Hopefully a simple question: is there a way to query what PCIe 'version' or 'mode' a device is actually using? I know lspci -vv can query what the PCIe root supports (e.g. to tell you your motherboard is 1.0 or 2.0 or what have you), but I want to query 'the other way' and see what a device (like an add-in card) is actually negotiating or using. I know the nVidia proprietary driver's 'X Server Settings' application can get this information for supported nVidia GPUs (on the 'GPU' page under 'Graphics Card Information'), but what about for other devices? (and not just GPUs)

If you check your box's specs, they usually big themselves up hurling those specs at you.

There's pcie stuff in /sys/bus/ and /sys/module - have you checked them?

A thing I thought would be more informative is how many lanes the board is using. Each lane doubles the bandwith, kind of.

Yes I know - I'm not interested in the maximum spec of the motherboard (and lspci -vv seems to report that), but what is actually being negotiated by the device itself with the motherboard. For example, most modern GPUs will not run at maximum revision x maximum lanes all the time, as a power saving feature. On Windows this can be monitored directly via software (like GPU-Z), can it also be done in linux? (apart from nVidia proprietary driver with certain GeForce GPUs)

No. Do you have more specifics?

Yes this also - each 'generation' doubles bandwidth (kind of) as well, so ideally I'd like to see both, and not just for certain nVidia cards. (In other words, what are *other* controllers using? like USB 3.0, non-nVidia GPUs, PCIe port multipliers/bridges, PCIe storage devices, etc) This is more of a 'trust but verify' kind of operation - I can find whatever the device is advertised to do independent of the machine, but what is the device actually doing?