2020-11-10 QCFractal User Group Meeting notes

Updates from MolSSI

• BP – Database migration tentatively thursday – Will probably be down all day.

  • Will require managers to go down during migration

  • BP will post about downtime in OpenFF’s #qcfractal channel with @channel announcement.

    • Announcement should say it’s a 24-hour outage, and BP can say it’s up sooner if tings go quickly.

• BP – New QCF release tomorrow (weds Nov 11), QCPortal won’t need to be updated.

• BP – Center where servers are hosted warned us about possible migration – Likely won’t come up for ~months

Queue Status

• Queue has 2+ weeks of work remaining

• Managers – about 1500 cores on PRP, lots on UCI HPC3 and Lilac

• JM – What’s the general process for submitting datasets

  • DD – Eg:

  • JW – Could have dataset submission session at next hackathon

User Questions

• HJ – Could I run a torsiondrive using one basis set/method, and then re-evaluate the optimized geometries with another basis set/method? This would save a huge amount of computational time.

  • JH – Could take a “default” scan, and start a constrained optimization from each of those final geometries.

  • DD – We can assign the benchmarking jobs high priority when submitted. The already existing ones can be hurried by increasing the number of services on the server, and there are some other techniques we can use to hurry them further.

  • DD – BP, could we up existing number of services from 900 to 1500?

•