本项目研究面向高能效和低延时的片上多核共享资源冲突约束优化模型。该模型包括三个主要优化目标：（1）基于缓存节能分区的Bank冲突延时最小化；（2）实时节能总线策略与支持机制；（3）片上共享资源的能耗最小化。本项目紧密结合高能效计算和WCET计算，重点研究一个Bank冲突延时最小化算法，在优选的缓存节能分区容量基础上使Bank冲突延时最小化；以及研究一个支持UBD优化值和时间距离的实时节能总线策略与支持机制，在减少的UBD改进值内保障硬实时任务及时得到总线服务，在时间距离值内提升非硬实时任务总线服务能力；同时在保障硬实时任务WCET满足定时要求的条件下，利用缓存节能机制和总线节能机制使片上共享资源的能耗最小化。研究成果可广泛应用于诸多电池供电的高端多核实时系统。本项目的可行性已在前期工作中得到了充分的验证。

This project studies the optimization model on bounding of interferences in on-chip shared resources for energy-efficient and low-delay embedded multi-core. The model includes three main optimizing goals: (1) minimizing bank conflict delay based on energy-efficient cache partitioning, (2) energy-efficient real-time bus strategy and its support mechanisms, and (3) minimizing energy consumption for on-chip shared resources. This project closely combines energy-efficient computation with WCET computation,the focus of our research is to construct the algorithm which minimizes bank conflict delay under the optimal sizes of energy-efficient cache partitions, and the energy-efficient real-time bus strategy and its support mechanisms with optimal UBD value and time-distance value,which ensure hard real-time task to get bus service timely in the reduced UBD value,and promote the bus service ability of non hard real-time task in the time-distance value, at the same time using cache energy-saving mechanism and bus energy-saving mechanism to minimize energy consumption of on-chip shared resources under the condition that hard real-time task WCET meet timing requirements. The research results can be widely used in high-end multi-core real-time systems with battery powered.The feasibility of this research has been verfied sufficiently in our preliminary work.