Hi Max,

On MilkyWay@home we do something very similar (using genetic algorithms).  We 
always keep a buffer of 500 workunits available, and generate new ones when the 
 buffer drops below 500.  That way there is always work available for the 
clients, and work is generated as needed.  You probably want to change the way 
your genetic algorithm works if you want to get the best results out of BOINC.  
I'd be more than happy to talk to you about this as we've got a lot of 
experience doing evolutionary algorithms over at MilkyWay@Home -- you might 
even be able to reuse our code.

--Travis


On Oct 6, 2011, at 8:28 AM, Max Dmitrichenko wrote:

> Hi!
> 
> My small project makes some kind of genetic algorithm, i.e. it
> searches for optimal parameter values. The algorithm is implemented
> in a following way: based on current best value work generator daemon
> generates new N workunits and waits until all of them are
> assimilated, then it determines the new best variant, generate another
> N workunits and so on.
> 
> The problem is: when clients submit their jobs, there are no new
> workunits available. And new communication is deferred for quite a
> long time which is enough to calculate a couple of workunits. So 2/3
> of time clients do nothing.
> 
> How can I change the communication interval? As far as I understand
> the server-side scheduler reports to the client the next time
> of communication. Is there any mean to change it?
> 
> --
> Max
> _______________________________________________
> boinc_dev mailing list
> [email protected]
> http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev
> To unsubscribe, visit the above URL and
> (near bottom of page) enter your email address.

---------------------------------------------------------------------------
Travis Desell,  Assistant Professor
University of North Dakota - Dept. of Computer Science 
[email protected] - cell: 518-867-1054
Streibel Hall Room 220 - office: 777-701-3477
3950 Campus Road Stop 9015
Grand Forks, North Dakota 52802-9015

Homepage ( http://people.cs.und.edu/~tdesell/ )
MilkyWay@Home ( http://milkyway.cs.rpi.edu/ )
DNA@Home ( http://dnahome.cs.rpi.edu/ )
Worldwide Computing Laboratory ( http://wcl.cs.rpi.edu/ )
----------------------------------------------------------------------------

_______________________________________________
boinc_dev mailing list
[email protected]
http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev
To unsubscribe, visit the above URL and
(near bottom of page) enter your email address.

Reply via email to