Abstract: Data transferring between clusters degrades the performance of clustered VLIW processors. This paper presents a novel two-dimension forced-directed schedule algorithm for Register-File-Connectivity Clustered Very Long Instruction Word (RFCC-VLIW) processors, in which the force expression varies as the cycle and cluster. Experimental results indicate that the schedule algorithm prevails against other ones for clustered VLIW processors when aiming to RFCC-VLIW architecture.

Key words: force-directed, Very Long Instruction Word(VLIW), schedule algorithm

摘要： 多簇超长指令字处理器的簇间数据传输会将导致处理器性能下降。该文针对寄存器堆互连的多簇超长指令字(RFCC-VLIW)结构，提出一种新的二维力量引导调度算法，其力量表达式为以周期和簇为自变量的二维力量。实验结果表明，以RFCC-VLIW结构为目标，该二维力量引导调度算法优于现有的其他多簇超长指令字处理器的调度算法。

关键词: 力量引导, 超长指令字, 调度算法

CLC Number: 

• TP302