From Audioasylum  http://www.audioasylum.com/forums/pcaudio/messages/8/81264.html

I think there are different "64 bit" things going on, not to be confused:

1. The precision used in the DSP calculations (e.g. filters). Here 64 bit floating point is essential because it provides the needed 48 bits of resolution to reduce round-off errors in the complex calculations,

2. The precision used in audio file formats (e.g. 24 bit fixed point, 32 bit floating point). Some software uses 64 bit internal calculations but "bounces" the final data to 32 bits before going into a file,

3. The address size used in CPU instructions, e.g. 32 bits in length or 64 bits in length. You will need a 64 bit processor to directly address more than 4 GB of RAM memory.

It is possible to use 64 bit calculations and 64 bit file formats in with a 32 bit processor and operating system. Indeed, the 32 bit processors will also do 64 bit calculations. It's only if you have very memory intensive applications that there is much to be gained by going to a 64 bit system, and the extra size of instructions is likely to actually reduce performance unless the extra memory capability is exploited. I run the 64 bit iZotope SRC in Soundforge Pro 10, and here it will support both 32 bit and 64 bit floating point. The (slightly newer) version in iZotope RX Advanced uses 64 bit calculations, but doesn't support the 64 bit floating point in the file format. A further confusing factor is that the WAV file format itself is limited to 32 bit addressing in the file (4 GB) if the standard Microsoft WAV format is used. (There are variants that allow for extended addressing.)

Tony Lauck

Peter: And indeed, only a matter of the volume control being in there (though not active at all) is enough to "destroy".

As in two channel two-way?

The DAC is connected to the PC. For now and some visualisation, let's say it is proprietary i2s. This one cable can carry 8 cannels of high speed data. One channel could be a normal left speaker, another channel the right.

.. while the xover math always will be done in the PC (nothing strange here). But, like with normal XXHighEnd operation, the processing and inherently degrading (SQ) effect of such a thing - will be covered for by XXHighEnd, and playback will be as good as always. But, now spread over (max) 8 channels, each with their own bandlimit and roll offs.

For me this means I can't hold up the further process just because of one person. But, obviously it means that you will have to wait, but it will be for your good reason.
Don't be afraid I won't do it, if you only imagine that the 8 channel version has been developed for one person only also, and it really took a very large deal of the development time (still not finished, and he will have to wait some longer too). But, I seriously think there will be quite some people wanting it, if you only know what it can bring for sound quality (see previous post ... I'm fairly sure).
When the headphone output is there, more people will take it, I'm sure. But don't underestimate this feature, because doing it really well is quite costly, and I won't go for less than "really well". Could be in between 350 - 500 Euro.

There's 2 channel input and 8 (or 6) channel output (via XXHighend, the software to be programmed yet).

I forgot to let create a (dummy) headphone output in the prototype of the cabinet and it really should be there for later.