Parametric Estimation of the Ultimate Size of Hypercomputers

TL;DR

This paper analyzes how power, size, and other physical constraints influence the design and efficiency of future hypercomputers, emphasizing the importance of massive parallelism for applications with random memory access.

Contribution

It introduces a parametric model to estimate the maximum size of hypercomputers considering physical and performance factors, highlighting the role of multithreading.

Findings

Massively-parallel computing is essential for random memory access applications.

Physical constraints significantly influence hypercomputer design.

High-performance computing relies on extensive multithreading.

Abstract

The performance of the emerging petaflops-scale supercomputers of the nearest future (hypercomputers) will be governed not only by the clock frequency of the processing nodes or by the width of the system bus, but also by such factors as the overall power consumption and the geometric size. In this paper, we study the influence of such parameters on one of the most important characteristics of a general purpose computer - on the degree of multithreading that must be present in an application to make the use of the hypercomputer justifiable. Our major finding is that for the class of applications with purely random memory access patterns "super-fast computing" and "high-performance computing" are essentially synonyms for "massively-parallel computing."