WSSP

The future of computing is parallel but it may not be easy

Michael J. Flynn
Stanford University

Abstract

Today’s emphasis on multi core architectures comes about from the failure of frequency scaling due to power density limitations not because of breakthroughs in parallel programming or architecture.

Progress in automatic compilation of serial programs into multi tasked ones has been slow. A look at parallel projects of the past illustrates problems in performance and programmability. Solving these problems requires both an understanding of underlying issues such as parallelizing control structures and dealing with the memory bottleneck.

For many applications performance comes at the price of programmability and reliability comes at the price of performance.

Fifty Years of High Performance Computing -- From 1964 to 2014 --

Yoshio Oyanagi
Education Center for Computational Sciences, Kobe University

Abstract

The history of high performance computing and computational sciences is briefly summarized and the characteristics of the Japanese HPC development are compared with those in the U.S. and Europe. When we youngsters started computational sciences in the 1970’s, senior professors were skeptic about it. In the 1980’s, however, Japanese vendors started to produce high-quality vector computers and Japanese government deployed them at university and laboratory computer centers. They were so user friendly that few dared to harness parallel computers. The success of vector computers in Japan caused on the other hand the delay of parallel computer development. In the U.S., a number of venture companies started to build massively parallel computers in the 1980’s and users welcomed them, while there were then no Japanese commercial parallel products, although parallelism research was active in the Japanese academia. The situation drastically changed when the K Computer started to work in 2011. We are looking forward to the Exascale computer around 2020.