Abstract: The paper investigates floating-point and logarithmic number representations for computing with FPGAs. The key issue is to select the best number format for an application to improve ...