A General Theory of Computational Scalability Based on Rational Functions

TL;DR

This paper introduces a universal rational function model for computational scalability, proving its equivalence to a queue-theoretic throughput bound and demonstrating its necessity and sufficiency for practical performance modeling.

Contribution

It establishes a general theoretical framework linking rational functions to computational scalability and performance bounds, unifying existing models like Amdahl's law.

Findings

C_p is equivalent to the throughput bound of a machine-repairman model.

Simpler models like Amdahl's law are special cases of this framework.

C_p is necessary and sufficient for modeling practical scalability characteristics.

Abstract

The universal scalability law of computational capacity is a rational function C_p = P(p)/Q(p) with P(p) a linear polynomial and Q(p) a second-degree polynomial in the number of physical processors p, that has been long used for statistical modeling and prediction of computer system performance. We prove that C_p is equivalent to the synchronous throughput bound for a machine-repairman with state-dependent service rate. Simpler rational functions, such as Amdahl's law and Gustafson speedup, are corollaries of this queue-theoretic bound. C_p is further shown to be both necessary and sufficient for modeling all practical characteristics of computational scalability.