DocumentCode :
2215533
Title :
Efficient power analysis of combinational circuits
Author :
Krishnamoorthy, Shankar ; Khouja, Adel
Author_Institution :
Synopsys Inc., Mountain View, CA, USA
fYear :
1996
fDate :
5-8 May 1996
Firstpage :
393
Lastpage :
396
Abstract :
Designers of low power chips are finding it increasingly useful to obtain estimates of average power consumption early in the design phase. They need a fast power estimation capability which gives reasonably accurate power estimates. In this paper, we present some ideas to obtain an efficient implementation of probabilistic power analysis for combinational circuits. These ideas are based on the observation that in order to obtain a fast implementation, it is very important to control the sizes of the Reduced Ordered Binary Decision Diagrams (ROBDDs) that are created during analysis. However, controlling sizes of ROBDDs could impact the accuracy of the result. We explore the consequences of making the trade-off between performance of the implementation and accuracy. Our experimental results illustrate that our implementation can provide estimates that are within 5% of a commonly used probabilistic method and runs 10 times faster, on average. Our method also scales very well for large circuits
Keywords :
Boolean functions; circuit analysis computing; combinational circuits; integrated logic circuits; ROBDD; average power consumption; binary decision diagrams; combinational circuits; fast power estimation capability; global propagation power analysis; low power chips; probabilistic power analysis; reduced ordered BDD; Boolean functions; Circuit analysis; Circuit simulation; Combinational circuits; Data structures; Energy consumption; Phase estimation; Research and development; Size control; Switching circuits;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Custom Integrated Circuits Conference, 1996., Proceedings of the IEEE 1996
Conference_Location :
San Diego, CA
Print_ISBN :
0-7803-3117-6
Type :
conf
DOI :
10.1109/CICC.1996.510583
Filename :
510583
Link To Document :
بازگشت