Problem 2: Given the following code:

LOOP:

lw $5, 100($1)

lw $6, 800($1)

add $5, $5, $6

sw $5, 100($1)

addi $1, $1, -4

bne $1, $Zero, LOOP

Assume the initial value in $1 equals 400, such that the loop executes 100 iterations.

1. Execute the code on a pipelined computer based on the following assumptions:

i) Forwarding is not available;

ii) The branch decision is made during the EX phase, and if the branch is taken, the new PC is available by the end of the MEM phase (Fig. 4.51);

iii) Do not alter the code sequence.

a) Filling in the slots of the following table. Insert stalls where they are necessary. b) How many cycles does the whole loop take to execute? Show your calculation.

**This is the lw in the next iteration.

2. Do 1. again based on the following assumptions:

i) Forwarding is available;

ii) The branch decision is made during the ID phase, and if the branch is taken, the new PC is available by the end of the ID phase (Fig. 4.62 or 4.66);

iii) Do not alter the code sequence.

a) Filling in the slots of the table above. Insert stalls where they are necessary.

b) How many cycles does the whole loop take to execute? Show your calculation.

3. Do 1. again based on the following assumptions:

i) Forwarding is available;

ii) The branch decision is made during the ID phase, and if the branch is taken, the new PC is available by the end of the ID phase. Re-schedule the instructions in the loop so that the total number of clock cycles is minimized.

a) Filling in the slots of the table above. Insert stalls where they are necessary.

b) How many cycles does the whole loop take to execute? Show your calculation.

Instruction lw S5, 100($1) lw $6, 800(S1) add S5, $5, $6 sw $5, 100 ($1) addi $1, $1,-4 bne S1, SZero, LOOF **lw S5, 100(S1) 123 4 5678910 11 12 13 1415 16 Instruction lw S5, 100($1) lw $6, 800(S1) add S5, $5, $6 sw $5, 100 ($1) addi $1, $1,-4 bne S1, SZero, LOOF **lw S5, 100(S1) 123 4 5678910 11 12 13 1415 16