Experiences in autotuning matrix multiplication for energy minimization on GPUs

摘要

In this paper, we report extensive results and analysis of autotuning the computationally intensive graphicsrnprocessing units kernel for dense matrix–matrix multiplication in double precision. In contrast to traditionalrnautotuning and/or optimization for runtime performance only, we also take the energy efficiency intornaccount. For kernels achieving equal performance, we show significant differences in their energy balance.rnWe also identify the memory throughput as the most influential metric that trades off performance and energyrnefficiency. As a result, the performance optimal case ends up not being the most efficient kernel in overallrnresource use.