we can increase the utilization of silicon by overlapping some instructions with pipelining.

number of machine cycles per instruction = pipeline depth
typically pipeline depth is 5

other problems:

solution: introduce NOPs into the pipeline

examples of NOPs:

using these NOPs, the above may be rewritten:


what are some of the problems with this implementation of NOP?
structural hazards?
use of A and B, etc...
note that on many microprocessors, R0 is always 0, therefore can be used without structural hazard, e.g. AND R0 R0 R0.

a better solution: do some useful instruction instead of a NOP:

continuing onward (store register C):

final example