Den 09.11.2014 18:18, skrev Sergey Kurdakov:
Hi Sergey,
it's difficult to figure out while you want exactly compute shaders,
but with OpenCL you might have several OpenCL threads - one which will
communicate with OpenGl and another one which will perform independent
computations
yes, OpenCL is an option and if Compute Shaders really have limitations
compared to OpenCL I will need to go this way. Last time I have touched
OpenCL - some years ago - it was simply not ready.
Since Compute Shaders and Shader Storage Buffer Objects came with OpenGl
4.3, i think that is what I need . App development can be done by using
GLSL syntax (version 430) and so I hope that I can bypass all the
OpenCL stuff, simply base everything on GL 4.3 or GL ES 3.1. After
doing some successful tests (adding simple SSBO support to my OSG ) it
solved my computation taskes, so I'm absolutely positive to use
ComputeShaders instead of OpenCL. Right now I see no limitation , OpenGL
4.3 and the nvidia drivers seem to support everything. So I just wonder
App's like this can be done with default OSG implementation. I see that
some GL 4.3 features are not known to OSG right now, but the main
question is how to get powerful, straight computation and visual
rendering under one Hat.
I can try to play around with 2 viewers with shared scene data between
them. Maybe this will simply work, just to call one the "ComputeViewer"
? One for the computation, the other one for visual rendering.
But else, there is no better, "builtin" way to run a "osg computation
thread" over the scene in parallel to the viewers renderthreads?
Another related question is if I need to do proper use of
glMemoryBarrier() to make sure everything is in sync ?
Thanks and regards, Markus
_______________________________________________
osg-users mailing list
[email protected]
http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
