RME Digiface USB - what latency did you manage to run?

[Yearofthegoat]

I have one also. It's great.

I run the phones/line out to active monitors and S/PDIF out to a headphone amp if I want to use phones.

Haven't tested RTL but those figures already mentioned seem usual. Reaper reports "~0.7/1.7ms" 32smp 48K. Likewise in FL Studio. Cubase says 0.73/1.77. No surprises as it's what is being reported. But RME units are known to report accurately.

Essentially I don't have to fiddle around to get usable MIDI keyboard input, I can leave it on 256smp and I can't really hear or feel any lag - which is why I bought the unit. The benefit is that at 256 I don't typically run out of CPU after adding VSTs.

Only exception is if I'm using my MIDI drum pad, in which case I usually take the buffer lower. Seems I notice delay more with drums for some reason. Even then I never need 32. 128 or 64 is just fine.

[lfm]

Essentially I don't have to fiddle around to get usable MIDI keyboard input, I can leave it on 256smp and I can't really hear or feel any lag - which is why I bought the unit. The benefit is that at 256 I don't typically run out of CPU after adding VSTs.

Only exception is if I'm using my MIDI drum pad, in which case I usually take the buffer lower. Seems I notice delay more with drums for some reason. Even then I never need 32. 128 or 64 is just fine.

I've been running mine now almost 2 years at 64 samples and is all good.
- but use a NEC based chip extra usb card with Renesas drivers
- I had a general problem with internal Intel usb + microsoft xhci drivers

Keyboard I find not so sensitive since mostly not that percussive sounds as drums are.
- and plugins like Addictive Drummer 2 add 64 samples latency
- midi is not included in calculations for asio
- but playing realtime drums in my setup on AD2 I am ok

Digiface can only report what it knows, and that is DAC latency for phones out.
- so I did a loopback test running metronome out and checking if recording is on grid
- that discovers that I have to put 42 samples adjustments on all recordings over my ADAT based preamps AD converters
- added to recording settings in daw
- then sample accurate

[Yearofthegoat]

That's interesting re. USB. I have a 5900X on AM4 and for quite a while there were people reporting USB dropouts, stutter etc. I think I was fortunate as the CPU-laned USB ports worked fine. As I approach an elective upgrade, I was thinking about going back to Intel. Maybe need to think more about that.

Re. the RTL etc. I haven't done a loopback test yet. Perhaps I will. I haven't noticed any issues with PDC in DAWs, which suggests that its reporting is - at least resonably - accurate.

Overall it's a great little unit. Wouldn't be without it!

[WatchTheGuitar]

My Macbook keeps warning me that the kernel driver for Digiface I use will need to be replaced with a ‘safer’ driver/system extension at some point. Has anyone tried this other method out?
My latency so low with the interface as it stands so loathe to lose that unless necessary.

[Funky40]

Q:

is there a dfference in latency between RME AIs ? (on mac)
i´m particularly interested in the digiface.
Is it as low in Latency as the "better" (bigger) ones ?

From searching the web, it looks to me like the RMEs use all the same drivers.
i need lowest latency as possible

[lfm]

i need lowest latency as possible

I run 32 samples with Digiface at 48k with no problem too, if so.
- only thing that happends is cpu goes up the lower you go
- estimate, half latency, double cpu
- no droputs or pops or anything
- with 64 samples I get just over 200 samples RTL as I recall
- less than 5 ms delay

Doing eDrums monitoring through there is no problem.
- if my playing go tighter than 30 ms I feel in the pocket
- RTL does matter being that low as 5 ms

I run Sonar on Windows then.

I would look at which daws you use instead if to go lowest!!!!!

Reaper, if monitoring through while recording, add an extra ASIO buffer size, as many as needed, until covering loaded plugins latency.
- 1 sample on a plugin, an extra asio buffer is added
- running Reaper 2010 I had to go from 128 samples to 64 to not get noticable delay.

StudioOne use standard one extra ASIO buffer for processing.

AFAIK Cubase does not, and I am on Windows so Sonar does not either.

[Funky40]

I would look at which daws you use instead if to go lowest!!!!!

Thanks for your detailed report !

I still wonder if all RME => USB AIs <= have the same latency, or are there differences between RME AIs ? since the digiface itself is quite old tech. Anybody here in the knows ?

What you´re saying sounds ok, or good, so far.


DAW ? i know what you mean I use no DAW, i use GigPerformer (host). This is best vs. Latency.
So, on the hosting-end am i perfectly fine. GP is king

[vitocorleone123]

I would look at which daws you use instead if to go lowest!!!!!

Thanks for your detailed report !

I still wonder if all RME => USB AIs <= have the same latency, or are there differences between RME AIs ? since the digiface itself is quite old tech. Anybody here in the knows ?

What you´re saying sounds ok, or good, so far.


DAW ? i know what you mean I use no DAW, i use GigPerformer (host). This is best vs. Latency.
So, on the hosting-end am i perfectly fine. GP is king

RME writes their own drivers (many companies rely on a white label solution or base it on one, apparently). It doesn’t matter how old tech is but how effectively it’s utilized - and the Digiface is exceptionally well utilized.

[Funky40]

RME writes their own drivers ....

I´m aware of that

- and the Digiface is exceptionally well utilized.

i´m in no doubt of that

But my point is, is the digiface as little in latency as some bigger RMEs. particularly the TB ones ?
Or at least same fast as *any* other USB-based RME, same as the big -expensive- ones ?


I need to get down a decission how to walk into the future v.s AIs.
I could afford a Digiface right now. But not something bigger from RME - for now.

[vitocorleone123]

RME writes their own drivers ....

I´m aware of that

- and the Digiface is exceptionally well utilized.

i´m in no doubt of that

But my point is, is the digiface as little in latency as some bigger RMEs. particularly the TB ones ?
Or at least same fast as *any* other USB-based RME, same as the big -expensive- ones ?


I need to get down a decission how to walk into the future v.s AIs.
I could afford a Digiface right now. But not something bigger from RME - for now.

There’s a latency database on GS.
Digiface USB is $385+shipping right now.

[WatchTheGuitar]

If anyone’s thinking about getting the Digiface USB may be worth mentioning that the real time TotalMix FX monitoring effects aren’t available on the Digiface, only on the more expensive interfaces that come with DSPs. It didn’t bother me as I hadn’t planned using them but it isn’t that clear it isn’t available because a lot of RME videos show features from the whole range.

[PAK]

My Macbook keeps warning me that the kernel driver for Digiface I use will need to be replaced with a ‘safer’ driver/system extension at some point. Has anyone tried this other method out? My latency so low with the interface as it stands so loathe to lose that unless necessary.

I’ve been using the Driver Kit drivers since just after Sonoma launched. RME themselves have said there is no performance loss since Sonoma (this wasn’t the case prior to Sonoma).

[PAK]

But my point is, is the digiface as little in latency as some bigger RMEs. particularly the TB ones ?
Or at least same fast as *any* other USB-based RME, same as the big -expensive- ones ?

I could afford a Digiface right now. But not something bigger from RME - for now.

USB devices tend not to report their true latency (this is also the case for RME). So a physical loopback cable is required to measure true RTL.

As mentioned, DawBench have tested various cards over the years, including the true RTL. But some were done on the V3 drivers, and some on V4, which it makes it difficult to know how much any latency differences are down to using the newer drivers, versus the hardware implementations actually making a difference.

What the numbers do say, however, is RME's USB devices typically have between a 0.5ms and 1ms disadvantage, VS PCIe / Thunderbolt. Which is probably less than many would expect?

BTW: For those on Macs.. Remember you can still use PCIe on all modern Macs via an external Thunderbolt enclosure..

[Funky40]

If anyone’s thinking about getting the Digiface USB may be worth mentioning that the real time TotalMix FX monitoring effects aren’t available on the Digiface, only on the more expensive interfaces that come with DSPs. It didn’t bother me as I hadn’t planned using them but it isn’t that clear it isn’t available because a lot of RME videos show features from the whole range.

Thanks for that.
i studied their page and videos and effectivly thought i would have access to total Mix, buying the digiface. Not that important for me either. But it was an optional point. Since i do PC-intern audio routing stuff. Can´t tell from a distance if i would switch over to use total mix, if i had it available.

There’s a latency database on GS.

I was there at GS the other day. guess that was a thread or so. The therad was to loong to cluttered etc. to be of real help. Where do i have to look ?

[Funky40]

What the numbers do say, however, is RME's USB devices typically have between a 0.5ms and 1ms disadvantage, VS PCIe / Thunderbolt. Which is probably less than many would expect?

Thats very valuable info ! Much thanks !
Yes, less than expected. ( i use right now a Quantum2626 with TB3)

The RME would replace my 2nd AI, respectivly might go to a 2nd Mac later.
I´m hesitant to just get me another Q2626, since they are now EOL.
and the ADAT based digiface would allow for quite some flexibility in my setup, by adding some Adat converter boxes to the whole setup.