Some issues related to double rounding (Q2434938)

It is well known that the calculation on the computer is slightly different from the calculations on paper with a pencil. Computer arithmetic only approximates real arithmetic. At each step of computing it is dealing with rounding. Double rounding is a phenomenon that may occur when different floating-point precisions are available on the same system. The potential influence of double rounding on the Fast2Sum and 2Sum algorithms, on some summation algorithms, and Veltkamp's splitting is analysed in this paper.