mike_mccue
The fact that a DAW engine can work smoothly with lots of tracks, efx and vsti with a couple hardware inputs and outputs doesn't really say much about the capability of the I/O driver or hardware.

There aren't many usb devices that are candidates for comparison if you expect 24 or more I/O as a base line for comparison.

Hey Mike,
 
All RME USB audio interfaces offer the same level of low-latency performance.

• Babyface has 10 channels of I/O
• Fireface UFX has 30 channels of I/O

 
Some audio interfaces allow you to reduce the number of I/O ports (Lynx AES-16e-50 and the X32 to name two).  
In my tests, reducing the number of I/O ports had no affect on low latency performance.
If the audio interface glitched running 32-channels of I/O, reducing it to 8-channels didn't eliminate glitches.
If you were talking massive amount of I/O, it would likely make a difference, but most folks here aren't running more than 32-channels of I/O.
 

I don't know about 24 or more, I've never used that many.  I've used 16 simultaneous inputs 8 on board 8 through ADAT, on my VSL without a glitch and that's only a USB 2.0 device, and that's using monitor sends using the VSL FX which run through the computer.
 
If someone other than RME would start making USB 3.0 devices (which USB 3.0 now supports 10G) there's more than enough bandwith to support 24 simultaneously A/D conversions.
 
There's even more than enough bandwith on USB 2.0 but it requires writing drivers to tweak the USB clock front and back buffers which if unattended account for 24ms of round trip latency on their own.
 
The audio world is slow at change and has mostly use TB as a way of extending the live of FW well past it's lifespan when it's a dead protocol for any other industry.  Now we are finally seeing TB interfaces introduced 4 years after TB was introduced.  USB 3.0 has been out longer than that and only RME has bit.

Hi Mike,
 
Running a well configured current generation machine;
If you set the ASIO buffer size to 64-samples, you shouldn't have glitches.
To successfully run below that, you need an audio interface with top-notch drivers (RME, MOTU, Lynx).
 
Prior to Z77 motherboards, all USB3 ports were provided via third-party add-on controllers.
This can cause compatibility issues (similar to using a non-TI chipset Firewire controller)... which could affect performance.
 

Mr Roseberry,
 
Didn't know that you responded to this post until tonight.  FWIW, I contacted Jim outside of SONAR and am in the process of having Jim do a build for me (for now and a bit future proof).  I do think that eventually the OEMs will write drivers to support TB on Windows machines and also decided to stay with SONAR (and run PT on the same machine).
 
My real disappointment is with UA and their APOLLO (rather lack of) support on this platform.  Not only do I think that the UA plugs are very good models of the vintage hardware, I like what they have done on the audio side with the Apollo... that print or monitor with their plugins with better than average A/D converters.
 
So, the bottom line after reading these very informative responses is that there is nothing to gain by switching to MAC for me, sticking with APOLLO (FW over TB) and using UAD2 PCIe cards for additional horsepower.
 
Thanks everyone and a big thank you for Jim R for putting up with all my questions.
 
Jim

200bpm
The reality is that current usb2.0 interfaces (can) have great latency and throughput.
 
TB is a solution for a problem that doesnt exist. 
 
 

Succinctly put 200 pm. Thanks for the info Jim Roseberry - great reading.
 
Having spent 5 troublesome years with 2 firewire digital desks and all if the headaches and controller compatability issues that accompanied them, I for one am thrilled to be running an RME fireface over usb. It just works....... so if it ain't broke...:)