Set up CUDA and MPI in CentOS 6.3 / RHEL 6.3

I have installed the CUDA 5.0 (full pack), ie, with driver, toolkit and samples in two machines. Operating system on the two machines are CentOS 6.3 x86_64. I installed OpenMPI 1.6.2 64 bit version in the two machines.

Machine Configuration:
Machine 1:
OS:CentOS 6.3
CPU: Core i7
RAM: 8 GB
GPU 1: Tesla C1060
GPU 2: GTX 260

Machine 2:
OS:CentOS 6.3
CPU: Core i5
RAM:16 GB
GPU 1: GT 440
GPU 2: GT 440

I want to setup a simple CUDA distributed system using these two machine. How i can run a CUDA program using Open MPI in these two machines.

Please help. I am very new to the LINUX using.
What are the basic steps done for using CUDA programs in openMPI.