[Users] Meeting Minutes May 20, 2021 ETK
babiuc at marshall.edu
Thu May 20 10:37:44 CDT 2021
Zach Etienne, Yosef Zlochower, Miguel Garcia, Leonardo Werneck, Ken Sible, Atul Kedia, Steve Brandt, Roland Haas, Peter Diener, Maria Hamilton
Chair: Peter Diener
Minute taker: Maria Hamilton
* pre-Agenda : nothing
* SPEC benchmark contribution: Steve no news
* Workshop update:
* Roland: deadline talks registration June 23
* ETK 2021 Workshop Webpage<https://einsteintoolkit.github.io/et2021uiuc/>
* ETK 2021 Workshop Announcement<https://hyperspace.uni-frankfurt.de/2021/05/20/north-american-einsteintoolkit-virtual-school/>
* Scientific committee - Yosef, tasked to contact speakers, Roland will provide email addresses.
* New release main features:
* POWER: Roland just pushed a new commit, added documentation, Maria will test.
* New release issues: BBH Gallery
* Steve reran BBH from Gallery, run on Deep(?) Bayou cluster, code dies with assertion failure on number of threads. Made ticket, reported problem to Erik, who said is not pssibe, his code is perfect!
* Zach suggested to try running with previous release, if yes, then revert to it
* Steve will try, and see if works, although expects is an issue with the cluster, not teh code
* Roland explained that it might come from an misunderstanding between threads and computing nodes, a problem due to hwloc1, hwloc2 conflict. He found a workaround for QueenBee and advised Steve to check the QueenBee optionlist.
* Steve mentioned that hyperthreading is disabled for most clusters, while here is enabled and threads per core = 1, thus this might be the issue.
* Other gallery examples:
* BNS: Atul NSTracker link provided
* asks what should be updated, if results identical
* Roland advises update everything
* Leo: works on documentation for VolumeIntegral and will update in the ETK arrangements.
* Maria asks if the way VolumeIntegral is set up to follow star center does depend on mesh refinement and if is there any need for AHFinder or PunctureTracker. Zach explains that it tracks the center of the quantity set up to track, for example the lapse of a BBH system, and it does not depend of regridding. As for AHFinder or PunctureTracker, it can use both.
* Zach computed constraint viloation, and set it up to avoid interior of horizon, by specifying radius and center, thus substracting off interiors, and also excising region outside some boundary like when tracking constraint violation
* Name for the release ->male -> Lorentz -> no bjection, no vote necessary
* Peter spoke up, Lorentz<https://en.wikipedia.org/wiki/Hendrik_Lorentz> it is
* Multipach Wave Equaton -> Bill travels, sent update to Zach, bumps regarding visualization, will update/improve web page, to make it easier for next person, will update ticket on Mon
* Poisson Equation: -> Miguel updated all results, will replace on the website
* Roland confirms that will be no more changes of the code after the freeze
* TOV -> Peter ran it, everything looks identical, can't use presink yet, and not uploaded yet. Did not know what was used to generate the plot, will use gnuplot and add script on web page, update source tree, rerun and add updates.
Zack: friendly reminder of crunch time, everything needs to be in place for next week, update all that needs to be updated. He will work on release announcement.
Release Process<https://docs.einsteintoolkit.org/et-docs/Release_Process> and Details<https://docs.einsteintoolkit.org/et-docs/Release_Details>
Roland reminded Zach to contact those who contributed and ask if they are willing to be listed - usually should be listed as authors if they are gallery runners
Zach - will take care of Zenodo link asks if its a new one or updated. Roland explains that it keeps track of versions, but needs token change to "become" Steve to be able to update.
Nothing else on the release
* Unanswered mailing list queries: Peter - no unanswered email found
* Roland couple of tickets, nothing to worry about
* Flang compiler Error #2529<https://bitbucket.org/einsteintoolkit/tickets/issues/2529/flang-error-in-cactus-src-util>
* Problem with Vectors compipation in gcc8, but we use gcc9
* Bug report about initialized variables, if you want some that don't exist.
* Peter Erick found Flang issue
* End of Agenda: Miguel reported about Gabriel's pull request for machine file Frontera simfactory2. Roland will look into it to be approved
* May 27 meeting: Minute taker: Bill. Chair: Atul
Maria C. Babiuc Hamilton, Ph.D.
Professor, Department of Physics
College of Science, Marshall University,
1 John Marshall Drive, Huntington, WV, 25755
Room S 257, Phone: (304)696-2754
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Users