Just a heads-up about some anomalous interactions between Rosetta's target CPU run-time setting and BOINC's application switch period.

As I set Rosetta to run for 1h, it happened to be the same period that BOINC would switch applications. As Rosetta does just a best effort to run for the set time, it can under or overshoot it slightly. If overshot, BOINC would pause it and start another application, if there is any.

The fact that an application is paused has a few drawbacks, especially considering that Rosetta is almost finished. Unless Rosetta is left in memory (per BOINC's settings), restoring its last checkpoint may throw away several run-time minutes. Moreover, even when Rosetta is left in memory, instead of returning the result ASAP, it'll hang around longer, up to a few hours (I attached 26 projects), wasting disk space and delaying the scientific result from running Rosetta.

So I increased BOINC's application switch period to a little over 1h, say 65min.

A welcome side-effect was that PrimeGrid happens to run for a little under 1h on some of my systems. However, that's a little under 1h of CPU time, not wall-clock time. For the CPU I use to do anything else more intensive will slow down the completion of BOINC WUs. So I often had Rosetta's and PrimeGrid's WUs paused just shy of 100% complete. Not anymore...

HTH

---

Not sure what you are getting at. You should not be surprised after 21,000 credits that Rosetta WUs can take 2 or 3 hours for some proteins to complete a single model. And so your 1 hr time preference will often go over.

The project presently recommends a an 8hr time preference. This is set in your Rosetta preferences. If you decide to do this, make the change gradually, as BOINC will request work assuming the old time preference. So it will tend to request too much if you have a large cache.

This is discussed in more detail in this QA item.

---

Add this signature to your EMail:
Running Microsoft's "System Idle Process" will never help cure cancer, AIDS nor Alzheimer's. But running Rosetta@home just might!
https://boinc.bakerlab.org/rosetta/