Going back to 2008 it was an action item to be worked on of supporting mobile GPU switching with X.Org / Linux and there have been talks about it at other times, but it really hadn't amounted to much. David Airlie though has started playing with this switchable graphics technology on a Lenovo ThinkPad, which combines an Intel and ATI GPU. David Airlie has got this graphics switching to work, but in a rather crude form at this time. He has also published his kernel patch to allow this ACPI-based switching support to work.
David got this working in one day, but this is not true switchable graphics like can be found on Windows. With this rudimentary support, the ATI and Intel kernel mode-setting drivers initial both GPUs upon start-up and for now both GPUs continue to run. Neither of the GPUs actually shut off or are otherwise put into any reduced power state, which actually is eliminating most of the benefits of this feature. Additionally, when switching graphics the X.Org Server needs to be killed and restarted, but it's not seamless switching.
There is also not a clean path to do the switching but rather to initiate the ACPI method switch you need to change to a virtual terminal and write to the switch on the kernel's debug file-system and then change back the run-level. This work also depends on not having a xorg.conf (or one that specifies the GPU/driver) to function.
This work by David is a pure kernel patch and is not changing anything with the X.Org Server itself, since it's simply killing and restarting the display server in the process. Ultimately the X.Org Server needs to gain better support to provide a clean and seamless switching experience.
David does know how to power-down the Radeon GPU using ACPI and for the Intel GPU to potentially put it in a suspended state, but that code is not found in today's patch called the vga_switcheroo. Lastly, David is also going to lobby the GDM developers to have a log-out button for GNOME that says "Switch GPU" to provide a bit more transparent experience without having to manually change run-levels and write to the debugfs switch.
In David Airlie's blog post he adds, "I'm really hoping interested community people can make this actually useful to them on other hw, I won't have permanent access to the W500 to keep this all tested in the future."