The more instruments you load the more your computer will struggle at lower buffer settings no matter what audio interface you use.

If you look into the manuals of the products that you mentioned (BBF Pro FS, UCX II, UFX II/III),
there is always a chapter "Latency and Monitoring" where you find the converter latency for all sample rates.

1. You will see that all products have converters with a very low converter latency between 5–7 samples for A/D and D/A.
This results in a converter latency which is in "the worst case of 44.1 kHz" only around 0.1 ms. This is negligible, even AD and DA together are not more than 0.2 ms.

2. You are talking a lot about RTL, but when playing VSTi one half is missing, A/D and transport of audio data to the computer. The VSTi is either being played by a MIDI track of your DAW or a keyboard connected through either USB or MIDI.
So in most of your use case with VSTi it won't be the full RTL.

3. A RTL of 3ms appears to me as too small to be a strong requirement for all use cases and every day use. When I am playing guitar through a VSTi, then ASIO buffer sizes up to 256 samples and by this RTLs of up to 13,1 ms are no problem to play at all. More comfortably is if you stay below 10ms. I think 3ms is not really required, especially if you do not always have the full RTL when playing with VSTi.

4. With the above-mentioned RME products and modern converters with only 5-7 samples converter latency, it is unnecessary to use double speed, as this saves you only a fraction of a millisecond (around 0.1 ms). But double speed definitively increases the stress on your system because now the double amount of data needs to be processed *in time* for all tracks, VST and VSTi.

5. So with one of the above-mentioned RME interfaces there might be no need to use double speed and this alone might relax the situation.

I know you have Apple, me Windows .. but maybe it helps you a bit to see the RTL at 44.1 kHz for different RME products that I had, owned, tried.
Also special setups with
- RayDAT as recording interface and an UFX in stand-alone mode as preamp
- MADI setup with preamp connected through MADI

What you can see, regardless of which technology you use for transport between Recording Interface and PC.
All RME drivers deliver excellent performance, not far from each other.

https://www.tonstudio-forum.de/blog/ent … cts-en-de/

Thanks for delivering more details, some things I was unable to understand correctly.

The RME design might be different, TotalMix FX has direct access to the internal FPGA (programmable/flashable CPU) of the recording interface. With it, you simply manage the interface in terms of routing and metering.

I do not remember exactly, MC mentioned it a while ago, that inside of the FPGA you have a very low latency of only 1–3 samples.

Otherwise, you only have to deal with the converter latency (which is also very low, 5–7 samples) and the latency needed for the transport between computer / recording interface. This is dominated by the buffer size that you choose in the application (Apple), for Windows this is the ASIO buffer size of the driver.
With Apple, you have additionally a little safe safety buffer for the processing of the internal audio system, as everything goes through it. With Windows / ASIO you fully bypass the computer's audio system, and you have direct access to the HW.

I would say, as your setup is maybe special, you should try it out. I think that with RME you might have a good chance, also, because the transfer over USB/… happens inside the FPGA, no 3rd party chip with potentially bad values or issues are being used inside the recording interface.

Under Windows (ASIO) two different driver exist where the lowest ASIO buffer size differs a little bit, not significantly.
- older USB driver (BBF Pro FS, UCX II): 48 samples @single speed
- MADIface driver (UFX II, UFX III and others): 32 samples @single speed
I am not sure how this is with Apple drivers.

Under Windows only the older USB driver supports pitch. The MADIface driver supports also interfaces with a high number of channels. Maybe the transfer modes that need to be used for USB3 do not allow for pitch. Not sure whether you need pitch anyway. I do not know how this is with Apple, just in case you would like to have pitch.

I think it could boil down more to the question of which features you need, as those interfaces have a quite different number of channels and features.

The Babyface Pro FS is very nice, but I would only recommend it for mobile applications. It does not offer a lot of I/O ports.
If you would still need/want to use double speed, then the one ADAT port appears to be a limitation for me because then you could only use 4 of the 8 ADAT channels.
Another thing is, that it is designed for mobile applications and that it can be operated USB2 bus powered.
This has certain limitations in terms of energy consumption, so the headphone outputs are well, but finally not so strong compared to the other interfaces. Also the effect section is only partially implemented and uses the PCs CPU through the driver.

The other recording interfaces offer a more round package. the UCX II is a big facelift compared to the old UCX.
It has the preamps of the UFX II with 75 db Gain range, nice display and also provides DURec recording which can be used as backup recording in parallel to DAW recording to have a safety belt in case the computer or the DAW has issues.
But also here a little limitation, if you work at double speed you only have an ADAT port where you can only use 4 of 8 channels due to port multiplexing.

To be on the very safe side I would therefore recommend the UFX II or UFX III.
Meanwhile with the newer version of UFX II devices the analog and digital section of UFX II is equal to the UFX III.
The only difference - due to the higher port count of UFX III (by MADI) - is, that the UFX III has USB3 and MADI.

Another thing to consider, UCX II, UFX II/III support the new Room EQ features and Crossfeed, the Babyface Pro FS not.

Maybe also of interest: UFX II and UFX III are using the AKM converter of the ADI-2 Pro FS (former version of ADI-2 Pro FS R BE).

To sum up, my advice would be to get either UFX II (USB2 based) or UFX III (USB3 based, but can also be used with USB3 as 30ch interface).

Also interesting, the UFX III got the 1st implementation of USB3 Class Compliant mode. Interesting for Linux, where no totalmix / driver exist. In such situation CC mode is needed and here for all channels now including MADI.

For further comparing the interfaces you can use my Excel here:
https://forum.rme-audio.de/viewtopic.php?id=35156

At the moment I am not sure, whether the changes are already in that the UFX II has the same analog section as the UFX III. Simply keep it in mind, newer versions of UFX II are same as UFX III except for USB3/MADI.

Sorry, I can only deliver the values of products, that I had access to under Windows.

But the values that I presented in my table (see below), cover already most of the Windows ASIO drivers that RME offers for their products: USB (2), Firewire, TB, PCI/PCIe, HDSPe MADI FX (supporting special features of the card). Only the TB values are from a forum member, I never had TB.

Look at the RTL times of the UFX at 44.1 kHz.
It uses the same USB driver as the Babyface Pro FS with 48 samples minimum ASIO buffer size.
Also keep in mind, that the AD/DA converter have a different latency:
UFX is from 2010, BBF Pro FS from 2019.
[Only for comparison also the values for UCX II, UFX II and UFX III]

UFX:            AD 12 samples = 0.27 ms // DA  28 samples = 0.63 ms
BBF Pro FS: AD   5 samples = 0.11 ms // DA    7 samples = 0.16 ms
UCX II:        AD   5 samples = 0.11 ms // DA 5,8 samples = 0.13 ms
UFX II+III:  AD   5 samples = 0.11 ms // DA     6 samples = 0.136 ms

UFX with 48 samples @44.1: RTL = 4.3 ms which is also a very good value.

BTW .. Careful interpreting the values of RayDAT and HDSPe MADI FX. Those PCIe cards are missing the time for AD/DA, their "final" RTL values depend on the AD/DA converter that you connect through ADAT or MADI. Therefore, the additional columns for these cases with an ADI-8 QS connected through ADAT / MADI.

I can only assume that you will have "compareable" values for Apple, you would have to ask people with Apple for their RTL values, best also at 44.1 kHz to make it a little more compareable to my values.

Quick reply, regarding link .. works for me, browser problem?
Simply use the excel file, its better anyway.

Regarding the issue, sorry, no idea and I am neither using Apple, nor Apollo devices.