5. Conclusions

The width and length dimensions of p-channel and n-channel transistors used are (1000, 100) and (500, 100) µm, respectively. We run PowerMill on these 200 vectors for each circuit. For comparison, we also run PowerMill on the original vectors. Note that in this experiment we only generate 1000 vectors as the reference vector set. We have also generated a much larger set of vectors and found that our program produces very similar results to those presented in Table 1. However, due to the long simulation time of PowerMill, we could not afford to simulate such long vector set for comparison. Table 2 shows the results of running PowerMill on V original (1000 vectors) and S (200 vectors). The average ground current are obtained by assuming the clock rate is 1MHz. The last column shows the ratio of current obtained by running PowerMill on S versus that obtained by running PowerMill on V Power simulators offer an accurate solution to estimate power dissipation for CMOS circuits. However, for larger designs, it is common that the given set of functional vectors which characterizes the typical operations of a circuit is too long for such tools to handle. On the other hand, logic-level approaches, though efficient, may not be able to provide accurate estimation due to some inherent limitations. In this paper, we propose a method that combines the advantages of the logic-level and the transistor-level Circuit Model error abs. transition error Time (seconds)