You might have already realized that there is more than one method because there are different opinions about how capability and performance should be calculated and interpreted. Does your organization have a standard? If so, then I can help you pick the appropriate method in JMP. If not, then I would use the default method in JMP.
We have recently been harmonizing the calculations that appear in different platforms so that they are the same. Which version of JMP are you using?
Finally, is your process data normally distributed? If not, which distribution model do you use?