Answer:
Explanation:
As per Amdahl's law :
[tex]\text {Speedup} = {\frac{\text{Old Execution time}}{\text {New Execution time} }[/tex]
[tex]\text {Speedup} = \frac{1}{( (1- \text {FractionEnhanced}) + (\text {FractionEnhanced} / \text {SpeedupEnhanced}) )}[/tex]
Here :
Design 1:
FractionEnhanced = 0.5 (50% of computation )
[ Note that I/O wait has nothing to do with speed ]
SpeedupEnhanced = 15 times
[tex]\text {Overall speedup} =\frac{1}{( ( 1- 0.5) + (0.5/ 15) )}[/tex]
[tex]\text {overall Speedup} = \frac{1}{(0.5 + 0.033)}[/tex]
[tex]\text {overall Speedup} = \frac{1}{ 0.533} = 1.876[/tex]
========
Design 2:
FractionEnhanced = 0.3 (30% of computation )
SpeedupEnhanced = 20 times
[tex]\text {overall speedup} = 1 / ( ( 1- 0.3) + (0.3/ 20) )\\\\\text {overall speedup} = 1/ (0.7 + 0.015)\\\\\text {overall speedup} = 1/ 0.715 \\\\\text {overall speedup}= 1.398[/tex]
========
So as we can see Design 1 is better with overall speedup of 1.876 times the original processor.