using Kokkos)
- WDM
- Submitted highlight on integration with EFFIS
- Expect to submit another highlight when used in a big run
- Possible highlight slide on particle advection with WarpX (Matt)
- Improvement with Kokkos on Spock
Porting Activities
- https://wiki.jlse.anl.gov/display/inteldga/ECP+2.3.4.13+VTK-m
- https://frontier-coe.atlassian.net/wiki/spaces/FCOE/pages/1161625609/Porting+Activities+O21
ECP task updates
Roundtable
Gunther brings up an issue where he tried contour trees with large data sets that ran slowly on CUDA but crashed with HIP/Kokkos. This might be an issue of lack of proper unified memory on the HIP/Kokkos end. We are not sure.
R&D100?
- Thinking of submitting for 2022
- Big stumbling block: needs to be "created" between Jan 2021 to March 2022
- Plan: Release VTK-m 2.0, do a copyright assertion
Rename master branch?
- Need to wait for some changes to git before the transition (for some automated features of repo).
- Can we for a time have a master that mirrors the new default?
- Don't know, but seems possible. Will have to look into that.
