As you saw on my last post, I have one GeForce 9600M GT and an IGP 9400M. This last one have one advantage, the Pinner Mapped Memory.
Pinned Mapped Memory is host memory that is Mapped into space on both host CPU and software, and GPU, enabling them to exchange data asynchronously while they are both working, without interruption or need to finish execution of a kernel.
For example in a Chess Engine, the CUDA kernel could send back real-time informations to the host application, to be displayed in real-time (such as main continuation, search-depth, evaluation, node counts, …), on the other side the host could send back informations to the kernel without interrupting it, and even enable kernel-kernel communication between different GPU!!!
This feature is shared among MCP79 (9300M/9400M) and G200 (GTX260/275/280/285 & 295) and is a great one to ease communication between host and kernel.
