I am disagree with new(?) rule- 3 days for counting workunit(s). Almost 50 % my jobs are canceled in last minutes, because 3 days are too few. Please change it to almost 4 days!
/ 
Krunchin-Keith [USA]Volunteer moderator
Volunteer tester
Send message
Joined: Nov 10 05
Posts: 3217
Credit: 5,500,434
RAC: 3,635
I am disagree with new(?) rule- 3 days for counting workunit(s). Almost 50 % my jobs are canceled in last minutes, because 3 days are too few. Please change it to almost 4 days!
This has been covered before as follows:
Maire wrote:
Most of the simulations running on malariacontrol.net serve the purpose of estimating parameters of the underlying mathematical models. In essence, we iteratively try a lot of different model parametrizations, and let an optimization algorithm decide which parameter set to try next. This process is more efficient if we use shorter deadlines. In other words, with the same number of computers, we can reach the same goals in shorter time if we use shorter deadlines.
If you are getting too many tasks that get canceled, you probably should lower your queue size so as many do not accumulate and so that they get turned in before deadline.
/
no... I work on two proeject: malaria control and Rosseta. It is very bad that 50 % of work on Malaria is canceled because 3 days dedalines and my low PC running time.
/
no... I work on two proeject: malaria control and Rosseta. It is very bad that 50 % of work on Malaria is canceled because 3 days dedalines and my low PC running time.
You may wish to consider another Project then, here is a link to all the active Distributed Computing Projects, the Boinc ones are noted:
http://www.distributedcomputing.info/projects.html
/
I've noticed wu CPU times (2.6 GHz Phenom, allowing 100%), from between 24 minutes to almost 8 hours. I can see a 3 day deadline being a problem if one of these monter wu's gets sent to a box not allocating so much CPU time to malariacontrol. Is there any way to smooth the work out?
/