CPFloat, a library for simulating low precision floating-point arithmetics with various rounding modes, was released today. This was developed in collaboration with Massimiliano Fasi. Preprint is available here.