I would like to know what is the best to do.

My computer is ALWAYS open.

I would like to know if it is better for Rosetta to run a longer(1 day) task or a shorter(3 hours) one?

Is there a difference on the RAC ?

Doesn't really affect RAC much. You gotta take into account, that if a WU fails @ 22hrs... those hrs will be lost... that's why I keep my WUs 2-3hrs...

---

At some point all the tasks are going to stop because there is a hard limit on the number of decoys allowed at 99.

*MY* paranoia says that running over about 3-4 hours, tops 6; is where you balance risk of loss of the task from crashes, bugs, moths, phases of the moon, etc. and getting the work done. Ot to put it another way, I like tasks that take about an hour to 3 hours so I can complete the task and report it and move on.

Tasks that take multiple days, or even more than 6 hours run far higher risks that you are going to invest a lot of time and if the task fails you get nothing for the effort. I have been having troubles with YoYo recently with some of their tasks "hanging" ... do I let them run for hour after hour to run into the hard limits, will they run for more time then complete, should I abort ... hard questions ... and sometimes it does happen here ...

I have chosen the 3 hour time and think that is a good balance between letting it run longer for the project, yet limiting my risks ...

YMMV

Thank you so I will put the time to 2 hours.

I would like to know what is the best to do.

My computer is ALWAYS open.

I would like to know if it is better for Rosetta to run a longer (1 day) task or a shorter (3 hours) one?

Is there a difference on the RAC ?

Someone once put up some numbers that implied RAC was slightly higher per hour for longer runs, but whether that was conclusive I don't know. Current WUs are giving lower credits than WUs from a few weeks ago so it may be within the normal variation of credits.

WU failure is an issue at times, but they tend to run within batches, so it may be a case of running more failing WUs of shorter length or less of slightly longer. It balances out and I don't think it's an issue that's affected by runtime.

Much less of an issue recently, some decoys are long-running and if one doesn't complete by runtime+3 hrs the watchdog cuts in and aborts the job, so having a short runtime may bring problems of its own.

The other factor is server load. The more often new WUs are requested due to short runtimes the more load it puts on the Rosetta servers. There was a thread here proposing that default runtime should be doubled to 6 hours to help with that.

So my view is that the default 3hr runtime is pretty good. If you can run an extra hour without errors then it helps the project with server load and it may (or may not) slightly improve your RAC. Only if you're having specific problems should you run below the default runtime.

Taking my own advice, after looking at my completed tasks (zero errors in the last week), I'm going to edge my runtime up to 4 hours from the default 3.

I would like to know what is the best to do.

My computer is ALWAYS open.

I would like to know if it is better for Rosetta to run a longer (1 day) task or a shorter (3 hours) one?

Is there a difference on the RAC ?

Someone once put up some numbers that implied RAC was slightly higher per hour for longer runs, but whether that was conclusive I don't know. Current WUs are giving lower credits than WUs from a few weeks ago so it may be within the normal variation of credits.

WU failure is an issue at times, but they tend to run within batches, so it may be a case of running more failing WUs of shorter length or less of slightly longer. It balances out and I don't think it's an issue that's affected by runtime.

Much less of an issue recently, some decoys are long-running and if one doesn't complete by runtime+3 hrs the watchdog cuts in and aborts the job, so having a short runtime may bring problems of its own.

The other factor is server load. The more often new WUs are requested due to short runtimes the more load it puts on the Rosetta servers. There was a thread here proposing that default runtime should be doubled to 6 hours to help with that.

So my view is that the default 3hr runtime is pretty good. If you can run an extra hour without errors then it helps the project with server load and it may (or may not) slightly improve your RAC. Only if you're having specific problems should you run below the default runtime.

Taking my own advice, after looking at my completed tasks (zero errors in the last week), I'm going to edge my runtime up to 4 hours from the default 3.

I have BOINC 6.10.58, running RAH exclusively. I do not see a setting for 'target CPU run time.' Where do I find it?
Does target run time affect a user's ability to predict the lowest-energy structures?

---

I would like to know what is the best to do.

My computer is ALWAYS open.

I would like to know if it is better for Rosetta to run a longer(1 day) task or a shorter(3 hours) one?

Is there a difference on the RAC ?

OK, it's kind of necrophilia to answer now on the original question, but....

In the current situation you would probably have been better of with longer runtimes, and the server is definitely better of with less connections that run alongside longer runtimes.

I've set my preferences to 20h on my 24/7 machine, and on 1h on my parents seldom running one.

Does target run time affect a user's ability to predict the lowest-energy structures?

No, it simply would allow you to crunch more models in to fewer work units. In the end, the hours of time your machine runs R@h is the best predictor of your ability to best predictor of the day. More hours yields improved chances.

---

Rosetta Moderator: Mod.Sense