The calculator is very wrong (off by a factor of 8 for a 4 meter seating distance for our discussion purposes) except for people listening out doors or in anechoic chambers who are not interested in the power needed to achieve reference level home theater.
_Sound Reproduction: The Acoustics and Psychoacoustics of Loudspeakers and Rooms_ summarizes various studies showing that indoor listening space volume neither looses 6dB per distance doubling nor acts like a diffuse field. It looses about 3dB.
To use the calculator for indoor use one should take the square root of the listening distance (ex : square root of 2 = about 1.4, 4 = 2) and use that.
Reference level home theater assigns a SPL to 0dBFS, which is the highest possible signal with a 0dB crest factor. Consumer amplifiers are rated using sine waves with a 3dB crest factor (a sine wave with 100W RMS power into 8 Ohms has a 28.3V RMS voltage with peaks at 40V and therefore peak power 200W).
To apply the calculator to power required for digital movie sound tracks you must consider this and get rid of its headroom compensation which has already been accounted for by adding 3dB to the speaker sensitivity rating box and setting the Amplifier Headroom box to 0dB (all negative values are treated as zero).
It takes less power than you'd think.
After reading Siegfried Linkwitz's opinion that rooms were like diffuse fields (once you get past the critical distance where direct sound and reverberant field strengths are equal the total SPL can't drop more than 3dB; with 2-4 feet matching the directivity index of typical speakers) I got curious and made some measurements with sub (corner loaded) and center (on a short stand 3' off my screen wall) both with claimed sensitivity a bit over 90dB/2.83V/1 meter and nominal impedance around 6 Ohms. My room was 13x19x8' with big openings to the rest of the house, seating 11' off the screen wall. The sub was about 12' away and center 9' away measured to the drivers. The sub was a Citation 7.4 with a 14.5" driver in a ~3 cubic foot ported box and should be typical for reasonably sized subs; the center a Definitive C1 which should be typical for a medium consumer market sized MTM speaker. I used my Rat Shack SPL meter with C-weighting so sub-output may be a little under reported. I used an HP true RMS volt meter which is spot on at no worse than -2% against the calibration terminal on my Tek scope which is .17dB. The sub tune is around 30Hz so the voltage within the sub's pass-band may be over-reported (pink noise has equal power in each octave and I'm missing at least part of one, and I never looked at the frequency spectrum from my Lexicon DC-1 to see where its cutoff was). Any errors led to sub sensitivity being measured as less than it really was. Extrapolation to reference level output disregards thermal compression although the Citation 7.4 uses a JBL pro-sound driver with a 4" voice coil and shouldn't suffer too much arround just 100W.
My mains were actively tri-amplified dipoles that would be too messy to make sense of so I didn't measure them.
I measured .88 VRMS for pink noise at 75dB SPL (at my listening position) from my 6 Ohm center which is .13W. A 101dB peak (reference level Dolby Digital main channel maximum with typical encoder settings) would take 50W; although that's 0dBFS and could be done with an amplifier rated at 25W into 6 Ohms or 19W into 8 Ohms.
I measured .34 VRMS to produce pink noise at 76dB SPL (at my listening position) from my 6 Ohm corner loaded sub-woofer which is .02W. A 101dB peak would take 6W, 111dB (reference level Dolby Digital LFE maximum with typical encoder settings) 60W; although that's 0dBFS and could be done with an amplifier rated at 30W into 6 Ohms or 23W into 8 Ohms.
In my room, 125W and 150W respectively would get me a full 105/115dB on a potential DTS track track with the volume knob in the same position and take amplifires good for 47W and 56W into 8 Ohms for center and sub.
LFE headroom is 10dB over the screen channels so the voltage is 10 ^ (10/20) = 3.2 times each screen channel and mixing in five screen channels with correlated bass (and nothing else happening!) could boost the sub-woofer output by 20 log ((3.2 + 5) / 3.2) = 8.2dB or multiply the power requirement by 10 ^ (8.2/10) = 6.6X thus suggesting a sub-woofer amplifier good for 370W into 8 Ohms.