This tutorial presents system level design approaches for implementation of high-performance and low-power DSP systems. The emphasis is on tradeoffs with respect to area, power and speed using design approaches ranging from algorithm level to architecture level to implementation levels. The topics covered include transformation techniques like pipelining, parallel processing, sub-expression sharing, retiming, and number splitting. Experimental results on power consumption using different implementation styles like carry-save and redundant arithmetic are presented. A novel cell replacement transformation is also presented which results in reduction of power consumption in multipliers by a factor of 15.
|Original language||English (US)|
|Number of pages||20|
|Journal||Proceedings - IEEE International Symposium on Circuits and Systems|
|State||Published - Jan 1 1997|
|Event||Proceedings of the 1997 IEEE International Symposium on Circuits and Systems, ISCAS'97 - Hong Kong, Hong Kong|
Duration: Jun 9 1997 → Jun 12 1997