Tried USB 3.0 and with 192k the 8192 appears.
However at 44.1K using USB 3.0 the maximum asio panel is 2096 buffer size. But with thunderbolt at 44.1k its 4096.
So there is a slight advantage to using thunderbolt and that is you can have a higher buffer on a lower project sample rate.
Is there any chance in the future that 8192 could be added to thunderbolt at 44.1K ?
You are mixing causalities and priorities somehow.
The objective is really not to compare or to kind of "rate" USB and Thunderbolt technology by its max buffersize.
Thunderbolt IS external PCIe and MC explained already that it does not need so high buffersizes for this reason.
There is nothing better than PCIe to connect devices to a PC !
Besides this RMEs driver quality and performance is so well that you do not have to worry whether to use the one or other.
At the end it simply boils down for you what works best on your system, whether Thunderbolt or USB3 works better.
Depends on factors like BIOS, mainboard designs, HW/Driver quality of mainboard / thunderbolt cards.
Simply use what works best and fits your demands best.
I personally like USB3 because:
- you have a much broader choice of mainboards
- cable length up to 5m when using good quality USB3 cables from Lindy with 3 shields
- the plugs are IMHO more robust (not so tiny like with Thunderbolt)
Thunderbolt driver has a functional advantage, it still supports the pitch function of the driver if you need it. With the new USB transfer modes, which are required for this high number of channels, the pitch is not supported anymore under USB in the MADIface driver.
Simply use the ASIO buffersize that you need for your project, three examples:
A) If you record something with VSTi (guitar playing via VST amp loaded in the DAW), then you need very low ASIO buffersizes, because the RTL (round time latency) needs to be below 12ms otherwise you will feel uncomfortably and at a certain latency you will immediately slow down.
B) If you record life then simply take the higher ASIO buffersize, so that the CPU load is at minimum. Then the CPU gets lower amount of IRQs and is less unders stress. By this the likeliness to loose Audio is minimized.
C) If you are mixing it depends how many tracks, VST, VSTi you use and what CPU load they cause. Lower RTT can be nice in some situations, but the priority is stability, you should avoid having too high Spikes in ASIO and CPU load.
You can look here that I have no audio drops when playing back a Cubase Project with 400 Tracks and 2 VST per track
no matter whether I run this with 32 ASIO Buffers @44.1 or 64 ASIO buffers @88.2. https://www.tonstudio-forum.de/blog/ind … cks-de-en/
BR Ramses - UFX III, 12Mic, XTC, ADI-2 Pro FS R BE, RayDAT, X10SRi-F, E5-1680v4, Win10Pro22H2, Cub13