It is possible to deliver very intensive graphical workloads in a virtualized environment using vGPUs from NVIDIA, AMD or Intel. When moving these workloads to a centralized location most certainly will introduce network latency, since the remoting protocol needs to travel the wire. What is the effect of latency in these kinds of scenarios? This research will show the effect of moving a heavy 3D workload, like a racing simulator, to a virtualized environment.
Source: The effect of latency and RTT on 3D workloads in Microsoft Azure