Hi Guys, One multithreaded issue I am aware of is osgShadow::ShadowMap::cull method. Its definitely not thread safe. It has no critical section locks. I can imagine a scenario when two views cull threads execute concurrently and mess up _texgen matrix. Other similar scenarios are also possible I am afraid.
I noticed that all examples using FBOs were problematic when used with multimonitor/multithreaded Windows. As I said earlier I would blame drivers first but I would be good to further narrow/isolate the cases when the problems appear, so that at least some workaround could be devised. I have spent few days trying to track these problems in FBO and RenderStage but couldn't get to some unambigous conclusions. What may be interesting some examples (osgprerender) were OK when used with render to framebufffer camera. Cheers, Wojtek -----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Behalf Of Robert Osfield Sent: Thursday, April 17, 2008 8:50 PM To: OpenSceneGraph Users Subject: Re: [osg-users] When must you use SingleThreaded threading model Hi J-S, On Thu, Apr 17, 2008 at 7:36 PM, Jean-Sébastien Guay <[EMAIL PROTECTED]> wrote: > No solution was forthcoming when we investigated this, and the thread > kind of died for lack of time from both of us. I think going through the > examples and removing the explicit setting of SingleThreaded threading > mode where it makes sense might be a good way of generating interest in > this issue and eventually making progress towards a solution. > > What do you think? I think I need a nice period of calm so I can walk through the code and work out all the different possible failure modes and spot ways of plugging them. FYI, I do have RTT and multi-threading working at this end, under linux, but I have seen problems with osgShadow's RTT side of things so I believe its not all RTT techniques, just certain ones. The low level buffer support in osg::Camera and RenderStage is all down with multi-threading and multiple contexts in mind, but this is only part of the story, the high level code itself still needs to be aware of the issues of multi-threading and multiple contexts, my guess its the later which is going astray in some instances. I have a window of this week and and next week for making OSG-2.4 so I'm going to go for this, and once we get into May and beyond life at my end should start calming down (hopefully) so I'll be able to tend to items like multi-threading in osgShadow. Robert. _______________________________________________ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org _______________________________________________ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
