I've never seen a comparison between the two on the same card so I've had a go at it myself. I only bought an S2 originally because I couldn't find an S1 rev 2 in stock. The S2 won't be wasted as I'll try it on an older card.
The graphics card I tested on was a Leadtek 9600GT. The system it was tested in was the one in this thread in the gallery:
http://www.silentpcreview.com/forums/vi ... highlight=
The temperature in the case just behind the air intake was ~21C. Air is fed through the system by a pair of 120mm fans meaning the graphics card is a semi-passive setup.
S1 rev 2 31C
Load max (looped Crysis benchmark):
S1 rev 2 54C
There are other differences between them. If anything the S1 was easier to install being more stable on the GPU during screw installation. There was no difference in memory cooling (I didn't swap them) and no VRM cooling as none fitted.
From what I've seen there's no point in going for the S2 if you can use the S1. The price difference is less than a decent fan which makes the optional Turbo module even more pointless on the S2.
There is still very much a point in the S2's existence. The mounting hole positions are different which means it will fit more cards. It has a pair of slanted holes at 55mm centres which were a very popular mounting some years ago. Many Geforce cards from the original Geforce256 through to the 5600 used them and a lot of Radeon's from 7000-9000 did too. The sets of square diagonal holes are also different with sets for Geforce FX 5700 Ultra's, 5800 and 5900's, Geforce 4 Ti. None of these cards would be direct6ly compatible with the S1.