Abstract: With the development of HPCs,increasingly importance has been attached to reducing power consumption and raising productivity.We proposed a high productivity computing model to deal with the HPCs’ productivity problem,which adopts the concept of reconfigurable computing and is based on proactive cognition and decision system.This model apperceives the real-time states of application tasks,evaluates the matching degree of application state and current application structure,and then reconfigures the application structure to lower energy consumption and increase productivity.In order to verify the model’s effectiveness,we constructed a prototype experimental platform,implemented a video-copy-detection program and a password recovery program,and then used real Internet traffic statistical curve to simulate the programs’ loads.Experimental results demonstrate that under this environment the application system based on the model has raised its productivity by 58% compared with traditional method.

Key words: High productivity,Energy-efficient,High performance,Reconfigurable computing,Proactive cognition,Structure decision