Multi-Sub Optimizer (MSO) doesn't actually do room simulation at all. It figures out individual EQ, delay and gain for each subwoofer in order to get the flattest possible low-frequency responses of mains and subs together at multiple seating positions. Its input is a set of text files of frequency-domain low-frequency responses of main speakers and each sub at each listening position. The user performs these measurements using e.g. REW prior to running MSO. For example, if you energize the left and right main speakers together and have four subs, and want to do four seating positions, this would require 20 measurements. These would be mains together, Sub 1, Sub 2, Sub 3 and Sub 4 at each of four listening positions.
If, for example, you did these 20 measurements and let MSO do 1 million iterations to find the best configuration, it would be as if you had done 4 million measurements of mains and subs together (1 million EQ configurations, 4 listening positions) to find the best configuration.
Because it uses individual EQ for each sub, it is able to reduce seat-to-seat response variation, something that can't be done when you EQ all subs together.