The benefit of 32 bit DAC's with 24 bit source material should be minute, similar to dithering. Presumably the extra 8 bits will just be assigned all zeros and thus not contain any real data anyways. An extra bit or two can come in handy for the multiple serial calculations (which are likely with more sound processing, i.e. room EQ) to prevent build-up of error, and may give better tolerance to noise, but the difference certainly won't be as significant as the jump from 16 bits to 24 bits. Law of diminishing returns. I'd be impressed if someone could hear the difference between 23 bit vs 24 bit source material on an ultra-high-end system anyways.

Needless to say, 32 bit processing will give no benefit over 24 bit processing when using a 16 bit source.

As mentioned above, the implementation (and supporting circuitry) of the DAC can be far more relevant to the quality.