AFAIK it's the native apple driver that's reporting the latency to the RTL Utility.
The RTL utility then measures the actual latency and compare them.
In an optimal system reported latency should equal the measured latency.

RTL Utility is a tool for measuring Round Trip Latency in audio interfaces.
You can find it at https://oblique-audio.com/
You hook up your audio interface and connects your input to your output and let the utility measure the round trip latency. You can do the same in your DAW but I find this easier to work with.

No the reported RTL was greater than the measured.
For example at 44.1 khz with a buffer size at 64 samples the reported RTL was 411 samples and the measured RTL was 336 samples.

Waedi, download RTL Utility and play with it - this saves us all some time and obsolete posts.

Benzinger, we are not aware of a mechanism to send the latency numbers to macOS in CC mode, so the reported numbers are fake, we don't send any.

It's possible to override the latency figures published by Apple's CC USB driver for a device using a codeless KEXT, which is what Focusrite allows with some of its CC interfaces.  See this link:

https://support.focusrite.com/hc/en-gb/ … ace-on-Mac

Apple talk about customising some properties of their CC driver at the bottom of the following page under "Ways to Extend the Class Driver":

https://developer.apple.com/library/arc … index.html

And the sample code for the "SampleUSBAudioOverrideDriver" codeless KEXT is in the following link:

https://developer.apple.com/library/arc … Intro.html

EDIT:  Of course this is of limited use with RME devices and their custom drivers, and having to install a codeless kext driver kind of defeats the point of CC mode for devices with their own drivers, but I thought I'd point out that Apple does indeed provide a mechanism for customising its CC USB driver, including a CC device's reported latencies ...