一种基于bank-column缓存划分的多核共享缓存能耗优化方法

Bank-column Cache Partitioning for Shared Cache Energy Optimization in Multicore Systems

  • 摘要: 针对支持IABA (Interference-Aware Bus Arbiter)总线的多核共享缓存能耗问题,提出了一种基于bank-column缓存划分的多核共享缓存能耗优化方法.该方法在保证硬实时任务能够在截止期前完成的前提下,通过优化bank-column缓存划分来减少共享缓存能耗.实验结果表明,在保证硬实时任务能够在截止期前完成的前提下,提出的优化方法能够减少共享缓存能耗,与优化路-组缓存划分相比,35 nm制造工艺时至少能够减少约10%的共享缓存能耗.

     

    Abstract: For the shared cache energy in hard real-time multicores with IABA (Interference-Aware Bus Arbiter) bus, a bank-column partitioning optimization method was proposed for the shared cache energy. In this method, the bank-column cache partitioning was optimized to save the shared cache energy on the premise of hard real-time tasks which can be finished before their deadlines. Experimental results demonstrated that the proposal method can effectively reduce the shared cache energy consumption and guarantee that the hard real-time tasks can be finished before their deadlines. Compared with the way-set cache partitioning optimization method, the proposal method can save about 10% energy of the shared cache in 35 nm technology.

     

/

返回文章
返回