Question

Reorder the following code snippet to minimise execution time for the following configurations:

a . We have software technique , and have 2 delay slots.

b. We use interlocks , and predict not taken.

c. We use forwarding , and predict not taken.

add r1, r2, r3

sub r4, r1, r1

mul r8, r9, r10

cmp r8, r9,

beq .foo

add r1 and all is assembly language

Answer 1

add r1, r2, r3

mul r8, r9, r10

sub r4, r1, r1

cmp r8, r9,

beq .foo

We do such that execution can take Place Parallely in the above snippet given in question sub will have to wait for add since sub uses r1 result from addition.So,we will rearrange it such that it can be executed parallely.

In cmp after mul cmp uses result of r8 which will be the result of multiplication.So,we will rearrange it also, so that it won't depend on the previous result.