No doubt. However some fraction of the performance improvement between geforce and quadro is tied to the drivers. What said fraction happens to be is harder to judge, I don't suppose you have a fx3700 on an i7 motherboard that you could run a benchmark through?
thats more likely down to the fact that a workstation gpu driver is focused on work software rather than games and as such is more highly focused on delivering the highest quality performance in set environments rather than x number of games. It's only reasonable that a gpu used for work has drivers which are better suited to that task than those from a gpu aimed at games.
And no I haven't got a quadro fx3700 + i7 to benchmark or similar at present, my x2's wouldn't be a fair test.
What I can say with confidence is that benchmarks in maya, pro engineer etc show considerable improvements with the quadro drivers relative to the geforce ones. I can also offer the subjective opinion that it's nicer to work with when using the quadro drivers, and seems to cope with more complex models before grinding to a halt.
Like I said benchmarks mean squat, they're only useful for comparing like for like models in a range, you can't compare with last gen etc.
How much of your subjective opinion is down to the fact that you 'believe' it is faster due to the softmod.
Considering all 'benchmarks' I've seen for a softmod versus a full quadro show that the quadro is usually still twice as fast as the geforce with a softmod just proves to me you can't make a geforce into a quadro (you could with a gf6 series).
You've also got the possibility of additional issues being introduced as a geforce has like I said cut pathways compared with a quadro fx.
edit: fx3700 benchmark using a c2q (so slower) cpu is here -
link
fx1800 (newer but 1 down pecking order) benchmark using i7 870 is here -
link
your link shows sw01 as 39.42, these show 98.54 (fx3700) and 131.68 (fx1800) respectively
lol
when theflash looked as thogh it had worked!