The same caveats apply as for Windows legacy (and ALSA) outputs:
- Change latency is high and noticeable.
- Change between less than 100% and more than 100% occurs in two steps.
- Non-linear formats are not supported at all.
- Integer formats are prone to clipping/saturation.
- Floating formats may hit Windows peak protection, sounding awful.
In other words, it does not work properly and it never will. This will
be removed once minimum and maximum volumes are exposed properly.

As meager consolation, there are two advantages over endpoint volume:
- The code is much simpler.
- Other audio sessions are not affected.