Hi all, I have a question

What is better for Rosetta: set the highest WU runtime preference or a runtime of about 6 hours??

To clarify, I have set my WU runtime preference to 6 hours but I can set it to 12 hours or more if it is better for Rosetta, my conexion is bandwith and without restrictions

Thanks and best regards
Ricardo

---

Hi Ricardo.

It depends on two things; Machine speed and Report Deadline of the WU's

I have a Dell D3000 2.8G P4 with 1G dual channel ram. So I usually see how many model runs I can get per WU (This varies with proien size and protocol). Then ask yourself "How many models do I wish to run per WU?"

With a target CPU time of 6 hours I usually average 12 models per BIG protien and about 25 per small protien.

If you are running a slower machine then be sure that you give it enough time to run more than three models per WU.

Be careful about your work que. This is controlled by Rosetta Setting "Connect To Server About Every ? Days.

One day will give you 4 WU's. This is so you don't get covered so deep that you start missing deadlines (REDUCED TO ONE WEEK DUE TO CASP 7).

Hope this Helps.

---

"Connect To Server About Every ? Days." is actually a General Preference, not Rosetta.

The preference really gives the user the choice. The project science is in models processed. Whether you process them on 4 WUs per day or 1 really doesn't make much difference... so why not just do 1? I like the 24hr runtimes because it keeps the task list short and reduces the load on their server and my ISP.

The current recommendation is 8hrs. But that's a broad brush statement. If your machine is on 24hrs a day, then that's 3WUs per day... if your machine is one 4 hours each evening... that's two days... so if you run 24hrs a day, going longer than the 8hr guideline is exactly why they give us the choice.

The only exception would be if you happen to be one of the very few that run in to any problems when running for longer periods of time. Obviously if it should introduce any problem for you, then post about the problem and point out what your runtime preference is set to, and see if the answer comes back that a shorter runtime would be a workaround until they get it fixed. ...even if a problem occurs, the successful models are reported and significant to the project, and you get credit for all the time you spent (in a daily credit grant to errored WU reports).

So, set as you wish. But heed Keith's caution, if you have many WUs downloaded already or a larger cache, just beware that BOINC is going to download work assuming your old runtime, and the actual crunch will take the NEW runtime. So, best to make such a change when your cache preference is small and the number of existing WUs is small.

---

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/

Hi Ricardo.

It depends on two things; Machine speed and Report Deadline of the WU's

I have a Dell D3000 2.8G P4 with 1G dual channel ram. So I usually see how many model runs I can get per WU (This varies with proien size and protocol). Then ask yourself "How many models do I wish to run per WU?"

With a target CPU time of 6 hours I usually average 12 models per BIG protien and about 25 per small protien.

If you are running a slower machine then be sure that you give it enough time to run more than three models per WU.

Be careful about your work que. This is controlled by Rosetta Setting "Connect To Server About Every ? Days.

One day will give you 4 WU's. This is so you don't get covered so deep that you start missing deadlines (REDUCED TO ONE WEEK DUE TO CASP 7).

Hope this Helps.

Thanks for the answer Keith. I have 3 computers running Rosetta, 2 Celeron at 2,66G each and 1 P4 at 2,4G the two celeron are working for 6 hours per day and the P4 (at home) can run 12 hours or a bit more per day, its not a matter of deadlines, i am reporting 1 or 2 wu per day on each computer and each computer have only 2 wu (one running and the other waiting), my question is what is better for Rosetta (more models per protein or if its indiferent) or is this setting only to suit volunteers demand

Regards,
Ricardo

P.S. Sorry for my poor English

---

"Connect To Server About Every ? Days." is actually a General Preference, not Rosetta.

The preference really gives the user the choice. The project science is in models processed. Whether you process them on 4 WUs per day or 1 really doesn't make much difference... so why not just do 1? I like the 24hr runtimes because it keeps the task list short and reduces the load on their server and my ISP.

The current recommendation is 8hrs. But that's a broad brush statement. If your machine is on 24hrs a day, then that's 3WUs per day... if your machine is one 4 hours each evening... that's two days... so if you run 24hrs a day, going longer than the 8hr guideline is exactly why they give us the choice.

The only exception would be if you happen to be one of the very few that run in to any problems when running for longer periods of time. Obviously if it should introduce any problem for you, then post about the problem and point out what your runtime preference is set to, and see if the answer comes back that a shorter runtime would be a workaround until they get it fixed. ...even if a problem occurs, the successful models are reported and significant to the project, and you get credit for all the time you spent (in a daily credit grant to errored WU reports).

So, set as you wish. But heed Keith's caution, if you have many WUs downloaded already or a larger cache, just beware that BOINC is going to download work assuming your old runtime, and the actual crunch will take the NEW runtime. So, best to make such a change when your cache preference is small and the number of existing WUs is small.

Thanks Feet1st, i was writing my post when you posted yours, thanks again, i have it a bit more clear

Regards
Ricardo

---

...just for your info... you can actually edit your post once you see something like that happen.

---

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/