Accessing GPU from a kernel mode driver

Accelerated Computing CUDA CUDA Programming and Performance
1

Hello everybody,

I would like to acces the GPU from a seperate kernel mode driver in a real time application. Is this possible? if so is there a small example how to?

Thanks in advance

2

I also need the same functionality, but came to conclusion that direct use of GPU from kernel mode is unfeasible. As a workaround I transport all data for processing to user mode application and receive it back after being processed.

Anyway, I would like to know are there any alternatives.