Comparison of Multiplication Implementation in MIC-1 with that in MAC-1.

Development of a MULD M instruction implemented in MIC-1:

       sum = 0;
       for (i = 0; i < 15; i++) {
          if (high bit of a is 1)
              sum += b;
          sum += sum;
          a += a;
       }
       if (high bit of a is 1) 
          sum += b;

Pseudocode MULD implementation:

The operands are popped from the top of the stack and the accumulator holds the final result. It is assumed that the result fits into 16 bits.

         pop multiplicand from stack and put in register A
         pop multiplier from stack and put in register B
         I = -1     (Keep I in register C)
         sum = 0    (Keep sum in the accumulator)
     top_loop:
         shift sum left
         if (b is Negative) go to add_in:
         go to shift_B
     add_in:
         sum = sum + a
     shift_B:
         left shift B
         left shift I
         if (I is ZERO) go to 0
         go to top_loop

NOTE: The ac has no special role at the MIC-1 level so we don't need to reserve it for performing operations. (This is a point that often confuses people.)

MAC-1 instruction encoding:

         11111110yyyyyyyy                DESP y         
         11111111--------                MULD x
         0000xxxxxxxxxxxx

Analysis of the MIC-1 implementation:
- Fetch and decode (16 cycles): 0, 1, 2, 28, 40, 46, 50, 65, 73, 76, 80, 81, 82, 83, 84, 85.
- Execute (69 cycles):
  - Preloop (2 cycles): 86, 87
  - Loop (16 times 4 cycles): 88 through 92
  - Post loop (3 cycles): 93 through 96
Analysis of the MAC-1 implementation :
- Preloop: 7 + 12 + 12 = 31 cycles
- Loop [MULTLOOP to DONE] 164 times 15 iterations = 2460 cycles
- Postloop: 10 + 8 + 10 + 10 + 9 + 10 + 11 + 12 = 80 cycles
Total = 2571 cycles.