NVIDIA compute utilities

 This package provides utility binaries for paraller general purpose
 computing use cases with the NVIDIA driver.
 .
 CUDA MPS is a feature that allows multiple CUDA processes to share a single
 GPU context. CUDA MPS should be transparent to CUDA programs.
 .
 CUDA MPS requires a device that supports Unified Virtual Address (UVA) and
 has compute capability SM 3.5 or higher.
 Pre-CUDA 4.0 APIs are not supported under CUDA MPS.