[Clam-devel] Sample by Sample Processing in Network Editor

Pau Arumi parumi at iua.upf.edu
Thu Dec 21 10:33:21 PST 2006


En/na Xavier Amatriain ha escrit:
> After many successes using NetworkEditor for sample by sample 
> processing I have finally come to face its limitations. As soon as I
> have a complex network it becomes impossible to have any efficient 
> execution. There are two main problems:
>
> 1. It seems that the redrawing routines for the NetworkCanvas are 
> called at every Do from the QApplication::exec. Although I understand 
> that is necessary for Monitors this
> is an overhead that is not acceptable when doing sample by sample 
> processing and calling zillions of Do's.
>
> 2. (And more important) The overhead introduced by region management 
> is also unacceptable and many times even more costly than the
> execution of the Do itself. I have done a short spike and I think I 
> can optimize many of these functions by creating partial 
> specializations of
> some template functions when Token=TData. But I am not sure if that 
> can even solve the issue.
>
> Any thoughts?
>   
my first thought is that we need to integrate this 
development/experiments into subversion (you have flow-related changes 
in local, right?)
xavi, what is your "delta" with svn?

1. redrawing methods depend on timers (since NetEdit4) and it's not done 
on Do basis. what are the synthoms?

2. i think it could be more efficient if we get rid of 
different-port-size matching. but more in a policy way than a TData 
specialiation. but lets analyse first the situation.
are you compiling in release? all regions related checkings are done (or 
should) only in debug and they are expensive.  apart from checkings what 
operations of produce() and consume() are expensive? have you done any 
profiling?

-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.





More information about the clam-devel mailing list