• From: Sweden
  • Registered: 2008-03-14
  • Posts: 149

Topic: A few questions about Latency and Fireface UFX.

I'm trying to wrap my head around what the non-buffer roundtrip latency consist of in the Fireface UFX.
I'm doing all my tests with a Mac Pro connected to the UFX via firewire and using Logic 9.1.3 as my DAW:

Firmware: USB 343/DSP 1.34/FW 336
Firewire Driver: 2.91
USB Driver: 1.55
TotalMix FX: 0.94 MAC OS X Pre Release 6

At single speed I get a value of 149 samples when disregarding the buffer-size from Logic. (this is consistant regardless of the buffer size.)
The Safety Offset under OS X with firewire is 32 samples.
The Safety Buffer under OS X is 32 samples.

149-32-32-32=53 samples

Now to the part I don't understand.
According to your chart in the manual I should have a ADDA latency of 40 samples at single speed.
My measurements of this latency is not equal to 40 samples. When testing the Loopback function I got these results:

ADDA @ Single Speed:46 samples
ADDA @ Double Speed:43 samples
ADDA @ Quad Speed:40 samples

Question #1:
Why is there a difference of 6 samples (46-40=6)?
Has this something to do with the Loopback function?
The manual states that there is an Oversampling offset of about 2 samples for Loopback.


Question #2:
Looking back at the 53 samples messurement:

Eric Bradley wrote:

149-32-32-32=53 samples

What are those 13 samples difference consisting of (53-40=13)?
The manual states that there is a delay of 3 samples for digital reciever -> TotalMix -> transmitter.
How does it all add up?

Thanks!

/Eric

RME Fireface UFX
Mac Mini 2.6 GHz 16 GB Ram Areca RAID | MacBook Pro 2,2 GHz 16 GB Ram | OS 10.10.5
  • From: Sweden
  • Registered: 2008-03-14
  • Posts: 149

Re: A few questions about Latency and Fireface UFX.

Bump. journey
Matthias? Knut? Anyone?
I feel like I'm in Camerons "Aliens" - in space no one can hear your screams. wink

RME Fireface UFX
Mac Mini 2.6 GHz 16 GB Ram Areca RAID | MacBook Pro 2,2 GHz 16 GB Ram | OS 10.10.5

3 Reply by MC

  • MC
  • MC
  • Administrator
  • Offline
  • From: Germany/Thailand
  • Registered: 2007-04-06
  • Posts: 24,877

Re: A few questions about Latency and Fireface UFX.

That would be part 2 - can't remember here crying alone in the ship :-)

1. The latency differs hardware-wise in certain modes, due to the complicated routing and connections between FPGA and DSP. Whenever we find the time we will document this in detail. AFAIR the delay is 3 up to 7 samples, but don't quote me. Loopback is one of the mode that has its special latency behaviour. Someone else already asked about that and measured some (forum member Zavieri, I think).

2. More or less the same. With the final firmware, drivers and TM FX this part will then be checked again and documented.

Regards
Matthias Carstens
RME

MC's Website

  • Registered: 2010-07-11
  • Posts: 47

Re: A few questions about Latency and Fireface UFX.

Does this mean that whenever you care about sample accurate audio routing from the DAW to outside and back again, you need to do a fresh calibration in your DAW for the specific job? Or should this be a calibrate once for the type of routing that your are using and that should be good forever?

I have found that currently I need to calibrate every day (or every power cycle). A little annoying when you have a project in a DAW with external hardware routings and having to check the calibration every time you load it. (No idea of this is an RME quirk, or OSX quirk or even a Live quirk - I just don't remember having to calibrate often in the past).

We were routing through the UFX loopback yesterday just to get around a problem that a certain DAW was having with sample accurate internal bouncing and noted having to recalibrate a couple of times around DAW reloads and UFX power cycles. (Yes the bounce through the UFX was good once latency calibrated as we needed an exact copy of what it was doing when playing).

  • From: Sweden
  • Registered: 2008-03-14
  • Posts: 149

Re: A few questions about Latency and Fireface UFX.

MC wrote:

That would be part 2 - can't remember here crying alone in the ship :-)

1. The latency differs hardware-wise in certain modes, due to the complicated routing and connections between FPGA and DSP. Whenever we find the time we will document this in detail. AFAIR the delay is 3 up to 7 samples, but don't quote me. Loopback is one of the mode that has its special latency behaviour. Someone else already asked about that and measured some (forum member Zavieri, I think).

2. More or less the same. With the final firmware, drivers and TM FX this part will then be checked again and documented.

Thanks for your answer Matthias!
Looking forward to the new documentation.

RME Fireface UFX
Mac Mini 2.6 GHz 16 GB Ram Areca RAID | MacBook Pro 2,2 GHz 16 GB Ram | OS 10.10.5

6 Reply by MC

  • MC
  • MC
  • Administrator
  • Offline
  • From: Germany/Thailand
  • Registered: 2007-04-06
  • Posts: 24,877

Re: A few questions about Latency and Fireface UFX.

The latency within the UFX (and its drivers) is fixed per mode. So a simple power-cycling will not change it. That must have a different reason then.

Regards
Matthias Carstens
RME

MC's Website