Hardware-software Partitioning And Design Of Embedded Operating System

The tendency of computer hardware and software combine together is becomingmore and more clearly, software to hardware and software in place of hardware aretwo parallel orientation of computer system development.With the development ofembedded systems, the performance of traditional embedded real-time operatingsystem is difficult to meet the need of high-end applications. At the same time, sincethe developed of SOC technology, software in place of hardware get a broadlydevelopment space, some special purpose softwares are the in place of hardware untilnow. Therefore, making parts of embedded real-time operating system in place ofhardware, researching how to finish the Hardware-software Partitioning of embeddedreal-time operating system and realizing a Hardware-software hybrid operatingsystem is a problem which is new and very influential.In this paper, a simple analysis of μC/OS-II which is the reference operatingsystem is given, on this basis, some function modules part of μC/OS-II areprogrammed with the VHDL language and simulated by ISE, then got the time, spaceand power dissipation needed when these program are running by hardware directly.μC/OS-II is compiled by XPS and run on PowerPC405, then got the time, space andpower dissipation needed when these program running on microprocessor. Comparethe data which obtained from the hardware simulation and software implementation,we can find that there is higher efficiency when implemented programs withhardware take the place of software, of course, the hardware often take up morehardware resources; The software data and hardware data obtained are used inhardware-software partitioning.Select the genetic algorithm NSGA-II as the hardware and software partitioningalgorithm, NSGA-II has been improved to heal the defects of genetic algorithmwhich are early-maturing and need much time to finish. Improve the efficiency ofalgorithm operating in large-scale population and reduce the complexity of algorithm computational by using the fast non-dominated sorting algorithm; Maintainpopulation diversity with the crowding operator; Expand the sample space, Makesure the best individual is not lost and get rapid increase in population levels by usingthe elite strategy. Similarly, there are more repeat the individual in every populationof NSGA-II algorithm than other algorithms, the reason is using above measures inalgorithms, result in the lost of a large number of Pareto solutions. In this paper thedefect has been improved. The solutions of improved algorithm has betterdistribution than the solutions of original algorithm, the improved algorithm has amore comprehensive search solution space, At last,complete the hardware-softwarepartioning of task management part with the improved algorithm.