Low-power tradeoffs for mobile computing applications
Tiffany M. Mintz, James P. Davis · 2007
In this paper, we present a low-power estimation method for creating mobile computing applications on programmable logic devices forming the "fabric" of a reconfigurable computing platform. Today, the predominant method for creating mobile computing applications uses one or more embedded processors, selected for their cost, ease of programmability, and flexibility. However, concerns over performance and power consumption are causing systems designers to question this prevailing viewpoint. We present an approach that allows designers to assess the performance and energy consumption of key algorithms before committing to specific hardware-software partitioning decisions. We report on results using the Algorithmic State Machine (ASM) method, found now in most texts on digital logic design. Using a common Sorting algorithm as an example, we show how the ASM method can be used to estimate the impact of algorithm design on power consumption and, therefore, incorporate energy budgeting into the decision-making process for hardware-software partitioning.