Hardware reduction of DSP kernel Data Path using Carry Save Arithmetic operation in Fused Add Multiply Add unit

摘要

A high-performance data-path to implement DSP kernels is introduced in this paper. The proposed architecture of data-path is realized by FAMA (Fused Add-Multiply-Add (FAMA) unit). We differentiate from previous works on flexible computational unit using addition and subtraction process it takes that extra hardware. A mapping methodology, for datapath composed with the proposed flexible unit.it exploits the features of proposed units and enables fast computations. Comparing them with the datapath which use existing schemes, the proposed technique yields considerable reductions in terms of area and power consumption in datapath.