一种粗粒度可重构体系结构多目标优化映射算法

A Multi-Objective Optimization Mapping Algorithm for Coarse Grained Reconfigurable Architectures

针对多约束下的行流水粗粒度可重构体系结构的硬件任务划分映射问题,提出了一种多目标优化映射算法.该算法根据运算节点执行时延、依赖度等因素构造了累加概率权值函数,在满足可重构单元面积和互连等约束下,通过该函数值动态调整就绪节点的映射调度次序,当一块可重构单元阵列当前行映射完毕后,就自动换行,当一块阵列被填满,就切换到下一块,当一个数据流图映射完毕后,就自动计算划分块数等参数.实验结果表明,与层贪婪映射算法相比,文中算法平均执行总周期降低了8.4%(RCA4×4)和5.3%(RCA6×6),与分裂压缩内核映射算法相比,文中算法平均执行总周期降低了20.6%(RCA4×4)和21.0%(RCA6×6),从而验证了文中提出算法的有效性.

Based on rowpipelining coarse grained reconfigurable architecture（ CGRA）,we presented MOM（ multiobjective optimization mapping） algorithm to solve the multi-constraints hardware task partitioning-mapping problem. The cumulative probability weight function was constructed by the execution delay of computing nodes and the dependence between two nodes,etc. With the constraints of reconfigurable cell area and interconnection,the proposed algorithm could adjust dynamically the scheduling order of the ready nodes by thefunction values. When a rowof the RCA was mapped completely,MOM began on a newrow. When the RCA was filled,MOM switched to the next one. When a DFG（ data flowgraph） was mapped completely,the number of modules and etc were calculated automatically in MOM. Experiment results showthat the average execution total cycles of MOM decrease by 8. 4%（ RCA4 × 4） and 5. 3%（ RCA6 × 6） comparing with LBGM（ level based greedy mapping） algorithm. Comparing with SPKM（ split-push kernel mapping） algorithm,the average execution total cycles of MOM decrease by 20. 6%（ RCA4 × 4） and 21%（ RCA6 × 6）. Experimental evaluation confirms the efficiency of our approach.