[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [ns] random variables Vs simulation time



Yuri, Haobo, ns-users,

	thanks for the suggestions. I have the idea that it may be linked to the 
fact that I am using an Intel-Linux machine. I know scheduler has a numerical 
instability problem, and that there is numerical problems with Intel machines. 
Could it increase cpu utilization or something?
	
> X-Authentication-Warning: who.isi.edu: yuri set sender to [email protected] 
using -f
> To: Tarik Alj <[email protected]>
> Cc: [email protected]
> Subject: Re: [ns] random variables Vs simulation time
> From: Yuri Pryadkin <[email protected]>
> Date: 16 Mar 2000 16:43:46 -0800
> User-Agent: Gnus/5.0803 (Gnus v5.8.3) Emacs/20.3
> MIME-Version: 1.0
> 
> Do you have many upcalls to tcl?
> 
> Tarik Alj <[email protected]> writes:
> 
> > Yes, I was using uniform random variables with the min_ and max_ set to 
coherent 
> > values. Plus I was getting the "scheduler going backward in time" message. I 
was 
> > using scheduler.cc v1.42, after I changed to scheduler.cc v1.46 the problem 
> > disappeared but the simulation kept taking forever to complete.
> > 
> > > 
> > > Did you check your random packet size? Some ranvars may generate huge
> > > packets. Just a guess.
> > > 
> > > - Haobo
> > > 
> > > On Thu, 16 Mar 2000, Tarik Alj wrote:
> > > 
> > > > Hi everybody,
> > > > 
> > > > 	in my scripts I have been using a proc that uses 2 random 
variables to 
> > > > set random packet sizes and rate for cbr sources. About 20 CBR sources 
are 
> > > > generated like this, the topology is fairly small (5 nodes), and the 
> > simulation 
> > > > time is 3000 seconds.
> > > > 	It takes about 2 hours to complete, Vs a few minutes without 
using 
> > > > random variables. Has anybody experienced the same kind of behaviour? 
Does 
> > it 
> > > > have anything to do with the scheduler? 
> > > > 	
> > > > 	Answers, suggestions, pointers would be greatly appreciated.
> > > > 	
> > > > 	Thanks,
> > > > 
> > > > 
> > > > Tarik
> > > > 
> > > > 
> > > > 
> > 
> > Tarik

Tarik