Re: [Swarm-Support] Re: [Swarm-Modelling] xmalloc problem

[Doug Donalson]

As an alternate, you could make your own heap(s).  Instead of releasing a
"dead" agent back to the system, put it into a list and then when you need
to add a new agent, instead of malloc, pop the list of dead agents and reuse
the memory.  Just a thought...

Cheers,

D4

----- Original Message ----- 
From: "Steve Railsback" <address@hidden>
To: "support swarm" <address@hidden>
Sent: Monday, June 14, 2004 8:33 AM
Subject: [Swarm-Support] Re: [Swarm-Modelling] xmalloc problem


> José Antonio Pascual Ruano wrote:
>
> > Hello dear swarmers,
> >
> >
> > I`m running a simulation with JavaSwarm but I have a persistent problem.
> >
> > Every time the simulation reaches around 80000 cycles the system crashes
> >
> > The error that I get is:
> >
> > /src/Swarm/swarm/src/misc/xmalloc.c:95 xmalloc
>
> I posted a question about this problem several months ago because we
> also encountered this problem. The same error occurs when you create
> lots of agents instead of running for a long time. The problem does not
> occur running the same code in Linux.
>
> I think there is deep down in Windows a limit on how many times a
> process can access another block of memory- even if the process frees
> memory at the same rate it accesses it. In fact, I  think I saw a
> Windows parameter that controls this limit but stupidly did not write
> down where it was. Perhaps someone who really knows Windows can tell us.
>
> Steve R.
>
> -- 
> Margaret Lang / Steve Railsback
> 250 California Ave.
> Arcata CA 95521
> 707 822 0453; fax 822 1868
>
> _______________________________________________
> Support mailing list
> address@hidden
> http://www.swarm.org/mailman/listinfo/support
>