This is why I hate the Fresco Logic early firmware controllers.

If one device on the hub needs to be woken up, IT KILLS THE ENTIRE HUB. Okay, the 4 port powered hub works when nothing needs a power cycle from the root controller for certain flaky things, but still manages to crap out the ENTIRE HUB when another 2nd power cycle is issued for a port/device that went to sleep. Looking into getting this one from Orico: What I have not tested yet are powered hubs. You just have to be EXTREMELY PICKY on controller cards for VFIO. I’m sorry, but because the third party XHCI controller market is the minority now, this won’t be fixed.

I’ve not heard of any complaints from people using native Intel or native AMD USB 3.0, so I guess add-in card people are SOL. 5V is directly fed externally so the sleep/wake only has to worry about the data links, not the power management.Ĭhipset based (B450/X470/Z370) ones likely work far better than external ones and this crapshoot of power management. If there’s no supplemental power going to the plugs, the broken port sleep/wake functions break even further because it thinks 500mA is the limit and it pulls 600mA and it just breaks.Ĭontroller cards with supplemental power work great. High current devices just damn don’t work. Not unless they are extremely low current or a specific class of device that plays well, such as a card reader. Here’s the beginning of my headache regarding XHCI power management in xhci_hcd using third party USB 3.0 controllersĬontrollers that rely on standard bus power don’t sleep properly at all.

Via xhci driver windows 10 vl805