Memory management among multiple GPUs 9800 GX2 Shared Device Memory


I have a GeForce 9800 GX2 which has 2 GPUS of 512MB RAM each. Is there some CUDA utility to deal with sharing memory between these 2 GPU devices (not in global memory)? For example, is there some (fast) way to access an array in GPU2 device memory from a kernel running in GPU1.

Thanks in advance.

Nope, they have all the normal limitations of two distinct cards.

Thanks for the prompt response.