Among other factors, you might consider how many clock cycles are needed and how any two-step schemes are protected in the case of CPU interrupts.
A3:
How, for instance, is branch range determined? About how many bits would be required for a subcode to encode both integer and floating-point relations?
A4:
Yes.
A5:
Not equal is not equal (i.e., to zero), independent of interpretation of the bit pattern.
A6:
How many predicate registers are available (see Appendix D)? Could an inner compare instruction "unconditionally" modify the same predicate register that predicated that instruction itself?
A7:
It falls through only on the final traversal.
A8:
Infinite loop.
A9:
Consider the program in Figure 5-1 and substitute a subtract operation for the multiply operation. What other modifications would be needed for the specified situation?
A13:
What is the maximum amount of multiplication achievable using one shladd instruction?
A15:
Either A or D. Why?
A18:
Begin from MAXIMUM as a model program, but look instead for a minimum value.
