@Min Hee Jo I could be wrong, but here is my interpretation of the question: You have to calculate the execution time ratio first before you calculate the geometric mean. ExecutionTimeRation = ExecutionTimeReference / ExecutionTime. My guess would be also that CPU A is the reference here. So for CPU B its 40/15=2,6 and 40/1000= 0,04. The geometric mean of that sqrt(2,6 * 0,04) = 0,326. If u do the same for CPU C you get 0,7302.
The correct answer is on my opinion A. The correct formula is ExecutionTimeRatio = ExecutionTimeReference / ExecutionTimeNew. if it is normalised to CPU_A then Geometric mean CPU_A = 1 Geometric mean CPU_B = 0.326 Geometric mean CPU_C = 0.73 Biggest value is A therefore it is the fastest.
you should add units... seems not general without units
In the first Question is B better as these numbers are performances and not time ?
I think the correct answer is C. I normalized to A and as a result C has the biggest geometric mean of 1.369.
@Min Hee Jo I could be wrong, but here is my interpretation of the question:
You have to calculate the execution time ratio first before you calculate the geometric mean. ExecutionTimeRation = ExecutionTimeReference / ExecutionTime.
My guess would be also that CPU A is the reference here. So for CPU B its 40/15=2,6 and 40/1000= 0,04. The geometric mean of that sqrt(2,6 * 0,04) = 0,326. If u do the same for CPU C you get 0,7302.
@Min Hee Jo Hmm in "1 1 4 Summarizing Performance" he defines it the other way around.
The correct answer is on my opinion A. The correct formula is ExecutionTimeRatio = ExecutionTimeReference / ExecutionTimeNew.
if it is normalised to CPU_A then
Geometric mean CPU_A = 1
Geometric mean CPU_B = 0.326
Geometric mean CPU_C = 0.73
Biggest value is A therefore it is the fastest.