If it is good to run jobs for 24 hours, why isn't the limit higher? I don't mind running a job for a week if it is better for the project, and some other projects have had jobs that timed out after 30 days or had time limits of 100 hours.

When the WU time settings were first setup, we could choose up to 4 days per WU. That was reduced to a max of 1 day due to a number of errors and bugs that have been eliminated and worked around since that point.

Perhaps when we move back to 14 day WUs, the option of 4 days can be returned - for those that want more than 24 hours per WU?

---

Thanks for the background, BennyRop. I don't necessarily want longer jobs (WU), but I am willing to run them longer if it helps the project.

Ack.. it looks like I could clarify a little. Right now, we have 7 days to return WUs. Prior to CASP we had 14 days to return WUs. A WU run time of 4 days may be too long for 7 day returns for anyone with a cache.. (If you have 2ea 4 day run time WUs - it'll be tough to get 8 days of crunching done in the 7 days you have until their due date.)

As for WUs that take forever and a day - the largest proteins we dealt with for CASP were also the largest we've tackled; and I'd assume that the Athlon 64 3000+ cpus (and comparable) produced a fair number of decoys in a day.

---

The main benefit to the project is just that you don't have to hit their servers as frequently to request work. The main benefit to the efficiency of your machine is just that you don't have as much overhead of downloading which takes time away from the crunching (very small amount of time). The download itself runs while you crunch a WU, but there is a small amount of processor time required to handle each download too. I've not heard of anyone actually quantifying it. I'm sure it's <1%.

---

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/