Chapter 32. Offloading Graphics Display with RandR 1.4
Version 1.4 of the X Resize, Rotate, and Reflect Extension (RandR 1.4 for short) adds a way for drivers to work together so that one graphics device can display images rendered by another. This can be used on Optimus-based laptops to display a desktop rendered by an NVIDIA GPU on a screen connected to another graphics device, such as an Intel integrated graphics device or a USB-to-VGA adapter.
X.Org X server version 1.13 or higher.
A Linux kernel, version 3.13 or higher, with CONFIG_DRM enabled.
Version 1.4.0 of the xrandr command-line utility.
Using the NVIDIA Driver as a RandR 1.4 Output Source Provider
To use the NVIDIA driver as an RandR 1.4 output source provider, the X server needs to be configured to use the NVIDIA driver for its primary screen and to use the “modesetting” driver for the other graphics device. This can be achieved by placing the following in /etc/X11/xorg.conf:
Screen 0 "nvidia"
See “What is the format of a PCI Bus ID?” for information on determining the appropriate BusID string for your graphics card.
The X server does not automatically enable displays attached to the non-NVIDIA graphics device in this configuration. To do that, use the xrandr command line tool:
$ xrandr --setprovideroutputsource modesetting NVIDIA-0
$ xrandr --auto
This pair of commands can be added to your X session startup scripts, for example by putting them in $HOME/.xinitrc before running startx.
$ xrandr --listproviders
command to query the capabilities of the graphics devices. If the system requirements are met and the X server is configured correctly, there should be a provider named NVIDIA-0 with the Source Output capability and one named modesetting with the Sink Output capability. If either provider is missing or doesn't have the expected capability, check your system configuration.
Synchronized RandR 1.4 Outputs
When running against X.Org X server with video driver ABI 23 or higher, synchronization is supported with compatible drivers. At the time of writing, synchronization is compatible with the “modesetting” driver with Intel devices on Linux version 4.5 or newer. If all requirements are met, synchronization will be used automatically.
X.Org X server version 1.19 or newer is required to support synchronization. Without synchronization, displays are prone to “tearing”. See Caveats for details.
If synchronization is being used but is not desired, it can be disabled with:
$ xrandr --output
and re-enabled with:
$ xrandr --output
See Vblank syncing for information on how OpenGL applications can synchronize with sink-provided outputs.
What is the format of a PCI Bus ID?
Different tools have different formats for the PCI Bus ID of a PCI device.
The X server's "BusID" X configuration file option interprets the BusID string in the format "bus@domain:device:function" (the "@domain" portion is only needed if the PCI domain is non-zero), in decimal. More specifically,
"%d@%d:%d:%d", bus, domain, device, function
in printf(3) syntax. NVIDIA X driver logging, nvidia-xconfig, and nvidia-settings match the X configuration file BusID convention.
The lspci(8) utility, in contrast, reports the PCI BusID of a PCI device in the format "domain:bus:device.function", printing the values in hexadecimal. More specifically,
"%04x:%02x:%02x.%x", domain, bus, device, function
in printf(3) syntax. The "Bus Location" reported in the information file matches the lspci format. Also, the name of per-GPU directory in /proc/driver/nvidia/gpus is the same as the corresponding GPU's PCI BusID in lspci format.
On systems where both an integrated GPU and a PCI slot are present, setting the "BusID" option to "AXI" selects the integrated GPU. By default, not specifying this option or setting it to an empty string selects a discrete GPU if available, the integrated GPU otherwise.