Worldwide Trends in Computer Architectures for Data Science - Jeff Zais.pdf (1.34 MB)
Worldwide Trends in Computer Architectures for Data Science
presentationposted on 2020-03-10, 04:00 authored by Jeff Zais
High performance computing architectures continue to evolve along several dimensions. These changes are driven by the demand for more complex simulations and the ability to create, handle, and analyse ever growing volumes of data.
This paper will focus on the state of the art in computer architectures designed to server large academic research communities in countries around the world. Prominent examples will include NCI (Australia), LRZ (Germany), and SciNet (Canada).
Besides these examples that are in place, trends in technology will be summarized, to show what can reasonably be expected in the next five years. This will include expected advances in many of the key areas of computer architecture, including processors, memory, networking, and storage. Particular emphasis will be placed on the rapidly evolving area of storage technology.
ABOUT THE AUTHOR(S)
Jeff Zais recently joined NeSI and NIWA as the Senior High Performance Computing Architect and Science Advisor. His academic background includes a B.S. degree from the University of Wisconsin, and M.S. and Ph.D. degrees from Stanford University in Aerospace Engineering. Professional experience includes technical and management roles at Ford Aerospace, Cray Research, IBM, and Lenovo, focused on application performance and system architecture.