Virtualizing General Purpose GPUs for High Performance Cloud Computing: An Application to a Fluid Simulator

In this work we present an hypervisor-independent GPU Virtualization Service named GVirtus. It instantiates virtual machines able to access to the GPU in a transparent way. GPUs allow to speed up calculations over CPUs. Therefore, virtualizing GPUs is a major trend and can be considered a revolutionary tool for HPC. To test the performances of GVirtus we used a fluid simulator. Morover to exploit the computational power of GPUs in cloud computing we virtualized three different plugins for GVirtus Framework : Cuda Runtime, Cuda Driver and OpenCL plugins. Our results show that the overhead introduced by virtualization is almost irrelevant.