The run time on the MRIComparisonWU would depend on the size of your network, number of measures and which measures you will calculate, but it will also be highly dependent on the number of permutations that you choose to perform.
This issue arises due to the methods used to calculate the null distributions for the MRI covariance networks. Namely, the permutations are performed at a subject data level. For each permutation, two new groups are formed and the measures are recalculated for each group.
Therefore, I would suggest calculating your data measure by measure starting from a small number of permutations and then gradually increase the number of permutations. This should give you a robust estimate of how much it will take to calculate all measures. As a rule of thumb, the measures related to global and local efficiency as well as the small-worldness measure take longer computational time.
Please let me know if there is anything I can do to help additionally.