Chapter 8

A1:	Corresponding mathematical operations have the same numbers of operands, similar branching choices would be desirable, etc., so it makes sense to structure the instructions to be parallel, even if they cannot share much actual digital logic.
A2:	b. fmerge.ns f1=f3,f3; d. fmerge.s f1=f3,f3; f. same as d using Fr1 for both sources.
A3:	How is the second destination register accommodated?
A4:	Consider by analogy adding 0.03 three times onto 7.1; stepwise, with rounding, the result would be 7.1 considering the implied uncertainty beyond the first decimal place. Now consider adding 0.03 + 0.03 + 0.03 = 0.1, onto which 7.1 is added for a result of 7.2. Now can you do the binary case?
A5:	The absolute spacing between successive numbers jumps by a factor of two every time the binary exponent has to be incremented by one.
A8:	An integer has only one sign, at the left; the segmentation of the operands into, say, left and right halves is a computational convenience.
A9:	What is the precision of the binary significand for that format?