The upcoming arrival of high dynamic range imageand video applications to consumer electronics will force theutilization of floating-point numbers on them. This paper showsthat introducing a slight modification on classical floating-pointnumber systems, the implementation of those circuits can behighly improved. For a 16-bit numbers, by using the proposedformat, the area and power consumption of a floating-pointadder is reduced up to 70% whereas those parameters aremaintained for the case of a multiplier.
展开▼
