pipeline

Question

Answer 1

Average instruction execution time on the nonpipelined machine is

Clock cycle × Average CPI = 

                        10 ns × ( ( 40% + 20% ) × 4 + 40% × 5 ) = 10 ns × 4.4 = 44 ns 

In the pipelined implementation, the clock must run at the speed of the slowest stage plus overhead, which will be 10 + 1 or 11 ns ; avg instruction execution time.

Speed up  = avg time in pipe/ avg time in non pipe =  44/11 = 4 times

options (c) 

Comments

Yes,I would like to add

Average CPI in pipelined instr is 1 CPI

so CPI*Clock cycle time=1*(11)

Answer 2

Without pipelining :

Time = 10ns

With Pipeline :

Total no of cycle = Frequency of ALU*no of cycle for ALU + Frequency of Branch*no of cycle for Branch + Frequency of Memory*no of  cycle for Memory.

= 0.40*4 + 0.20*4 + 0.40*5 = 1.6 + 0.8 + 2

= 4 cycle

Cycle time = 10ns + 1ns = 11ns

Total time = 4*11ns  = 44ns

Speed UP = time in pipe/time in non pipe.     

                  = 44/10 = 4.4

OptionOption D.