Sign up for our newsletter and get the latest HPC news and analysis.
Send me information from insideHPC:

John Gustafson to host BoF on Posit Arithmetic at SC17

Dr. John L. Gustafson is currently a Professor at National University of Singapore and Visiting Scientist at A*STAR Computational Resource Centre.

John Gustafson from A*STAR will host a BoF on Posit arithmetic at SC17. Entitled, “Improving Numerical Computation with Practical Tools and Novel Computer Arithmetic,”  this BOF will be co-hosted by Mike Lam with discussions on tools for measuring floating point accuracy.

Efficient use of computer arithmetic is crucial to HPC. Unfortunately, few software analysis tools exist to analyze floating-point accuracy or to recommend robust mixed-precision modifications for full-sized applications. At the same time, next-generation arithmetic formats such as Gustafson’s latest Type III unums (posits and valids) offer the possibility of better accuracy, performance, energy efficiency, and reproducibility, but their implementations and tool ecosystems are still young. The goal of this BoF is to connect researchers and industry practitioners interested in improving the state of the art in this area, and to begin to identify concrete ways to work together.

In this video from the HPC Advisory Council Australia Conference, John Gustafson from National University of Singapore (NUS) presents: Beating Floats at Their Own Game.

Gustafson is also a speaker at the Computational Reproducibility at Exascale 2017 Workshop at SC17.  For this talk, he plans to explain why people get different answers from one run to another using IEEE floats. And then, obviously, why posits fix that problem.

Reproducibility is an important concern in all areas of computation. As such, computational reproducibility is receiving increasing interest from a variety of parties who are concerned with different aspects of computational reproducibility. Computational reproducibility encompasses several concerns including the sharing of code and data, as well as reproducible numerical results which may depend on operating system, tools, levels of parallelism, and numerical effects. In addition, the publication of reproducible computational results motivates a host of computational reproducibility concerns that arise from the fundamental notion of reproducibility of scientific results that has normally been restricted to experimental science. This workshop combines the Numerical Reproducibility at Exascale Workshops (conducted in 2015 and 2016 at SC) and the panel on Reproducibility held at SC16 (originally a BOF at SC15) to address several different issues in reproducibility that arise when computing at exascale. The workshop will include issues of numerical reproducibility as well as approaches and best practices to sharing and running code.

Dr. John L. Gustafson is an applied physicist and mathematician, currently Professor at National University of Singapore and Visiting Scientist at A*STAR Computational Resource Centre. He is a former Director at Intel Labs and former Chief Product Architect at AMD. A pioneer in high-performance computing, he introduced cluster supercomputing in 1985, and first demonstrated scalable massively parallel performance on real applications in 1988, for which he won the inaugural ACM Gordon Bell Prize. The principles used to achieve this have become known as Gustafson’s Law. He is a recipient of the IEEE Computer Society’s Golden Core Award.

Dr. Gustafson has recently finished writing a book, The End of Error: Unum Computing, that presents a new approach to computer arithmetic: the unum. The universal number, or unum format, encompasses all IEEE floating-point formats as well as fixed-point and exact integer arithmetic. “This approach obtains more accurate answers than floating-point arithmetic yet uses fewer bits in many cases, saving memory, bandwidth, energy, and power.”

Registration is now open for SC17, which takes place Nov. 12-17 in Denver.

See our complete coverage of SC17

Check out our insideHPC Events Calendar

Resource Links: