A GPU (or a cluster of GPUs) might be able to process, say, 10,000 frames in one second. This does not mean that the same GPUs can process one frame in (1/10,000) of a second.
Even with an infinite number of parallel GPUs, there will be an amount of latency required in copying memory to the GPU, running a job, and copying it back. After the frame is compressed, sent over the network, and picked up by the client, further delay (possibly tens of milliseconds) is added on before pixels appear on the screen.
Even with an infinite number of parallel GPUs, there will be an amount of latency required in copying memory to the GPU, running a job, and copying it back. After the frame is compressed, sent over the network, and picked up by the client, further delay (possibly tens of milliseconds) is added on before pixels appear on the screen.
See the discussion around John Carmack's superuser post: http://superuser.com/questions/419070/transatlantic-ping-fas...