Energy Efficient Approximate Bfloat-16 Floating Point Division Hardware for Image Processing

Energy efficient designs are need of the hour and image processing applications are error tolerant application. Where the error present in the computing does not impact the output visual quality. This work proposes an efficient approach to design a Bfloat floating point division. This work proposes three algorithms that are implemented at the hardware level. All the proposed algorithms will produce lesser error in the output and does not impact the output visual quality. The proposed work reduces the energy sianlficantly than the existing works.