The inputs were taken from ISCAS benchmark set. The optimization script performs only symbolic computations on the nodes of the circuits. Moreover, all the examples could fit in the memory. Our intuition would lead us to expect that memory effects would be secondary in this case and the performance will be a reflection of the MIPS rating of the machine. However, as shown in Figure 2 we notice an anomaly which is that R4400_60 is performing much better than its low clock speed would indicate. In Section 3.2.1, we make an attempt to explain this anomaly basing our arguments on cache simulation results. We notice that its running times are quite faster than the Sparc20_100. We also observe that both the Sparc machines perform poorly on this application. Since Sparc10_50 is the base machine, the performance ratios of the non-Sparc machines is much higher compared to those in other applications.