Compiling the CUDA_MPI sample

I have got an error when compiling the CUDA_MPI sample program using visual studio 2010, and when I used the wmpiexec.exe I got that error:

Running on 1 nodes
CUDA error calling "cudaMalloc((void **)&deviceInputData, dataSize * sizeof(float))", code is 38
application called MPI_Abort(MPI_COMM_WORLD, 38) - process 0
Test FAILED

job aborted:
rank: node: exit code[: error message]
0: Heba-PC: 38: process 0 exited without calling finalize

any clue??