Hello Andrey,
If you dare to question are chosen for high-quality sound system what would be your choice - MAC OS or WIN 7?
I personally use Macs running OS X, except when I need to develop and test on Windows, in which case I boot my Macs into Windows under Boot Camp.
My experience is that Macs running OS X provide a robust, almost compatibility issue-free, glitch-free platform for audio/MIDI that works and performs superbly 'out of the box' and remains that way, without any customisation, tweaking, maintenance or other technical intervention usually needed on the part of the user, since they're designed as a unit to achieve that.
PCs and Windows can of course perform just as well, but you might or might not need some technical experience, tweaking, troubleshooting, etc., to achieve it since generic PCs and Windows aren't normally designed or tested for reliable 'pro-audio'/MIDI performance. (If you buy a PC from a specialist company that makes, tests and installs PCs specifically for audio/MIDI/Hauptwerk then of course that should be fine because they should have ensured that everything's compatible and set up as necessary.)
And in what order do you recommend to use parameters with minimal loss of sound quality from the download menu sample?
Some general background information and recommendations can be found in the 'Loading organs' section of the user guide (pages 75-77 in the v4.0.0 guide).
People have individual preferences for how to load ranks within a certain amount of memory, and it will also depend on the nature of the sample set (wet/dry, etc.).
My personal preference for a wet set would be first to load with all default options (stereo, 16-bit, compressed, all attack+release samples, all loops, full release samples).
If there's still plenty of free memory, try 20-bit or 24-bit (might give less hiss in release tails for large wet sets), but the difference in hiss between 20-bit and 24-bit is probably extremely minor, so 20-bit is a good choice.
There's almost never any benefit in turning off memory compression, since it's entirely loss-less (no loss of audio quality at all). Memory compression typically saves 35-45% of memory and only reduces polyphony by about 10%, so I would always recommend keeping memory compression turned on (as it is by default).
If there isn't enough memory to load the sample set with all default rank options (stereo, 16-bit, compressed, all attack+release samples, all loops, full release samples) then my personal preference would be to try disabling rank features one-by-one until it fits, in the following order:
1. Multiple attack samples (except Salisbury, which needs them for tremulants).
2. Multiple loops.
3. Disable some ranks completely.
(Using mono or release truncation or disabling multiple release samples will have a big effect on the sound, so my preference is to avoid them.)
Hope that helps.
Best regards, Martin.
Hauptwerk software designer/developer, Milan Digital Audio.