54. Accessing GPUs from Containers in HPC
Authors: Lucas Benedicic (Swiss National Supercomputing Center)Miguel Gila (Swiss National Supercomputing Center)
Abstract: Thanks to the significant popularity gained lately by Docker, the HPC community has recently started exploring container technology and the potential benefits its use would bring to the users of supercomputing systems like the Cray XC30 Piz Daint of the Swiss National Supercomputing Centre (CSCS). Working in close collaboration with NERSC and an engineering team at Nvidia, CSCS is extending the Shifter framework in order to enable seamless GPU access from containers.
We explore the feasibility of using Shifter as a container framework for deploying applications using GPU devices with practically no overhead, thus achieving native execution performance. The tested examples include two GPU programming environments, namely CUDA and OpenCL.
Two-page extended abstract: pdf