Re: [Algorithms] General purpose task parallel threading approach

[<asy...@gm...>]

>But the fact you are stalling at all, is the bit that concerns me.
What stalls are you talking about ? On the contrary I'm trying to eliminate
all the cases where cpu has to wait on something.

>but it seems to >me< that you are attacking the problem of multiple cores
in totally the wrong way.
Explain. If I make things work at the same speed as a standard approach
(when there is no or very little contention) or faster (in case of
contention or blocking io), what is wrong here ?

Could you explain a bit more about your approach ? And what do you do when
physics iteration finishes faster than rendering iteration or the other way
around ?

Alexander.

2009/4/4 Oscar Forth <os...@tr...>

> Do you have a concrete example which bothers you (maybe I could provide a
>> better explanation of how it'll work on a concrete case) ?
>>
>
> Well I don't have a concrete example.  But the fact you are stalling at
> all, is the bit that concerns me.  I appreciate using a double buffered
> approach leads to latency but it makes much more sense to me especially as
> far as scaling to many cores goes.
>
> Thanks to using a display list style system for DirectX I'm also doing
> rendering from multiple threads.  Essentially each "task" decides what it
> wants to do ... sends its forces to the physics engine and then renders the
> previous frame's physics results.  This, of course, allows it to respond to
> things like walking into a wall before straight away and make alternate
> "decisions".
>
> This way I'm avoiding stalls except at the end of a frame when I
> synchronise everything.  I guess I could probably completely de-couple
> rendering from logic but I don't see the point of rendering at 300fps when
> you can only update the logic at 5fps.
>
> Anyway ... thats just my approach.  I'm not saying you are wrong (My
> knowledge is limited enough without telling other people they are wrong ;))
> but it seems to >me< that you are attacking the problem of multiple cores in
> totally the wrong way.
>
>
> ------------------------------------------------------------------------------
>
> _______________________________________________
> GDAlgorithms-list mailing list
> GDA...@li...
> https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list
> Archives:
> http://sourceforge.net/mailarchive/forum.php?forum_name=gdalgorithms-list
>



-- 
Regards,
Alexander Karnakov