The Research And Implementation Of Parallel Computing Method On MPCore Multicore Processor

With the improvement of embedded system performance from people and the improvement of embedded processor technology, embedded processor has entered the multicore processor time from single-core processor time. However, Although the count of processor core has improved largely, the traditional development method of embedded system has based on single-core model, which is unable to be full use of the performance of the embedded multicore processor, resulting in a waste of processor computing resources, not good to reduce the processor power consumption.Recently, MPCore multicore processor has been widely in embedded application system. Based on MPCore multicore processor, the paper can apply research results into a wide range of embedded system development universally. The method researched by the paper is a way of multicore parallelism optimization on the embedded multicore processor, which can transform the traditional single-core model’s software into multi-core parallel model to adapt to the embedded multicore processor environment, use as many as computing cores and get the best system performance by full use of multicore processor without change the original software architecture. The method can build a balance point between processor performance and power consumption, improving energy efficiency, ling with today’s low carbon environmental theme.Firstly, based on MPCore multicore processor hardware structure, the paper analyses how multicore processors support multiple computing cores simultaneously, and communication method among multiple cores, finds MPCore multicore processor support parallel computing bottleneck through comparative analysis between MPCore embedded multicore processor and the desktop multicore processors hardware structure. Secondly, the analysis of key technologies involved in parallel computing methods, including multi-core parallel methods, evaluation of multi-core parallelism, the support for parallel computing from embedded operating system and embedded multicore processor’s parallelization of difficulties, multi-core parallel methods section contain parallel model analysis and parallel programming tools, evaluation methods section is the way to the parallelization performance analysis of parallel algorithms, the support for parallel computing from embedded operating system section tells how to support multiple processors simultaneously, from the difficulty of development tools, hardware performance, and the parallel strategy, the section of embedded multicore processor’s parallelization of difficulties tells where is difficulties and bottlenecks and proposes a parallel multi-core processors for embedded the basic method of calculation. Finally, with the task parallelism and data parallelism in embedded multicore processors calculated on MPCore multicore processor platform, the paper transform the traditional serial model software into parallel model, improving the software performance and achieving good results through performance tuning and test analysis, verified this paper about MPCore multicore processor parallel computing method is feasible and can apply the method to actual parallelization of embedded systems development.