Notebook session questions | TFox – Dispatching calculations to slurm? JH – We’re working on dispatching local computation to QCEngine, which may be a bit better. TF: really has been nice to use the approach taken in benchmarking for use of HPC resources JH: yeah, this execution style is something we aim to support
MMackey: kernel crashed for me; on Debian testing TFox - couldn’t get git clone to work; had to download zip file instead CBannan - getting Errno Snowflake instance did not boot properly, try increasing the timeout on MacOS Catalina; restarting notebook also gives “There appear to be 5 leaked semaphore objects to clean up at shutdown”; ps aux | grep postgres after killing kernel gives nothing MMackey: from openff.toolkit.topology import Molecule kills the jupyter kernel Questions to fill time: Has everyone managed to execute the notebook? Will using Bespokefit always require a notebook to run? How much could usage be simplified/will it be wrapped in a CLI? What if I want up to k6? Could we pre-download QCA datasets and package them with the bespokefit code?
KMeier: having issues with Snowflake, colleague wasn’t, however, so doesn’t appear to be systematic to our infrastructure MMackey: would be nice if you could tell ForceBalance to not spit out color codes in the log output CBannon: wasn’t clear how long executor would run, so killed it BSwope: just watched, may give this a shot later KMeier: thank you for putting together this tutorial/workshop; will give this more a play to see if I can get around issue, colleague didn’t have it so pretty confident CBannon: will give this more a play if have time; not the highest priority in my group right now, so have to squeeze it in BSwope: very impressive work! CBannon: indeed, great stuff here!
|
