Cost-Effective Matching of Task and Computer Device

Abstract

The primary goal was policy guidance for the proportions of resources to be spent on men and computing devices to accomplish engineering tasks. The secondary goal was an analysis for evaluating the cost of performing a spectrum of engineering tasks on a spectrum of devices. This paper attempts to parameterize the task descriptions and then to analytically model the computational environments. Seven types of devices were studied. These were: slide rule; mechanical desk calculator; stored-program desk calculator; small time-sharing system; large time-sharing system; small, local batch computer; and large, centralized batch computer. These devices were simulated in terms of task input man-time, computation man-time, output man-time, man waiting time, machine time. The major findings were: (1) The current large batch oriented system has minimized the cost of computing equipment but has unfortunately maximized task total cost; and (2) the alternates, essentially time-shared systems, require that more money be spent on computing equipment, nevertheless, they drastically reduce the task total cost, because they reduce task flowtime.