The NVIDIA Linux Driver supports GLX when Xinerama is enabled on similar GPUs. The Xinerama extension takes multiple physical X screens (possibly spanning multiple GPUs), and binds them into one logical X screen. This allows windows to be dragged between GPUs and to span across multiple GPUs. The NVIDIA driver supports hardware accelerated OpenGL rendering across all NVIDIA GPUs when Xinerama is enabled.
To configure Xinerama
Configure multiple X screens (refer to the xorg.conf man page for details).
Enable Xinerama by adding the line
Option "Xinerama" "True"
to the "ServerFlags" section of your X config file.
Using identical GPUs is recommended. Some combinations of non-identical, but similar, GPUs are supported. If a GPU is incompatible with the rest of a Xinerama desktop then no OpenGL rendering will appear on the screens driven by that GPU. Rendering will still appear normally on screens connected to other supported GPUs. In this situation the X log file will include a message of the form:
(WW) NVIDIA(2): The GPU driving screen 2 is incompatible with the rest of (WW) NVIDIA(2): the GPUs composing the desktop. OpenGL rendering will (WW) NVIDIA(2): be disabled on screen 2.
NVIDIA's GLX implementation only supports Xinerama when physical X screen 0 is driven by the NVIDIA X driver. This is because the X.Org X server bases the visuals of the logical Xinerama X screen on the visuals of physical X screen 0.
When physical X screen 0 is not being driven by the NVIDIA X driver and Xinerama is enabled, then GLX will be disabled. If physical X screens other than screen 0 are not being driven by the NVIDIA X driver, OpenGL rendering will be disabled on them.
Only the intersection of capabilities across all GPUs will be advertised.
The maximum OpenGL viewport size is 16384x16384 pixels. If an OpenGL window is larger than the maximum viewport, regions beyond the viewport will be blank.
X configuration options that affect GLX operation (e.g.: stereo, overlays) should be set consistently across all X screens in the X server.