Here are some numbers: With a nominal -40 dBu test tone at 400 Hz into the input of my Rumble 100 combo, all controls flat (12 o'clock position), I plugged the XLR output into a balanced input of a line level mixer and set the level to read -6 on the VU meter. Rotating the master clockwise from 12 o'clock to 3 o'clock raises the level by 9 dB to about +3VU. Rotating the master counter-clockwise from 12 o'clock to 9 o'clock reduces the level by about 12 dB to -18VU. So under these admittedly primitive test conditions, the sweep from 9 o'clock to 3 o'clock induces a change of about 21 dB (with steady-state tones, VU are equivalent to dB for all intents and purposes).
BUT, if you set the level on your sound reinforcement system after ballparking the master at, say, 12 o'clock, you're only changing that level by +9 or -12 dB even with fairly substantial changes in the master level. If your mixer channel fader is set at the unity gain position, there would be no adverse effects on clipping by raising the amp's master level as above, or on audible noise by lowering the amp's master level, with just about any well designed mixer. It's noteworthy that rotating the master control past 3 o'clock to the extreme clockwise position has little to no effect on the measured output level from the XLR line output (less than 2 dB), with my 100. That may be a result of the built-in delta-comp limiter, but with a -40 input signal, I doubt it.