Precision Modeling and Bit-width Optimization of Floating-Point Applications

We present a floating-point precision modeling methodology that can be used to develop application adaptive arithmetic precision models for variable bit-width floating-point computing. We also develope