What is the science difference for allowing 24 hours vs say 4 hours per workunit.

What is more beneficial to the project: crunching 1 WU for 24 hours or 4x6 hour WU in the same time?

What is the science difference for allowing 24 hours vs say 4 hours per workunit.

What is more beneficial to the project: crunching 1 WU for 24 hours or 4x6 hour WU in the same time?

same result for the project each task contains a number (1+) of models - the longer the run, the more models contained. If you're machine runs 24/7 then you might want to increase the run-time to reduce bandwidth use (although feet1st pointed out this isn't always the case), but longer run times mean you risk losing more work if the task errors out (which doesn't happen (to me) often, fingers crossed!)

HTH
Danny

---

What is the science difference for allowing 24 hours vs say 4 hours per workunit.

What is more beneficial to the project: crunching 1 WU for 24 hours or 4x6 hour WU in the same time?

same result for the project each task contains a number (1+) of models - the longer the run, the more models contained. If you're machine runs 24/7 then you might want to increase the run-time to reduce bandwidth use (although feet1st pointed out this isn't always the case), but longer run times mean you risk losing more work if the task errors out (which doesn't happen (to me) often, fingers crossed!)

HTH
Danny

If you already downloaded the models and have them on your computer and then change the time on the website and update, the lengths for these models changes as well. So I am thinking there is something else here?

What is the science difference for allowing 24 hours vs say 4 hours per workunit.

What is more beneficial to the project: crunching 1 WU for 24 hours or 4x6 hour WU in the same time?

same result for the project each task contains a number (1+) of models - the longer the run, the more models contained. If you're machine runs 24/7 then you might want to increase the run-time to reduce bandwidth use (although feet1st pointed out this isn't always the case), but longer run times mean you risk losing more work if the task errors out (which doesn't happen (to me) often, fingers crossed!)

HTH
Danny

If you already downloaded the models and have them on your computer and then change the time on the website and update, the lengths for these models changes as well. So I am thinking there is something else here?

the length of the models doesn't change - the number of models within a task changes, and it's tasks (work units) that you see in BOINC manager.

---

When a task is sent to your computer, it is a lot like sending you a city map and asking you to search for routes from here to there. There are many many many possible routes. It will take a person a few minutes to plan and document each route.

Now, what would be better for a mapping project?? To plan routes for 3 hours on 8 different maps? Or to plan routes on one map for 24 hours? That just depends on how many route calculations are desired on each map right? And the project needs more models computed on the task or it would not have been sent to you.

So, for the science, the question is the number of hours of computing done per day, and the models produced with that computing. Not the number of tasks.

---

Rosetta Moderator: Mod.Sense