随着多核处理器片上集成核数的不断增多,并行任务的调度能力越来越成为制约性能提升的关键因素。文章设计一种面向异构多核计算系统的动态任务调度控制器,主要实现动态监控处理单元的负载情况、动态任务唤醒、乱序任务发射、任务写回安...随着多核处理器片上集成核数的不断增多,并行任务的调度能力越来越成为制约性能提升的关键因素。文章设计一种面向异构多核计算系统的动态任务调度控制器,主要实现动态监控处理单元的负载情况、动态任务唤醒、乱序任务发射、任务写回安全管理等功能;研究一种降低计算任务结果数据回写双倍数据速率(double data rate,DDR)外存储器次数的方法,大幅节省了访存开销,进一步提升了计算性能。仿真及性能测试显示,在典型应用场景下,与已有的无动态调度功能的任务发射控制器相比,实现了显示并行化编程向任务并行的自动化控制过渡,编程友好度显著提高,在不同类型的测试案例中,分别提升了11.3%~37.9%的计算性能。展开更多
This paper analyzes the threat of TCG Software Stack(TSS)/TCM Service Module(TSM) deadlock in multi-user environment such as cloud and discusses its causes and mechanism.In addition,this paper puts forward a dynamic p...This paper analyzes the threat of TCG Software Stack(TSS)/TCM Service Module(TSM) deadlock in multi-user environment such as cloud and discusses its causes and mechanism.In addition,this paper puts forward a dynamic priority task scheduling strategy based on value evaluation to handle this threat.The strategy is based on the implementation features of trusted hardware and establishes a multi-level ready queue.In this strategy,an algorithm for real-time value computing is also designed,and it can adjust the production curves of the real time value by setting parameters in different environment,thus enhancing its adaptability,which is followed by scheduling and algorithm description.This paper also implements the algorithm and carries out its performance optimization.Due to the experiment result from Intel NUC,it is shown that TSS based on advanced DPTSV is able to solve the problem of deadlock with no negative influence on performance and security in multi-user environment.展开更多
文摘随着多核处理器片上集成核数的不断增多,并行任务的调度能力越来越成为制约性能提升的关键因素。文章设计一种面向异构多核计算系统的动态任务调度控制器,主要实现动态监控处理单元的负载情况、动态任务唤醒、乱序任务发射、任务写回安全管理等功能;研究一种降低计算任务结果数据回写双倍数据速率(double data rate,DDR)外存储器次数的方法,大幅节省了访存开销,进一步提升了计算性能。仿真及性能测试显示,在典型应用场景下,与已有的无动态调度功能的任务发射控制器相比,实现了显示并行化编程向任务并行的自动化控制过渡,编程友好度显著提高,在不同类型的测试案例中,分别提升了11.3%~37.9%的计算性能。
基金supported by the State Key Program of National Natural Science Foundation of China(Grant No.91118003)the National Natural Science Foundation of China(Grant No.61173138,61272452,61332019)+1 种基金the National Basic Research Program of China("973"Program)(Grant No.2014CB340600)the National High-Tech Research and Development Program of China("863"Program)(Grant No.2015AA016002)
文摘This paper analyzes the threat of TCG Software Stack(TSS)/TCM Service Module(TSM) deadlock in multi-user environment such as cloud and discusses its causes and mechanism.In addition,this paper puts forward a dynamic priority task scheduling strategy based on value evaluation to handle this threat.The strategy is based on the implementation features of trusted hardware and establishes a multi-level ready queue.In this strategy,an algorithm for real-time value computing is also designed,and it can adjust the production curves of the real time value by setting parameters in different environment,thus enhancing its adaptability,which is followed by scheduling and algorithm description.This paper also implements the algorithm and carries out its performance optimization.Due to the experiment result from Intel NUC,it is shown that TSS based on advanced DPTSV is able to solve the problem of deadlock with no negative influence on performance and security in multi-user environment.
