I have not looked into the code in too much depth such a check could either
be cached at context creation (besides the result won't change during the
lifetime of the context) or you could share the glcontext and use the new
extension mechanism if OpenGL version is greater than 3.0. Or you could
pro
This is why we can't have nice things. :) You're absolutely right.
The overhead is easily measurable by running glxspheres with VirtualGL
on the local display. Now I remember why I designed the system to share
the app context in the first place.
I checked in a new patch that creates the re
Hi, thanks for looking into this.
I think it will work. I don't know the specifics of the fix (I should
proobably download the svn version) though if a readback context is created
every frame it could hamper performance. If it's only created once, it
should work quite nicely.
--
Blerg. Never mind. I can reproduce the issue quite readily with
VirtualGL 2.3.1. I wasn't able to reproduce it with the current SVN
trunk because a fix I just checked in for another problem fixed this as
well.
Basically, VirtualGL's readback function was trying to share the OpenGL
context e
I believe you, but can you show me how to reproduce this? I would think
that running a test program such as:
http://www.opengl.org/wiki/Tutorial:_OpenGL_3.0_Context_Creation_%28GLX%29
with VGL_READBACK=pbo would repro it, but no such luck.
Reproducing the problem is important, because otherwis
