Abstract
An improved classification methodology for sorting data-processing instructions for ARM7TDMI CPU core is presented in this paper. Main discussion here is related to the process of creating appropriate training sets for neural network (NN) based estimation of power consumption. We have proposed instructions’ sorting methodology according to the binary instruction representation and the resources being used for the overall system model. Thus separate instructions groups are obtained for NN-based estimation of power consumption. Experimental results of the proposed method confirm successful usage of this sorting methodology for providing higher accuracy estimation of power consumption.
Highlights
One of the problems, which require detailed analysis during the design of embedded and autonomous systems, is power-aware design
According to the results of the previous work it was decided to use preliminary sorting methodology related to the maximum available data as well as grouping various instructions within separate subgroups, according to their power consumption performance
The step in the described procedure is to form groups according to the proposed method of the sorting instructions and taking into account available data about CPU’s “basic” power consumption during the data-processing instructions execution: 1. Arithmetic-logic instructions
Summary
One of the problems, which require detailed analysis during the design of embedded and autonomous systems, is power-aware design. There are known power-optimization methods [3] [4] which are based on the fastest running of the routines and putting microprocessor into powersaving mode (usually, by disconnecting unused blocks of the CPU). These methods are highly effective when the program must run few times, and it doesn’t involve detailed analysis of CPU’s analysis of operation according to the powerconsumption. An improved data sorting and analysis methodology will be represented in this paper aiming to complete the approach of using neural networks for the software-related power consumption estimation [7]
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have