System and methods for optimizing energy efficiency in programmable devices

摘要

A system for optimizing energy efficiency of an embedded system. A hardware abstraction layer (HAL) is coupled to hardware of an embedded system. An energy hardware abstraction layer (eHAL) couples to the hardware and to the HAL. A hypervisor couples to the HAL and to the eHAL. One or more applications couple to the hypervisor, the HAL and the eHAL. The eHAL measures energy use of the embedded system, constructs a model of energy use of the embedded system and, using the model, determines how to one of reduce and minimize energy use of the embedded system. The hypervisor is configured to alter one or more hardware parameters to one of reduce and minimize energy use of the embedded system as the embedded system performs one or more tasks. In implementations a real time operating system (RTOS) is coupled to the one or more applications and to the hypervisor.

主权项

1. A method of real time energy efficiency optimization for an embedded system, comprising:
using an energy hardware abstraction layer (eHAL) of an embedded system to measure power consumption of the embedded system during execution of the embedded system's tasks, the embedded system comprising: hardware, a hardware abstraction layer (HAL) coupled to the hardware, the eHAL coupled to the hardware and to the HAL, a hypervisor coupled to the HAL and to the eHAL, a real time operating system (RTOS) coupled only to the hypervisor and at least one application, the at least one application coupled to the hypervisor and to the RTOS, the HAL and the eHAL; calculating, using one of the eHAL, the hypervisor and both the eHAL and the hypervisor, from the measured power consumption, electrical energy use of the embedded system; generating, using a processor coupled to the embedded system and using one of the eHAL and the hypervisor, a model of electrical energy use of the embedded system, the model relating electrical energy use of the embedded system for tasks performed by the embedded system to values of hardware parameters of the embedded system; storing the model in the eHAL for use by the hypervisor; in response to the embedded system being required to perform a specific task, determining, with the hypervisor, using the model of electrical energy use, optimized hardware parameter settings to one of reduce and minimize electrical energy use of the embedded system while performing the specific task; using one of the hypervisor and the eHAL, implementing the optimized hardware parameter settings, and determining, using the hypervisor, whether one or more optimization steps is likely to result in a net reduction in electrical energy use of the embedded system, taking into account an amount of electrical energy required to implement the one or more optimization steps.