Approximate Booth Multipliers using Compressors and Counter

Approximate computing is a technique to elevate performance of digital circuits which curtail hardware requirements such as digital-signal processing, multimedia, and machine-learning. Approximate multipliers and different techniques which can benefit from the use of such convolutional neural networks. The proposed method uses multi- level compressors such as 4:2, 5:2 and 6:3 counter to build truncated booth multiplier. Compensation circuit is provided by changes in k-map, eliminating carry from missing region. When the mapping is done correctly output errors and hardware errors may both be decreases. Even the most modern multipliers cannot compare to TACBM outcome in terms of accuracy and area-power savings.