RME Multiface Distorted with GP3


I’ve just logged a ticket for an issue with my RME Multiface / GP3 on my main DAW - It won’t work in ASIO mode without digital distortion. All other DAW apps. work great on this PC.

To further investigate this I tried installing a BabyFace Pro - This works at buffer setting 1024, but below that does exactly the same as the Hammerfall. This is not the case on my laptop.

Update - After a reboot the BabyFace Pro is now working great down to 128 samples, below that gives the digital distortion.

These results repeat in GP2 and GP3.

The DAW PC having the issues with GP3 is pretty much state of the art - A new 3xs i9 9900k based DAW.


256 samples is by far enough and it will give you more security with regard to glitches, than 128. Don’t try to go below it doesn’t make sense.

Yes, but the Multiface is not exactly state of the art
Which driver do you use for the Multiface ?


The comment about achievable latency was nothing to do with best practice at gigging, more about the capability of the device. (for live I’ll run as high a buffer that I don’t notice. I can live with 512).

Nothing wrong with a Multiface even in 2019. Driver version is 4.26 from 07/02/2018.

It may be old but it’s rock solid especially at lower latencies for tracking - it beats the (currently produced) BabyFace Pro in that respect. It easily outperforms the vast majority of current USB interfaces with their generic USB chipsets. Regardless of age it’s simply one of the best interfaces in its class. This is why I still use it.

The point is that everything else on the PC works great with it, as one would expect. So the issue is unique to GigPerformer.

I just hope that this issue is a symptom of something that may have an impact elsewhere so is worth fixing, and not just unique to a 15 plus years old soundcard, in which case it may be perceived to be not worth the effort to fix, sadly.

After more testing this issue is linked to GP running on my main DAW only. The Multiface uses a Cardbus adapter running in a Magma Chassis (PCI to PCI bridge, so no drivers in the way here).
When I tried the Cardbus in the laptop using an ExpressCard to Cardbus adapter, the Multiface ran flawlessly down to a buffer of 256 samples.


Just out of interest - Why would you reply with this generic comment without considering the context.
In fact your comment of “doesn’t make sense” is frankly invalid. I have a real issue here that only impacts GigPerformer on a proven stable DAW system.

Who are you anyway? do you work for Deskew? If not please stay out of this thread going forward, thanks!


When a conversation starts like that, it never turns out well. That’s why I already stayed out of this thread after your first answer. So, no need to warn me just to ask me to keep out of the topic, I did so by myself. Also don’t expect me to get into a sterile verbal fight now, I have too much real problems in my live for that. I think you already decided « who am I », and your question is not one. I never meant to be haughty with you, but whatever, because you already decided that I was. I won’t try to explain why I said this or asked that, this doesn’t make sense anymore now given the way the discussion is turning out.

If you already opened a ticket for your issue, you are already in good hands with the Deskew team. So the only additional benefit when posting in the community forum is to have some kind of contacts with other community members. But, even if you warn me again, you can be assured, it won’t be with me in the future. :zipper_mouth_face:


I wanted to keep this on track because so many threads on forums get derailed quickly, and it’s frustrating having this issue.

Apologies for any offence caused - I certainly did not mean to do so, but on re-reading I can appreciate how my comment could have been taken that way. Sorry.

I’ll stay off here and let tech support check on this.


Thank you for that but please don’t stay off. There’s a far better chance of solving this with lots of eyeballs and experienced users, to say nothing of the benefit of future users who run into the same problem.
Can you summarize exactly the current status? It sounds like everything works down to 128 samples. Is that correct?