NVS 290, Windows 10, and DisplayPort

[Assimilator]

Since Ryzen CPUs have no iGPU, that means if your primary GPU dies you have no way to get into Windows to do stuff while you wait for a replacement. To prevdent such a terrifying scenario, I purchased a Quadro NVS 290 off eBay and a DMS-59-to-2x-DisplayPort cable to go with it (since my monitors are connected to my primary RTX 2060 SUPER via DP). Problem is, it doesn't seem to work:

* Try to run the NVS 290 alongside 2060 SUPER in Windows 10 = NVS 290 gets disabled by Windows because it "reported errors" (Windows did download and install the latest supported Quadro drivers, 342.01, for Win10)
* Try to run the NVS 290 alone in the primary PCIe x16 slot = no output via either of the DP outputs on the cable, motherboard post code however indicates the system has booted into Windows successfully

Is it possible that the NVS 290 somehow doesn't support the DMS-59 to DP adapter? I was under the impression that the adapters are device-agnostic, but being wrong in this regard would explain why things aren't working.

 

[StefanM]

You cannot run 2 different NVIDIA drivers at the same time; Tesla architecture drivers don't support Turing and vice versa.
You need at least a Kepler GPU as sidekick.

 

[Assimilator]

Purchased a DMS-59-to-DVI cable and it works perfectly with the NVS 290. Will be returning the DMS-59-to-2x-DisplayPort cable for a refund. I'm not sure if the cable is faulty, or if the card itself needs to be able to "understand" DisplayPort to output it, but either way it doesn't seem to work as I expected.

 

[lexluthermiester]

You cannot run 2 different NVIDIA drivers at the same time; Tesla architecture drivers don't support Turing and vice versa.
You need at least a Kepler GPU as sidekick.

Um, yes you can. Quadro's and Geforces run just fine side by side for example. I just threw in an old Quadro FX1800 next to my 2080 and it works perfectly. This thread made me curious and sure enough it works.. Granted I don't have a Tesla card to test, but still.

 

[Assimilator]

Um, yes you can. Quadro's and Geforces run just fine side by side for example. I just threw in an old Quadro FX1800 next to my 2080 and it works perfectly. This thread made me curious and sure enough it works.. Granted I don't have a Tesla card to test, but still.

Well, the NVS 290 and the 2060 SUPER definitely don't like each other.

 

[Toothless]

Um, yes you can. Quadro's and Geforces run just fine side by side for example. I just threw in an old Quadro FX1800 next to my 2080 and it works perfectly. This thread made me curious and sure enough it works.. Granted I don't have a Tesla card to test, but still.

The only time cards won't run together is a legacy GeForce and a newer GeForce. Drivers straight up said "this card is too old to run with this newer card."

 

[Assimilator]

The only time cards won't run together is a legacy GeForce and a newer GeForce. Drivers straight up said "this card is too old to run with this newer card."

I didn't actually check in the NV control panel with both cards installed. Will look at that later this week.

 

[StefanM]

Um, yes you can. Quadro's and Geforces run just fine side by side for example. I just threw in an old Quadro FX1800 next to my 2080 and it works perfectly. This thread made me curious and sure enough it works.. Granted I don't have a Tesla card to test, but still.

Interesting, the FX1800 is also Tesla architecture.
What happens when you read out the CUDA capabilities from both GPUs in the advanced tab of GPU-Z?

 

[lexluthermiester]

Well, the NVS 290 and the 2060 SUPER definitely don't like each other.

The only time cards won't run together is a legacy GeForce and a newer GeForce. Drivers straight up said "this card is too old to run with this newer card."

I'll admit, I only installed the drivers to see if they would take and they did, I mistakenly assumed they would work. However, I didn't check the NV control panel earlier to play with any settings. You both are right. As soon as I opened the NVCP it crashed and I mean hardcore crashed, as in I needed to take out the Quadro, remove the drivers for BOTH cards and reinstall the RTX drivers(needed to update them anyway).

So I retract what I said above. Sorry folks.

Interestingly enough, I threw the Quadro into one of my other systems with a GTX670, and not only did it take, but everything is well and good. Could the RTX line be enough of an architectural departure from past designs that the NVidia unified driver system is not longer completely unified?

Interesting, the FX1800 is also Tesla architecture.

Right?

What happens when you read out the CUDA capabilities from both GPUs in the advanced tab of GPU-Z?

Didn't get that far, and I don't want to go through the hassle again.