云计算环境下浮点和整数同时多线程资源配置方法研究  

Cloud Computing Environment Floating-point and Integer Multithreading Resource Allocation Methods at the Same Time

作　　者：崔嘉[1] 

机构地区：[1]海军航空工程学院,烟台264001

出　　处：《科学技术与工程》2017年第23期238-242,共5页Science Technology and Engineering

摘　　要：云计算环境下数据规模较大,当前资源配置方法依据优先级、按照顺序对资源进行配置,具有资源利用率低、配置效率低、能耗高等弊端。为此,提出一种新的云计算环境下浮点和整数同时多线程资源配置方法,分析了计算操作和访存操作的并发性,给出云计算环境下和每片外访存load操作相应的发射执行指令数的定义,得出在相同时间段内IPM值越大,计算型与访存型操作同时执行的机会越大,资源配置性能越高的结论。依据该结论,通过IPM实现云计算环境下浮点和整数同时多线程资源配置,将较多的资源分配至IPM值较大的浮点或整数线程中,将较少的资源分配至IPM值较小的浮点或整数线程中,使得云计算环境下的共享资源得到最佳配置。利用实验测试所提方法的实际应用性能。实验结果表明,所提方法资源利用率高、能耗损失低、资源配置效率高,整体资源配置性能很高。Under the cloud computing environment,the data size is large,the current resource allocation method based on priority,in accordance with the order of the allocation of resources,with low resource utilization,low allocation efficiency,high energy consumption. To this end,a new cloud computing environment and multi thread floating point and integer allocation of resources were proposed,analyzes the calculation operation and access concurrency,cloud computing environment and each chip memory load operation corresponding emission instruction definition,obtained in the same period of time the value of IPM the greater computational and memory access operations executed at the same time the greater the chance, the higher the conclusion resource allocation performance. On the basis of the conclusion,through the realization of IPM cloud computing environment floatingpoint and integer simultaneous multithreaded allocation of resources,more resources will be allocated to the IPM of the value of the floating-point or integer thread,the less resource allocation to IPM smaller floating-point or integer thread,the cloud computing resource sharing environment to get the best configuration. The experimental results show that the proposed method has the advantages of high resource utilization,low energy consumption,high resource allocation efficiency,and high resource allocation performance.

关 键 词：云计算环境 浮点 整数 多线程资源配置 

分 类 号：TP391.75[自动化与计算机技术—计算机应用技术]