That sounds backwards to me. EDIT: Sorry, I think I misread your comment. Yes, I think you have it right.
A performance contains 16 parts-- normally associated with each of the 16 MIDI channels-- although you might not need to use all 16 parts. If you're playing notes with one of the parts, you can sustain those notes and then switch to playing a different part. In that scenario, the notes sustained with the old part can continue to sound without any interruption as you play with the new part.
However, if you're playing notes with one of the parts and sustain them, then switch to a different performance, the synth will need to load all of the new voice selections, voice parameter settings, effects selections, and effects parameter settings for the 16 parts of the new performance. I believe this will usually cause the synth to cut off the notes being sustained from the first performance.