【摘要】 充分利用多核结构的共享三级缓存、私有二级和一级缓存,将采用Modeling算法求解(l,d)-Motif发现问题过程中产生的大量不同位置组合的数据,以多轮方式调度分配到各级缓存中,通过线程绑定以平衡核心负载,设计实现了一种访存高效的(l,d)-Motif发现线程级并行算法.实验结果表明,与已有的并行求解(l,d)-Motif发现算法相比,本文给出的算法更高效、具有较好的加速比和可扩展性,并且能够成功求解更长l的(l,d)-Motif发现问题.更多还原

【Abstract】 By utilizing fully the shared L3Cache,private L2Cache and L1Cache on the multi-core architecture,the produced large-scale data with the different combination of positions are distributed to the multi-level caches while the Modeling algorithm is used to solve(l,d)-Motif finding problem,and a cache-efficient thread-level parallel(l, d)-Motif finding algorithm is designed on multi-core computer by binding the threads to the cores to balance their computational loads.The experimental results show that,compared with the existing parallel algorithm to solve(l, d)-Motif finding problem,the presented algorithm is efficient and has good speedup and scalability,and it can solve successfully the(l,d)-Motif finding problem with longer l.更多还原