Accessing Central GPU for Graphical Processing in a Network

WE have a “GPGPU” enabled Server with three “Tesla C1060” and one “Quadro NVS”. All the other Computers are connected to this Server in Star Topology through 1GBPS LAN.
We want our client machines to execute their GUI based CUDA programs on Server and the output will be rendered back to the clients. We are curreently using CentOS 5.5 (64 bits) on Server and on Client as well.
I have tried to execute my programs through ssh -X, Xnest and vnc client server apps. but they are giving an error as described in the image. All my command based programs runs well. I request ou to give some concrete Solution on this issue… Am I going wrong in the Configuration…