Not all benchmark functions have the same range. But I think that you need to minimize them. So the value 784 is ..way too far from the minimum of each function.

Some of the schemes you mentioned are not SI (you mean Swarm Intelligence, right?), but you can always compare these kind of algorithms (Nature Inspired) by giving fixed number of iterations, population and specific values of the parameters that each algorithm has.

That's what I'd like to know, too - how the comparison is made is crucial. By the way, these test problems are not really "benchmarks" anymore - maybe they were in the 1980s. 

However, in literature are still used.

The performance of an algorithm can be measured depended on how close of the known minimum it gets.

Then the literature needs to grow up, in my opinion.

No-one is *really* interested in solving those problems, so focus needs to shift to those methods that could potentially solve much harder or larger problems.

@Michael Patriksson. Thanks for suggestion. Could you tell what could be used instead of benchmark functions?

@Alexandros Tzanetos. Thanks for suggestion. I know that different SI(Swarm Intelligence) techniques have different range but for comparison I am keeping fixed range of [-10,10],iterations=100 and such other parameters. But Since I received value as 784,I was confused whether it was True?

@Lokesh Gagnani. Let me correct you a bit: the SI techniques do not have range. A functions does.

As far as the value, you have to do multiple independent runs of the algorithm to see the average performance. However, I think 784 is a high enough value to consider that something is missing (while the known minimum is zero).

@Michael Patriksson. You are right. So the first step is these functions (to check if the algorithm does what it was programmed to do) and then the real challenge is real world problems.

In conclusion, dear Lokesh, I think that first of all you have to correct possible bugs in your code and check if the algorithms find a value near optimum. Secondly, evaluate their performance on a real world problem (based on what you aim to do in your research).

You can evaluate those optimizers according to several benchmark problems provided by IEEE Congress on Evolutionary Computation (CEC). There are some details that I think researchers in the field of optimization should be aware of them. Here, I list them again (I also have mentioned them for other researchers) because it can be very useful for you and other researchers. 

CEC'05: Evolutionary Real Parameter single objective optimization

CEC'06 : Evolutionary Constrained Real Parameter single objective optimization

CEC'07: Real-parameter MOEAs

CEC'08: large scale single objective global optimization with bound constraints

CEC'09: Dynamic Optimization

CEC09: real-parameter MOEAs

CEC10: large-scale single objective global optimization with bound constraints

CEC10: Evolutionary Constrained Real Parameter single objective optimization

CEC10: Niching Introduces novel scalable test problems

CEC11:  Real-world Numerical Optimization Problems

CEC2013: Real Parameter Single Objective Optimization

CEC2014: Real Parameter Single Objective Optimization (incorporates expensive function optimization).

CEC2014: Dynamic MOEA

CEC2015: Real Parameter Single Objective Optimization (incorporates 3 scenarios)

CEC2016: Real Parameter Single Objective Optimization (incorporates 4 scenarios)

CEC2017: Real Parameter Single Objective Optimization (incorporates 3 scenarios)

• Optimal reactive power dispatch benchmark problems: IEEE 30-bus, 57-bus and 118-bus test systems
• Truss design problems: 10-bar plane truss, 25-space truss, 72-bar truss, 120-bar truss dome, 200-bar plane truss, 26-story truss tower
• Non-truss design problems: Welded beam, Reinforced concrete beam, Compression Spring, Pressure vessel, Speed reducer, Stepped cantilever beam, Frame optimization