Target Run CPU Time

Questions and Answers : Preferences : Target Run CPU Time

To post messages, you must log in.

AuthorMessage
boinc_qc

Send message
Joined: 12 Oct 05
Posts: 1
Credit: 230,975,310
RAC: 0
Message 25412 - Posted: 29 Aug 2006, 6:47:23 UTC

Does anyone know what this setting does ?

thanks in advance
ID: 25412 · Rating: 1 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Keck_Komputers
Avatar

Send message
Joined: 17 Sep 05
Posts: 211
Credit: 4,246,150
RAC: 0
Message 25415 - Posted: 29 Aug 2006, 7:43:48 UTC

It sets approximately how long a rosetta task will run for. After the time has expired the client will finish the current model/decoy then end and report the task.
BOINC WIKI

BOINCing since 2002/12/8
ID: 25415 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
SekeRob

Send message
Joined: 7 Sep 06
Posts: 35
Credit: 19,984
RAC: 0
Message 28865 - Posted: 3 Oct 2006, 15:33:33 UTC - in response to Message 25415.  
Last modified: 3 Oct 2006, 15:50:46 UTC

It sets approximately how long a rosetta task will run for. After the time has expired the client will finish the current model/decoy then end and report the task.


I've tried to find an answer, which the previous does not satisfy nor the FAQ.... we know we can set the WU to process according this "Target CPU run time" preference. Previously i had not set it, now i have put it to 4 hours CPU time, so it does more 'angles/models' on a single WU, for which proportionate more credit is granted. Now the more cycles are run on a single WU, the more chance of finding a very low energy one.....now the question..... How do the scientists determine which of the returned WU's crunched for say only default 3 hours identify themselves as having the potential to go much lower i.e. had it been run 4, 6, 8, 24 hours?

If i got it all wrong, fine....disregard the question.
Coelum Non Animum Mutant, Qui Trans Mare Currunt
ID: 28865 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Keck_Komputers
Avatar

Send message
Joined: 17 Sep 05
Posts: 211
Credit: 4,246,150
RAC: 0
Message 28869 - Posted: 3 Oct 2006, 20:18:24 UTC

The way I understand it each workunit is composed of a varying number of models/decoys/trajectories each of which is basically a complete and discreate unit. It does not matter to the science if 10 workunits do 100 decoys or 100 workunits do 10 decoys, either way it is still 10,000 attempts to find the best configuration.
BOINC WIKI

BOINCing since 2002/12/8
ID: 28869 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
SekeRob

Send message
Joined: 7 Sep 06
Posts: 35
Credit: 19,984
RAC: 0
Message 28893 - Posted: 4 Oct 2006, 9:42:43 UTC - in response to Message 28869.  

The way I understand it each workunit is composed of a varying number of models/decoys/trajectories each of which is basically a complete and discrete unit. It does not matter to the science if 10 workunits do 100 decoys or 100 workunits do 10 decoys, either way it is still 10,000 attempts to find the best configuration.


Don't know if these WU's are generated real time at sending, but if my prefs are set for 4 hours which can do X models in the given time frame at given CPU speed, then how does the system track for it not to duplicate models of the same WU send to someone else (if the target is to get a preset global total of models for each WU)?

Coelum Non Animum Mutant, Qui Trans Mare Currunt
ID: 28893 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
FluffyChicken
Avatar

Send message
Joined: 1 Nov 05
Posts: 1260
Credit: 369,635
RAC: 0
Message 28910 - Posted: 4 Oct 2006, 18:16:31 UTC - in response to Message 28893.  

The way I understand it each workunit is composed of a varying number of models/decoys/trajectories each of which is basically a complete and discrete unit. It does not matter to the science if 10 workunits do 100 decoys or 100 workunits do 10 decoys, either way it is still 10,000 attempts to find the best configuration.


Don't know if these WU's are generated real time at sending, but if my prefs are set for 4 hours which can do X models in the given time frame at given CPU speed, then how does the system track for it not to duplicate models of the same WU send to someone else (if the target is to get a preset global total of models for each WU)?

Random seeds are set by the the servers.
though we are going way beyond the intial question.

It is really just a bandwidth saver if you run it longer than the default, well probably 8hrs to 24hrs. Though that is not a suitable length for people that don't have their computers on for very long where a shorted run time is beneficial, they get the results sent back before it goes over the deadline.
Team mauisun.org
ID: 28910 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Mod.Sense
Volunteer moderator

Send message
Joined: 22 Aug 06
Posts: 4018
Credit: 0
RAC: 0
Message 29275 - Posted: 13 Oct 2006, 3:37:37 UTC

The runtime preference doesn't make your models better. It just makes more of them. So the science about how to produce a good model from a given random point is the same either way.
Rosetta Moderator: Mod.Sense
ID: 29275 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Keith T.
Avatar

Send message
Joined: 1 Mar 07
Posts: 58
Credit: 34,135
RAC: 0
Message 40401 - Posted: 6 May 2007, 7:24:04 UTC

Hi, I hope it's OK to re-open this old thread.

I'm fairly new to Rosetta, I'm running it a lot more at the moment while SETI is out of work.

I currently have my run time set to to the max 24 hours, is this optimal?

I have an AMD Athlon XP 2200+ with 256MB RAM, running Windows XP Pro SP2.

I also have a Pentium II 233MHz with 512MB RAM running Windows 98SE which I know is below the minimum reccomended spec, but it does manage to do some work sucessfully. I would not mind this machine running each task for >24 hours if it is possible.

So in summary, what is the optimal run time per task, and is there a way to go beyond 24 hours?
ID: 40401 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile anders n

Send message
Joined: 19 Sep 05
Posts: 403
Credit: 537,991
RAC: 0
Message 40419 - Posted: 6 May 2007, 12:17:00 UTC

Hi Keith T

For the moment 24H is max on run time.

The servers get less trafic the longer runtine we users select so in that way
your in top of the class :)

Your PII should do just fine with a long runtime, slow computers have trouble staying within runtime pref. if you have 1 or 2 H setting.

Anders n
ID: 40419 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote

Questions and Answers : Preferences : Target Run CPU Time



©2024 University of Washington
https://www.bakerlab.org