Poster: Statistical Power and Energy Modeling of Multi-GPU Kernels

摘要

To improve the energy efficiency of parallel ap- plications on GPGPUs, a better understanding of the energy behavior of various applications is mandatory. In this study we employ statis- tical methods to model power and energy con- sumption of some common optimized high per- formance kernels (DGEMM, FFT, PRNG and FD stencils) on a multi-GPU platform.