Energy-Performance Trade-off Analysis of Parallel Algorithms

Energy consumption by computer systems has emerged as an important concern, both at the level of individual devices (limited battery capacity in mobile systems) and at the societal level (the production of Green House Gases). In multicore architectures, applications may be executed on a variable number of cores and these cores may operate at different frequencies. The performance and energy cost of a parallel algorithm executing on a multicore architecture have different trade-offs, depending on how many cores the algorithm uses, at what frequencies these cores operate, and the structure of the algorithm. We show how algorithm designers and software developers can analyze the energy-performance trade-off in parallel algorithms. We believe that such analyses should be applied to parallel algorithms to facilitate energy conservation.