The next three exercises deal with the normalization shifter NORMSHIFT used by the floating point rounder FPRND. The functionality of the shifter is specified by Equation 8.6 (page 393); its...

The next three exercises deal with the normalization shifter NORMSHIFT used by the floating point rounder FPRND. The functionality of the shifter is specified by Equation 8.6 (page 393); its implementation is described in section 8.4.2.

The shifter NORMSHIFT gets as input a factoring s₁
e_r
f_r
; the significand f_r
[-1 : 55]has two bits to the right of the binary point. The final rounded result may be a normal or denormal number, and f_r
may have leading zeros or not.

Determine the maximal shift distanceσ for each of these four cases.

Which of these cases require a right shift?