Render Device selection and information

I was wondering if there is anyway to select the
Rendering device. I know that this isnt included
in OpenGL (alteast not yet), but ive heard that
you can do this through the lower layer
(for example GLX in Linux).
I havent read the entire OpenGL
Redbook but in the first chapters it sez that the
OGL program doesnt have to be on the same machine
as the rendering device (like X11).
What i want to do is query what devices there are
locally and maybe on some other machine through
the network and get specs like amount of graphics
RAM, GPU proccessing power et.c. and then select
the device i want. Im currently developing for
Win32 and Linux, but it would be nice to port my
code for MAC, FreeBSD et.c. in the future.

Cheers // Alex

Originally posted by lostsoul:
I was wondering if there is anyway to select the Rendering device. I know that this isnt included in OpenGL (alteast not yet), but ive heard that you can do this through the lower layer
(for example GLX in Linux).
I would be happy if someone could prove me wrong but I fear there’s no way (at least on common hardware and drivers). What I’ve read was about using multiple renderers for multiple monitors and rumors were that only the AGP renderer will be avaiable.
Actually, the discussion was much longer but it boiled down to that. Try search this forum, maybe you’ll find it.
I think the same problem applies to what you’re doing.
Originally posted by lostsoul:
I havent read the entire OpenGL Redbook but in the first chapters it sez that the OGL program doesnt have to be on the same machine as the rendering device (like X11).
What i want to do is query what devices there are
locally and maybe on some other machine through
the network and get specs like amount of graphics
RAM, GPU proccessing power et.c. and then select
the device i want. Im currently developing for
Win32 and Linux, but it would be nice to port my
code for MAC, FreeBSD et.c. in the future.
Cheers // Alex
Sure you can’t do that using WGL/GLX. Doing this remotely is complicated at least, if possible at all. Even on the local machine, the most OS-portable way to get CPU infos is by using CPUID. Executing this on a remote server would get you some nice informations… provided it’s x86.
Getting informations about the GPU is utterly overcomplicated. Theorically, /dev/proc is what you would like to use but sometimes this filesystem is not there. On win32, it does not exists at all (even though the registry can do the same thing in most cases).
I feel this is all but related to GL.