Productivity Metrics and Models

1. Introduction A conceptual framework is required for understanding the relative merits and utility of alternate computer systems for high performance: their architecture, execution model, and software methods for programming and operation. The general concept of " productivity " has been articulated as that attribute of the entire process of computing that delivers ultimate value to the end user mission. Productivity is defined to reflect the rate, value, and costs of producing computational results in contribution to achieving mission objectives of the using institution. While there is general consensus as to the importance and abstract nature of a productivity value of merit, there is not a common viewpoint of the specific metric and its formulation by which to quantitatively characterize it. The contribution of this paper is to propose a conceptual framework within which to consider productivity as a parameter and a formulation by which to quantify it. The objective of this framework is to establish a set of mutually consistent and complementary metrics that satisfy the discipline of dimensional analysis and that together provide a rigorous definition of productivity. It is the intent of this work that its results may serve as a tool for evaluation and comparison of alternate high performance computing systems.