- Present: Steve, Helvi, Peter, Leo, Cheng-Hsin, Keith, Roland, Yosef, Zach, Johnny - seminar proposal: Adam J. Peterson, "Code generation for AMReX with applications to numerical relativity", 2301.08354 https://arxiv.org/abs/2301.08354, which uses AMReX for Z4c evolution - March 09th, recording available on seminar https://www.einsteintoolkit.org/seminars/ page and youtube - Unanswered questions on mailing list - email http://lists.einsteintoolkit.org/pipermail/users/2023-February/008847.html from Roland: nobody used Anvil before - [SB] Another email regarding running on Rome powered supercomputer: [RH] Delta is powered by Rome, Steve or Roland will ask the user to provide the files for compiling (optionlists, etc.) and send this email to the mailing list - New GRMHD library: take IllinoisGRMHD and NRPyEOS and make it into a library, [LW] No need to review NRPyEOS and NRPyLeakage, will need reviewers for other parts. Terrence will review particle code, Peter can review NPSolverSX (?), [HW] have reviewers and champions in contact - [RH] People who agree to test the cluster, Steven Brandt - deep bayou, Peter - sunrise, Roland - stampede2, expanse, Summit, Frontera, [LW] add detail to the release - Open tickets - SummationByParts https://bitbucket.org/einsteintoolkit/tickets/issues/2700/summationbyparts-does-not-ensure-unique: [RH] prefix needs to be fix to not pollute namespace, fix name conflict when Fortran code add underscore at the end, warning: might break something in Llama - Mac os Ventura https://bitbucket.org/einsteintoolkit/tickets/issues/2699/etk-fails-on-mac-os-ventura: [RH] duplicate from the mailing list, could be listed as solved - Compile on the M1 system https://bitbucket.org/einsteintoolkit/tickets/issues/2682/updates-to-compile-et-on-macos-m1-systems [RH] PAPI needs significant changes, HomeBrew changes the links - Slab fails https://bitbucket.org/einsteintoolkit/tickets/issues/2683/slab-fails-to-compile-without-mpi [RH] Slab claim to optionally use MPI and Kranc tutorial did without it, but it actually does use MPI so a pull request is made to fix this - Include NRPyLeakage https://bitbucket.org/einsteintoolkit/tickets/issues/2689/include-nrpyleakage-in-next-et-release [LW]: will be included for the next release - EHFinder parfile https://bitbucket.org/einsteintoolkit/tickets/issues/2698/misner_22_80_3dpar-in-ehfinder-par-cannot: [PD] will look into it - [YZ] Large ET jobs on Frontera, startup overwhelming the system, every MPI proc starts reading files. Proposed solution: at a checkpoint, only a group of MPI procs to open files, workaround by staggering when the job starts, issue might also occur with CarpetX. [RH] Assign token barrier to MPI. Roland can send the token to Yosef. Frontera should fix their file system. - Next week: Chair: Zach Minute: Roland