...
首页> 外文期刊>Journal of Computers >An ILP-based DMA Data Transmission Optimization Algorithm for MPSoC
【24h】

An ILP-based DMA Data Transmission Optimization Algorithm for MPSoC

机译:基于ILP的MPSOC DMA数据传输优化算法

获取原文
           

摘要

—With the rapid development of integrated circuitdesign technology and the processed tasks and data volumesgrowing, MPSoC is becoming increasingly popular in avariety of applications. In MPSoC design, parallelism is avery important issue, for example, how to realize taskparallelism and data parallelism. Focusing on this issue, thispaper analyzes the role of DMA and presents an ILP-BasedDMA data transmission optimization algorithm to reducethe pipeline time when employing multi-stage pipelinescheduling method to solve task parallelism and dataparallelism. The proposed ILP model integrates taskallocation/schedule and data transmission and thus realizesthe optimal parallelism of data transmission and dataprocessing. In addition, we divide data transmission of ILPmodel into four cases: (1) DMA0, do not use DMA tooptimize data transmission; (2) DMA1, use DMA totransmit data between SPM and off-chip memory; (3)DMA2, use DMA to transmit data between SPM and SPM,SPM and off-chip memory; (4) DMA3, use DMA to transmitand prefetch all data. Simulation results show that the ILPmodel with DMA3 can reduce the pipeline time 17.8%compared with that of the ILP model with DMA0.
机译:- 在集成电路技术的快速发展和处理的任务和数据体积的快速发展中,MPSOC在似乎越来越受应用的应用中变得越来越受欢迎。在MPSOC设计中,并行性是Avery重要问题,例如,如何实现任务平行主义和数据并行性。关注此问题,此纸纸分析DMA的作用,并提出了一种用于在采用多级流水线的方法来解决任务并行和数据识别中的管道时间来减少管道时间的ILP基本数据传输优化算法。所提出的ILP模型集成了TaskAllocation /计划和数据传输,从而实现了数据传输和Dataprocessing的Realizesthe最佳并行性。此外,我们将ILPModel的数据传输分为四个情况:(1)DMA0,不使用DMA来优化数据传输; (2)DMA1,使用SPM和片外存储器之间的DMA TOTRANSMIT数据; (3)DMA2,使用DMA传输SPM和SPM,SPM和片外存储器之间的数据; (4)DMA3,使用DMA传输和预取所有数据。仿真结果表明,与DMA0的ILP模型相比,DMA3的ILPMODEL可以减少17.8%的管道时间。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号