Don't bother, it's not as performant as VNC (VNC overhead is higher but latency may be lower, depends whether you want to optimize for low latency and throughout versus server resource usage), especially once you take into account stuff like VirtualGL.