Topic: Fireface UC / Babyface latency

I'm considering either a Fireface UC or a Babyface for live work, and I'd like to know the real world amount of safety buffers on those on both OS X and Win 7. There's a little app called HALlab.app (or the audio configuration page in AULab.app) included in the OS X developer tools that gives exactly those kinds of information about an audio device. Could someone here (or a mod) please run it and tell me what it reports? Or some loopback test results with the Centrance utility under Win7. Thanks!

Re: Fireface UC / Babyface latency

Search is your friend! tongue

http://www.rme-audio.de/forum/viewtopic.php?id=7235

Re: Fireface UC / Babyface latency

Thanks Timur tongue

I came across that some time ago, but yesterday I was reading the PDF manual, and it's not as clear on that one. It talks about the 32 samples safety buffer, an extra 28 samples on OS X without saying if it's on input or output, etc.

Just to make sure then, a complete AD/DA trip with both the FF UC and the Babyface under OS X is 69 + buffer + buffer + 70? And 55 + buffer + buffer + 63 on Win7? Do the values stay the same when increasing the sample rate?

Re: Fireface UC / Babyface latency

Here's a more informative list with the readings of AU Lab. These were made a year ago or so with an older Firmware and driver though, so values could have changed (there was an overhaul of the OS X driver for both the FF and UC).

"Hardware" means ADC/DAC conversion plus some internal processing/buffer offset that is *not* the Safety Buffer.

Input OS X 10.5

HDSPe: Hardware 45, Safety Offset 24
FF 400: Hardware 45, Safety Offset 64
FF UC: Hardware 45, Safety Offset 24

Output OS X 10.5

HDSPe: Hardware 79, Safety Offset 24
FF 400: Hardware 32, Safety Offset 64
FF UC: Hardware 46, Safety Offset 24

These are the latencies your interface begins with *without* audio buffers being added. So in the above example a Fireface UC with a buffer setting of 64 samples would sum up to 46 + 24 + 64 = 134 samples = 3.04 ms output latency.

And yes, the complete round-trip latency is just as you calculated. Windows seems to have a slight edge here, but it's rather small and you have too keep in mind that OS X allows to use audio buffers down to 16, while the lowest in Windows is 48.

With the current drivers the audio-buffer size is doubled when you double the sample-rate, I don't know about the safety-buffer though. The "Hardware" part is mostly ADC/DAC (except for the HDSPe output and FF input on OS X) so it should not change. The latter values are listed in the manual for all sample-rates (in ms).