Actual RTL numbers for Axe-Fx III?

koitsu

New Member
Has anyone posted actual round trip latency figures for the Axe-Fx III as an audio interface?
I mean actual values measured via manual loopback or through an RTL utility.

Kind of excited about getting either a III or an FM3 and replacing my current recording setup, which is a Mark V head into an RME interface. The catch is that I'm used to excellent round trip figures of <5 ms, even at medium buffer settings @ 48kHz. And yes, I do realize that the specs for the FM3, such as converters, have not been released yet (but hopefully, it'll be comparable or equal to the flagship model in that regard).
 
The Axe Fx III, RME Fireface UC (USB2) and Presonus Quantum (Thunderbolt) all on the same Mac mini.
32_Axe.png 32_Fireface.png 32_Presonus Quantum_2018.png
Problem is that even if the Axe Fx III gives a ok RTL - it can't really run on a 32 samples buffer without dropouts. The RME and of course the Quantum does that without any hiccups.

As I understand it the "driver" part of the Axe III is a generic driver from the supplier of the USB port on the Axe and thus not optimized like on the RME interface.
 
The Axe Fx III, RME Fireface UC (USB2) and Presonus Quantum (Thunderbolt) all on the same Mac mini.
View attachment 55025 View attachment 55026 View attachment 55027
Problem is that even if the Axe Fx III gives a ok RTL - it can't really run on a 32 samples buffer without dropouts. The RME and of course the Quantum does that without any hiccups.

As I understand it the "driver" part of the Axe III is a generic driver from the supplier of the USB port on the Axe and thus not optimized like on the RME interface.
That's an Axe II in the screenshot. IS the III the amp or better?
 
Thanks for the comparison. Granted that is DAW reported so we're trusting what is being reported to the DAW is completely accurate. But still nice to see those three relative to each other in the same DAW/computer.

There was another thread where the Fractal mod mentioned that there is a new USB driver for the Axe III, so I was hoping there might be improved stability and speed with the new hardware. Especially since they made the effort to beef up the audio interface hardware.
 
Thanks for the comparison. Granted that is DAW reported so we're trusting what is being reported to the DAW is completely accurate. But still nice to see those three relative to each other in the same DAW/computer.

There was another thread where the Fractal mod mentioned that there is a new USB driver for the Axe III, so I was hoping there might be improved stability and speed with the new hardware. Especially since they made the effort to beef up the audio interface hardware.

Yes is the reported round trip latency from Studio One Pro - maybe not accurate - but good enough for comparison.

As you can see - the Axe III gives the same numbers as the II.
 
Wow, you're right--identical! So maybe it is indeed a generic USB driver as you say. Thanks.
Wouldn't surprise me. I remember when the latency improved so much between the Focusrite Scarlett 1st and 2nd gen interfaces, then some clever people realized you could get the latency improvement just by installing the 2nd gen drivers with the 1st gen hardware. The improvement was software optimization, obviously.

Drivers can be some persnickety things.

Anyone have latency measurements for Windows? I'm assuming that the Mac drivers are the class compliant default drivers that ship with OS X, as there's no need to install special drivers for Macs.
 
Last edited:
Has anyone posted actual round trip latency figures for the Axe-Fx III as an audio interface?
I mean actual values measured via manual loopback or through an RTL utility.

Kind of excited about getting either a III or an FM3 and replacing my current recording setup, which is a Mark V head into an RME interface. The catch is that I'm used to excellent round trip figures of <5 ms, even at medium buffer settings @ 48kHz. And yes, I do realize that the specs for the FM3, such as converters, have not been released yet (but hopefully, it'll be comparable or equal to the flagship model in that regard).

Define medium buffers :)

I've measured the whole loopback latency of the Axe-Fx III using Linux;
  • I have a cable going from Output 3 L to Input 3 L
  • USB Buffer Size in the Axe-Fx III itself is set to 8 (This contributes 8x4 = 32 samples of loopback latency from the axe-fx Itself)
I setup a patch like this:
Screenshot from 2019-05-16 13-37-09.png

The total loopback latency with that configuration is around 390 samples = 8.125ms

The latency measurement includes,
  • The USB Buffer size in the Axe-Fx III itself (set to 8 here, 2 buffers in, 2 buffers out, 32 samples total)
  • Jack Buffer sizes (You can think of Jack as the ASIO equivalent in linux) set to 64 (Total sizes for complete loopback is 64x3 + some USB guard which I don't know how much that is).
  • Measures DAC and ADC latency since I'm going out and in gain.
  • Measures any grid latency (IN usb-> to Out 1)
  • Includes latencies incurred by USB itself.
Note that the "software reported" latency will be much less.

Decreasing JACK buffer sizes to 32, decreases total loopback latency to 276 samples = 5.75ms but I get occasional glitches in my system with that.
 
Last edited:
OK, booted windows.

Same configuration as above (same patch, loopback, same USB Buffer Size)

ASIO Config, Safe buffer disabled
8 samples -> RTL 258/5.375ms
16 samples-> RTL 320/6.67ms
32 samples->RTL 332/6.917ms
64 samples->RTL 358/7.458ms
64 samples, safe buffer enabled - RTL 402/8.375ms
 
Define medium buffers :)

Heh, heh, I was thinking 128-256 as "medium" and anything below that as "low."

I've measured the whole loopback latency of the Axe-Fx III using Linux...

This is the same type of method I've been using lately. I've used the Oblique Audio tool before but just to be absolutely sure, I connect an instrument input and speaker output on my interface and record a synth beep or metronome click, move that to a new track and re-record via loopback to a fresh track. At 64 samples, I get about 4.4 seconds (212 samples), but there's not much else going on other than the AD/DA with just that. If I start throwing plugins at it, then that'll naturally increase. And this is all with the optimized RME drivers, of course.

Thanks for testing it in Win 10! I guess those numbers are similar to a lot of current USB interfaces, right? Kind of reinforcing what Sustainerplayer was suggesting earlier. Do you notice much difference in feel playing into your DAW like that versus direct from the Axe's analog outputs?
 

Attachments

  • latency.png
    latency.png
    98.1 KB · Views: 8
*Correction to my earlier statement: I just measured again and actually got 7.08ms at 128, not 5ms. Sorry, overestimated.
 
Heh, heh, I was thinking 128-256 as "medium" and anything below that as "low."

This is the same type of method I've been using lately. I've used the Oblique Audio tool before but just to be absolutely sure, I connect an instrument input and speaker output on my interface and record a synth beep or metronome click, move that to a new track and re-record via loopback to a fresh track. At 64 samples, I get about 4.4 seconds (212 samples), but there's not much else going on other than the AD/DA with just that. If I start throwing plugins at it, then that'll naturally increase. And this is all with the optimized RME drivers, of course.

Thanks for testing it in Win 10! I guess those numbers are similar to a lot of current USB interfaces, right? Kind of reinforcing what Sustainerplayer was suggesting earlier. Do you notice much difference in feel playing into your DAW like that versus direct from the Axe's analog outputs?

Yes that's actually what I did with Reaper, Insert->Click source. In Audio->Recording I uncheck "Use audio driver reported latency" and put 0 in the offsets. Then I output the click source at USB Output 7 (IN_USB L) and record output 1 (OUT1 L).

RTL utility was in the ballpark.. and jack_delay in linux was also in the ballpark of what I measured with Reaper.
 
Do you notice much difference in feel playing into your DAW like that versus direct from the Axe's analog outputs?

I never do that with guitar (I don't see a need). But I do play PianoTeq over the interface, which needs 64 samples minimum - so assuming it's half of the loop-back latency - should be around 4ms - it feels great.
 
Given siund's rate of travel in air, 1mS = 1.13' at sea level.

8ms is 9.04'.

So, at 8mS, the latency is the same as if the speaker was 9.04' away.

Pretty typical distance for many stages.
 
Thanks again for checking it, Albert. That sets my mind at ease.

I know some people are very sensitive with latency, but I personally don't notice it below 10ms. Around 12-15ms, I can start to feel a little detachment but can still manage fine. I think when I switch to (probably) the FM3, I'll end up doing everything but maybe reverbs on it and then direct monitoring while recording.
 
I just noticed that FAS released an Axe III USB firmware update on April 12 and "Axe-Fx III USB Driver Setup for Windows" at the end of January this year. AlbertA (or anyone else), are you running those latest versions?

The driver package is for Windows only. They note on the download page that no driver is needed for macOS, so I guess it's using Core Audio as expected.
 
Back
Top Bottom