Thiago Bastos wrote:
> Hi Gerrit,
> 
>     I'm looking at it and are about to change the refcounting to atomic
>     operations. It's just compiling right now so I will know more
>     soon ;) Hopefully you'll have a stable ref counter as you hit office
>     tomorrow morning ;)
> 
> 
> Hmmm, and what are the impacts of that change?
> 
> Clearly, atomic ref-counting is needed if you're running more than one 
> thread in the same aspect. But if it's going to make things slower for 
> programs that do not have multi-threaded aspects, I'd advocate for it 
> not making into the trunk (or at least not as a default option).

I don't think we need to worry, as there are fast cpu-specific 
instructions for this that most compilers support. They differ a bit, so 
one needs to code for different platforms/compilers etc.

On MSVC there is an (IIRC) InterlockedIncrement() intrinsic function 
(i.e. one that compiles down to a single instruction) that handles this.

No mutexes needed, which is nice. Some extra memory barriers may be 
needed, but I think that's pretty negligble (it's never been a problem 
for me in previous apps). If it's a problem, then you need to look at 
your reftptr usage pattern. :)

I don't know what Gerrit went with in his first implementation, but it's 
easy to optimize.

Cheers
/Marcus


------------------------------------------------------------------------------
Apps built with the Adobe(R) Flex(R) framework and Flex Builder(TM) are
powering Web 2.0 with engaging, cross-platform capabilities. Quickly and
easily build your RIAs with Flex Builder, the Eclipse(TM)based development
software that enables intelligent coding and step-through debugging.
Download the free 60 day trial. http://p.sf.net/sfu/www-adobe-com
_______________________________________________
Opensg-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/opensg-users

Reply via email to