Can I utilize Concurrent Kernel Execution among processes with the same context?

I want to share the Kepler GPU among several processes, so that the GPU kernels from different processes can execute concurrently on the GPU. Can I use the CUDA driver API to create a common CUDA context and then share it with all the processes? Besides context, are there any other constrains that blocks multiple processes to do concurrent kernel execution on the GPU?

Thanks.