Effective Acceleration of Iterative Slack Distribution Process
Xinjie Wei, Yici Cai, Xianlong Hong · 2007
Iterative slack distribution is a prevalent method in timing analysis and clock scheduling. Finding minimum mean cycle is the most time consuming step in the each iterative process. We present a practical strategy that can speed up the loop. A fast negative cycle detection method is used for examining whether the cycle of length two is really the minimum mean cycle. Traditional complicated minimum mean cycle algorithm can be skipped during the iterative slack distribution process. Experimental results show that our method can reduce running time of the iterative slack distribution process. The percentages are from 47% to 90% for different benchmark circuits.