My previous tests involved using the VIA controller for USB 3.0 and the motherboard chipset (Intel P35/ICH9R) for USB 2.0. Additionally the USB 3.0 tests were done via the hub built in to my monitors as it's conveniently placed. So let's mix things up a bit...
USB 2.0 → USB 3.0 hub → USB 3.0 port
I've got a USB 2.0-only cable (actually a power meter) which I can stick between the USB 3.0-SATA adapter and any USB 3.0 ports, forcing everything to drop down to USB 2.0 speeds.
There's a surprise. Running USB 2.0 into a USB 3.0 hub/port runs significantly slower than using a USB 2.0 port.
USB 3.0 port
What if I take the hub out of the equation and plug directly into the USB 3.0 card?
Well it's nice and fast again, but at the cost of a fully loaded CPU core.
USB 2.0 → USB 3.0 port
Again the USB 2.0 performance is nearly a third slower than it should be, though it's a smidgen better than it was with the hub in the way. Ironically this means in trying to make the USB 3.0 ports on the monitor hubs go faster I've actually slowed them down for USB 2.0 - which, given that most widgets I own are USB 2.0 is not an improvement.
|Cabling||Controller||Max (MB/s)||Average (MB/s)|
|mSATA → IDE → USB 2.0||Intel||33.3||32.8|
|mSATA → IDE → USB 3.0||VIA||81.7||79.1||(through hub)|
|mSATA → IDE||Intel||82.1||81.1||(as installed in Aether)|
|mSATA → SATA → USB 2.0||Intel||33.3||32.9|
|mSATA → SATA → USB 2.0||VIA||26.2||25.1|
|mSATA → SATA → USB 3.0||VIA||138.4||129.5|
|mSATA → SATA → USB 2.0||VIA||25.4||24.7||(through hub)|
|mSATA → SATA → USB 3.0||VIA||135.4||129.4||(through hub)|
|mSATA → SATA → eSATA||JMicron||64.2||62.9|
|Plextor 256GB M5S SATA||Intel||220.5||217.3||(as installed in Nyx, limited to SATA-II)|
I do have newer drivers to play with for both the USB and the eSATA controllers so those are next on the list to try.