Beyond C++ and Python

Session Chair

Michael Heroux (Sandia National Laboratories, USA)

Abstract

The HPC and Data Sciences communities have developed very distinct programming and workflow environments.  While Linux underpins the environments for each community, and members of one community have familiarity with the components and concepts of the other’s, the differences represent fundamental impediments to collaboration.  At the same time, learning about and adapting the strengths of each community will enable dramatic improvements for everyone.  Turnkey, cloud-based computing, commonly used in data sciences promises to revolutionize the way HPC works.  Alternatively, the effective harnessing of scalable computing accomplished by the HPC community, including a rich software environment, can enable data science computations that are presently impossible.
 
In this session, we discuss what the Data Sciences and HPC communities bring as programming and workflow strengths to the combined challenge of HPC+Data Science, the impediments that must be overcome, and the promise of convergence.

Go back to Agenda