For me it was 67% if windows control panel set to 5.1 speakers and 91% if speaker config set to "stereo" (windows control panel). SBX worked better with stereo speakers + 91% than 5.1 with 67% though to my ears but I'm unsure whether for most games simply using 5.1 speakers with SBX disabled vs stereo speakers + 91% SBX -- like which truly was better.
With 5.1 speakers it sounded most natural at 67%, moving to either ends just seemed to get more unnatural positioning of things. When using Stereo speaker config again the sound/staging with SBX becomes quite different (more natural sounding over headphones at least) and I really felt SBX worked better with stereo only with slightly less impact on SQ too. It's a bit shame though because a large amount of games I play sound best with 5.1 speakers so that SBX works less good if using 5.1 was a bit dissappointment for me. I need virtual processing configs that ideally should also be optimized for 5.1 speaker config use because I use that a whole lot of time and I don't want to constantly keep swapping around configs depending what I'm doing.
To test whether which setting worked best for surround I simply used Windowed mode with UT3 with hardware openal disabled so that I can better test software processing and kept running around for a little while and tested different values until I got something which seemed most accurate for me.
NOTE: Unreal Tournament 3 is probably among the ideal testing scenarios because it allows you to record "demos" at which you will be "playbacking" the exact same scene with the exact same things happening on the screen, perfect for making comparisons with different software sound processing settings.
Check this video I once made between Audigy 2 ZS (with 3rd party kX Audio drivers) vs Titanium HD:
Edited by RPGWiZaRD - 8/19/13 at 1:29am