[galib] memory usage of galib

kaare@ehp.dk kaare at ehp.dk
Thu Jan 22 14:57:40 EST 2004


hello all

as part of a bachelor project at DTU in denmark, we have written a program for
optimizing task graph assignments to multi-processor architechtures using
galib. this all works like a charm, only we have a serious problem on memory
usage, which increases rapidly as evolution progresses through the generation.
 
intuitively we would guess that a genetic algorithm will only use a constant
amount of memory as it maintians a population of constant size (we use fixed
size populations only), plus possible some temporary data for crossover etc.
is this a correct assumption (and in particular is this true for galib)?
 
if, as we expect, the answer to the above is yes, the error must be in our
program. does anybody have any examples of classic blunders when dealing with
galib, are there any objects one should pay special care to delete or free for
each generation or similar???

hope that someone can help, it would greatly help our project if we could find
a cure, or at least an explanation.

many thanks in advance
kåre poulsen and kasper jørgensen
copenhagen, denmark

> 
> 
> 
> ________________________________________________________________
> 
> This message was sent using Egmont Hall of Residence webmail system.
> For more information visit http://www.ehp.dk
> 



-- 






________________________________________________________________

This message was sent using Egmont Hall of Residence webmail system.
For more information visit http://www.ehp.dk



More information about the galib mailing list