Present: Lucas (minutes), Steven R Brandt (chair), Peter Diener, Beyhan, Helvi Witek, Jamie Bamber, Maxwell Rizzo, Roland Hass, Yosef Zlochower
# Hackathon post mortem
- Steve suggests that high level "how does the infrastructure works" talks may be useful for participants. Helvi added it as feedback for next talk. - Steve says he is open to having the next one at LSU. Steve suggests also that Canada is now a possibility.
# Agenda items
- BBH gallery example. Peter has not been able to get around to it yet. - Backports to Kruskal: Roland and Steve worked on various changes to make libs compile in various Linux distros. Roland reports to be testing it now on ACCESS clusters, and will merge it into master as soon as he is done testing. - Helvi asked Jay, Lucas and Cheng-Hsin to give updates on the projects of the Hackathon: 1. Jay: Ported Radicces M1, into a new code, called NUX. Removed GLS dependencies, close to having a GPU ready. Team effort, Allen, Jamie and Maxwell were the working forces. 2. Cheng-Hsin: Work on adding subcycling support to CanudaX. Got it into the evolution thorn, but Multipole output is not being output correctly. NewRadX performance and profiling the codes (Liwei, Lucas and Erik). 3. Lucas: Worked on getting higher order interpolation working on CapyrX. Initial tests indicate that evolution is working. He mentions wanting to run more tests using Multipole to see if the extracted wave forms are correct. He also worked on adding advective derivatives (working) and matter coupling (still needs work) to CottonmouthBSSN. 4. Steve: Created a JSON-like tags system for interface files. This feature can improve the way information is passed to the driver and aid in Cactus reflection mechanisms. He also created a reflection struct, that coalesces all useful information that Cactus has about a grid hierarchy.
# Unanswered questions on the mailing list.
- No questions
# Tickets
- 2858: Lucas says he can try the machine files for Marenostrum. If they work we will add them to simfactory. - 2867: Apparently, this happens when one does not use the SILO library that is distributed with the ETK. We discussed possibilities for making it work with spack. Roland has shared the options he uses while compiling. That may help. - 2872: Yosef points out that the problem went away with a new version but that does not mean it is solved. Steve says that there is not much we can do unless it shows up in the current or newer versions.