iCNS Distinguished Lecture: Evolving Programming Models for Massively Scalable Scientific Applications

ABSTRACT

Modern scientific discovery depends on the development of large scale applications to extract meaning from the petabytes of data produced by particle accelerators, genome sequencers, astronomical observatories and other large scale experiments. These applications must sophisticated enough to scale up to thousands of nodes on HPC clusters, but at the same time, be easy enough for a single researcher to design, operate, and troubleshoot without a staff of experts on hand. In this talk, I will describe our experience in developing the TaskVine workflow system and its evolution to meet the needs of ambitious researchers in high energy physics. By collaborating closely with our active user community, we have gradually developed effective programming abstractions that yield the right blend of performance, expressiveness, and reliability.

BIOGRAPHY

Prof. Douglas Thain is a Professor in the Department of Computer Science and Engineering at the University of Notre Dame. He received the Ph.D. in Computer Sciences from the University of Wisconsin in 2004 and the B.S. in Physics from the University of Minnesota in 1997.

He directs the Cooperative Computing Lab at Notre Dame, which focuses on the design of large scale distributed computing systems applied to grand challenge problems in science and engineering. His team is active in publishing open source software systems such as Floabilitya>, TaskVine Makeflow, Parrot, and others.

Prof. Thain has received multiple teaching awards at Notre Dame for his courses in distributed systems, operating systems, and compilers, and has published an introductory compilers textbook (compilerbook.org), an open source operating system kernel (Basekernel), and an online course in Data Intensive Scientific Computing. (DISC)