On the use of the Infinity Computer architecture to set up a dynamic precision floating-point arithmetic.