Yes, I believe that the amount of input and output variables can impact processing time. This also adds to the amount of fuzzy rules created. The more input, output and rules, the greater the number of times that it will be necessary to aggregate the membership degrees. However, I do not believe that the impact over time is very high given the simplicity of the operations.
As for the membership function, I believe that the type adopted does not influence the processing time. The number of linguistic variables can influence depending on how many were active in the process. If several linguistic variables are active for a given input value, the greater the number of times that the membership degree aggregation will be done.