Dynamic Fixed-Point Arithmetic: Algorithm and VLSI Implementation

from significant drawbacks such as limited dynamic range. A fixed-point hardware does not provideacceptable accuracy levels when simultaneously deals with large and small numbers. Although thefloating-point arithmetic greatly addresses this problem, it is not widely used because it faces importantchallenges when realizes in hardware. A novel computational paradigm named as Dynamic Fixed-Point(DFP) is proposed in this paper which provides improved precision levels while has a similar VLSIimplementation costs when compared to traditional fixed-point. The accuracy simulation results andVLSI synthesis costs of the new method is presented and compared with fixed-point to prove itsefficiency.