Power-aware Computation and Programming with GPU Accelerators

Present concerns about the amount and intensity of power in modern processors have been led to significant computer architecture research efforts in power-aware and temperature-aware calculations. Sincepower dissipation becomes an irritating growing problem, the graphic processor unit (GPU) analysis of power and its components is very difficult to design software and hardware systems.At present, GPU has become to an integral part of computing systems mainstream with multi-core multidisciplinary processors, with high symmetry and balance and high computational power and memory very high-bandwidth.To find the rules and principles of power-aware programming with GPU accelerators, we exploit a set of the basics fromprogramming languages. These power consumption amounts are efficient to estimate and evaluate the power during high-level programs development.

کلیدواژه ها:

GPU ، Power-aware Programming Principals