为了更有效地优化粗粒度可重构单元阵列映射加速性能,提出了一种行节点无依赖约束的空域映射调度方法,基于相同条件下,采用时延Petri网对若干个按约束已经被划分映射到可重构单元阵列的数据流子图的运行情况进行了分析,通过一个实例比较了行节点有依赖和无依赖的运行结果,结果表明该种空域映射方法具有可行性。
In order to more effectively optimize the mapping acceleration performance of coarse grained Reconfigurable Cell Array(RCA), a method of row nodes no-dependency spatial mapping scheduling constraint is put forward. Timed Petri nets are used to analyse several data flow subgraphs which have been partitioned and mapped on RCA by constraint based on the same conditions. The running results of row nodes dependency mapping and row nodes no-dependency mapping are compared by an example. The experimental results show that this spatial mapping method is feasible.