• Registered: 2014-07-09
  • Posts: 6

Topic: Latency of Fireface UFX (FF-UFX)

Hello,

I am trying to measure the latency of Fireface UFX (FF-UFX).  I want to know if  FF-UFX  has also an extra latency like in Multiface II  (here http://www.rme-audio.de/forum/viewtopic.php?id=12952   is reported that the  MF-II produces an extra latency of 15 samples. With the same setup and procedures I am describing below, I also made a measurement with MF II, I can confirm that it occurs an extra latency of 15 samples in MFII…)

The measurement is done with Window 7 PC via USB. The procedures I did are as follow:
- Generate  the white noise from MATLAB.
- Route that signal to one output of FF UFX.
- That output is connected to one input of FF UFX with an external cable
- Record the output by compensating ASIO-delay. The ASIO-delay is 1119 samples for a Buffer-Size of 512 samples (sample rate is 48 kHz).
- Compute the latency by performing cross correlation between the generated signal and the recorded signal

The result is that the latency is 0 if DSP EQ+D is ON. However, the latency is -4 if DSP EQ+D is OFF.
Here, I do not understand why setting DSP EQ+D to OFF causes a delay of 4 samples? And why the recorded signal comes earlier, which does not make any sense to me ?

The above result is the outcome of the measurement by a Desktop PC. However, when I tried with an another Desktop-PC and also with a laptop, which have the same OS, the latency is 0 regardless of the DSP EQ+D setting.

So, my questions are:
1)Why the setting DSP EQ+D (OFF, unchecked)  causes sometimes  (not in all PCs) a delay of about 4 samples?

I know what does the setting DSP EQ+D mean. If you activate this option, the EQ on the signal on the input channels will be recorded with the EQ. If there is no EQ - only the unchanged signal will be recorded. But why this setting (if it is OFF) causes a delay between the generated and the recorded signal?

2) Why is this latency negative?  Does it have a special meaning?

3) Does that latency depend on the measurement PC as well???

With the same setup and procedures I described above, I also performed measurements with a Babyface. In this case the measured latency was +1 sample regardless of the DSP EQ+D setting and also in all PCs, I measured with.

4) Why the DSP EQ+D settings does not make any latency difference in the case of Babyface but it does with FF-UFX?

5) How does this ASIO-delay of 1119 samples come about? As I understood from manual:

2*Buffersize                              =1024 samples
AD/DA                                =40 samples
Digital-Receiver --> TotalMix --> Transmitter     =3 samples
Safety-Buffer                             =32 samples
1119-(1024+40+3+32)=10 samples

Where do these 10 samples come from?  What I am missing?  Is there any USB-Bus-delay , which is reported  also here http://www.rme-audio.de/forum/viewtopic.php?id=18625?

Best,
Lin

2 Reply by MC

  • MC
  • MC
  • Administrator
  • Offline
  • From: Germany/Thailand
  • Registered: 2007-04-06
  • Posts: 25,599

Re: Latency of Fireface UFX (FF-UFX)

1. Which driver version did you use on the desktop computer with the UFX? Did you read out the driver reported ASIO latency with and without DSP for record? It seems not properly, as the reported latency will change by the mentioned 4 samples, so the files would match again.

5. Yes, USB.

Regards
Matthias Carstens
RME

MC's Website

  • Registered: 2014-07-09
  • Posts: 6

Re: Latency of Fireface UFX (FF-UFX)

Thx a lot for the answer ....
Now I can see that the latency is not dependent on the PC itself but on the driver-version of the RME …   
My question (3) in my first post is hence answered.

I checked the driver-versions again. Here are the resaults of the latency DelayIO (between the generated signal and the recorded signal)  and the  reported ASIOdelay for different soundcards and driver versions:

                                                                              Driver                       Driver
                    DSP-EQ                                                1.0.36.0                    1.0.53.0
                                                         
Babyface           OFF            DelayIO                              1                          1
                                      ASIOdelay                          1142                        1142

                   ON               DelayIO                              1                           1
                                      ASIOdelay                           1142                      1142



Fireface          OFF                DelayIO                           -4                         0
                                        ASIOdelay                         1119                      1115

                    ON                  DelayIO                           0                         0
                                         ASIOdelay                         1119                   1119

For Babyface:

2*Buffersize                                                           =1024 samples
AD/DA                                                                      =43+28=71samples
Digital-Receiver --> TotalMix --> Transmitter             =3 samples
Safety-Buffer                                                           =32 samples
ASIOdelay                                                                  =1142

1142-(1024+71+3+32)=12 samples  ?Regardless of the DSP EQ+D setting and also regardless of the driver version.

•    Can u confirm me that the USB-Bus-delay is 12 samples in Babyface?

For FF-UFX:

2*Buffersize                                                           =1024 samples
AD/DA                                                                      =12+28=40 samples
Digital-Receiver --> TotalMix --> Transmitter     =3 samples
Safety-Buffer                                                           =32 samples

For the older driver version ( 1.0.36.0)  

ASIOdelay does not change by changing the EQ+D settings.
But the measured DelayIO changes by changing the EQ+D settings.

Hereby I still have the same questions:

•    Why the setting DSP EQ+D (OFF, unchecked)  causes  a delay of about 4 samples?
•    Why is this latency negative?  Does it have a special meaning?

For the newer  driver version ( 1.0.53.0)   

ASIOdelay changes by changing the EQ+D settings. In contrary the measured delay does not change and is 0.

For EQ+D is ON

1119-(1024+40+3+32)=20 samples (sorry for tipp mistake in my first post I wrote “10” … )
•    In this case is USB-Bus delay is 20 ? Can u confirm this?


For EQ+D is OFF

1115-(1024+40+3+32)=16 samples

•    Why the USB-Bus-delay changes now about 4 samples?


- Best

4 Reply by MC

  • MC
  • MC
  • Administrator
  • Offline
  • From: Germany/Thailand
  • Registered: 2007-04-06
  • Posts: 25,599

Re: Latency of Fireface UFX (FF-UFX)

pyei.phyo.lin wrote:

•    Can u confirm me that the USB-Bus-delay is 12 samples in Babyface?

Yes.

pyei.phyo.lin wrote:

For the older driver version ( 1.0.36.0)  

ASIOdelay does not change by changing the EQ+D settings.
But the measured DelayIO changes by changing the EQ+D settings.

Of course.

pyei.phyo.lin wrote:

Hereby I still have the same questions:

•    Why the setting DSP EQ+D (OFF, unchecked)  causes  a delay of about 4 samples?l

Internal design of the hardware, it works like that. The data is tranferred between FPGA and DSP several times. That also explains why the values don't change with the Babyface - it does not have an 'external' DSP chip. Otherwise: In the old driver we simply forgot to give matching ASIO Latency values for the devices with external DSP and this condition.

pyei.phyo.lin wrote:

•    Why is this latency negative?  Does it have a special meaning?

No.

pyei.phyo.lin wrote:

For the newer  driver version ( 1.0.53.0)   

ASIOdelay changes by changing the EQ+D settings. In contrary the measured delay does not change and is 0.

I think first you confused yourself and then me smile

The hardware delay changes, so the driver reports matching ASIO delay values, so after all the error is zero in both cases. Perfect, isn't it?

Regards
Matthias Carstens
RME

MC's Website

  • Registered: 2014-07-09
  • Posts: 6

Re: Latency of Fireface UFX (FF-UFX)

Thx for answers ... very helpful ...  I really want to understand where the latencies are coming ….

I was just comparing the older and the newer driver versions and saw that for older one ( 1.0.36.0)  changing the setting DSP-EQ (ON or OFF) does not make any change on the ASIOdelay  but it does for newer version. But now I understand why.  So simply there is no matched ASIO Latency values for the devices with external DSP and this condition. Now it is clear for me.

Back to the samples-matches in FF-UFX  for the newer driver version.

2*Buffersize                                                           =1024 samples
AD/DA                                                                      =12+28=40 samples
Digital-Receiver --> TotalMix --> Transmitter     =3 samples
Safety-Buffer                                                           =32 samples 

For DSP EQ+D is OFF

1115-(1024+40+3+32)=16
USBbusDelay=16

•    So it means the USBbusDelay in FF UFX is 16 samples (in Babyface 12)? Can u also confirm this?

•    The  condition/setting DSP-EQ produces a latency value of  DSPEQ=4 samples ?

Proof:

For DSP EQ+D is ON

1119-(1024+40+3+32)=20
20=USBdelay+ DSPEQ=16+4

-Best and thx again for ur answers

  • Registered: 2014-07-09
  • Posts: 6

Re: Latency of Fireface UFX (FF-UFX)

Hello Matthias,

Just the final confirmation from you.

Is the USBbusDelay in FF UFX really 16 samples?

Thank you very much and looking forward to hearing from you soon. smile

Best
Lin