In this paper, we present a pipelined, high throughput single precision floating point implementation of the exponential function with a latency of 19 cycles. We also present the Application Specific ...