KGPU is a project that aims to port GPU capabilities directly to the linux kernel. Utilizing CUDA or OpenCL, userspace application can offload calculations to the GPU card, however KGPU extends this now transparently to the kernel.
As stated by the project’s site
The idea behind KGPU is to treat the GPU as a computing co-processor for the operating system, enabling data-parallel computation inside the Linux kernel. This allows us to use SIMD (or SIMT in CUDA) style code to accelerate Linux kernel functionality, and to bring new functionality formerly considered too compute intensive into the kernel. Simply put, KGPU enables vector computing for the kernel.
KGPU is currently developed by the university of Utah and sponsored by NVIDIA.