Abstract— Convolution neural network is basically an algorithm which is used to train a model to perform a specific task, which is inspired by the working of the human brain. Nowadays, its being extensively used in various applications. Its implementation over the FPGA board not only reduces the development time, but also increases the programmability. In this paper, Fast Convolution unit is described, which are based on parallel fast finite impulse response algorithm (FFA). This paper also includes the derivation of the algorithm.