Aparently, 32bit float also has the advantage of being less precessor intensive because most computer processors themselves run at 32bits.
Not true. Floating point operations are more processor-intensive than integer operations.
By the way, when you choose to "record at 32 bits," you are not getting 32 bit resolution out of your 24-bit ADC. It just means the data is stored in the file as 32-bit floating-point values instead of 24-bit integers. The advantage is that anything you do whcih involves processing by the software -- mixing, adding plug-in effects, etc. -- takes place at the highest-possible accuracy with the smallest-possible roundoff error.
So yeah, it's beneficial if you are doing all your mixing and such "in the box," as Teacher noted, but pointless if your computer is just the recorder and you send the signal out to mix with outboard gear in the more traditional way.
This same issue came up recently in another thread about the SB Live and the kX drivers and CEP, where some guy thought he was getting 32-bit, 96 KHz data out of the SB Live's converters...
Beneficial is relative, though. You will be able to record fewer tracks and use fewer plug-in effects overall because of the added resource burden. And if you are not recording fantastic quality instruments with tons of high-frequency components (bells, piccolos) with absolutely top-drawer microphones that can capture information in this part of the spectrum, anything you gain from 32-bit float processing (and indeed, many would argue even with 24-bit over 16-bit) is imperceptible.
