You guys probably thought I wasn't serious when I said I was doing this. But in this square lump is exactly what I said I'd do. Beside my nice, shiny GTX 980 there is a Radeon R7 360.

First, some background for anyone who's not aware; There seems to be a bug in rFactor based simulators which causes nVidia cards to throw a 'display driver has stopped responding' error, and it's been there for many years now. It manifests in game as a black screen for a short time, and then it usually resumes like normal a few seconds later... Usually. Sometimes it doesn't recover at all and the only way out is to kill the process with task manager. One thing's for sure, it seems to be track sensitive, as I've never had it at many tracks. But there's some specific tracks where it's happened multiple times. VirtuaLM's Le Mans 1991 for rF2 is probably the most famous example.
rF1 is affected, rF2 is affected, SCE is affected (see this year's Bon Anni), and I expect AMS is affected too (though I've not experienced it yet in that).
I did get some new info last night on this from a friend with a slightly overclocked 970. It may be possible to stop the problem from happening by increasing the voltage on your graphics card slightly, and that can be enough to stabilize it. That does suggest the voltages are slightly too low in these cards when they come from the factory (which does sound unlikely, I know), but it fits in with how overclocking generally works; increasing Ghz makes it unstable at high work loads, and that stability is regained by increasing voltages.
The first post in
this thread supports this theory IMO, as pCars for this guy is perfectly stable at 90% GPU usage. rFactor 2 is also stable at that range, and only becomes unstable at near maximum usage, which is where the lack of voltage would cause instability. pCars only stays stable because it never reaches that point. Though it does mean the games aren't responsible IMO.

I didn't go that route though, I only started looking into this theory last night. Instead, what you see through that window is the two graphics cards. The GTX 980 lights up and extends the longest, and sits nearest the CPU in the 16x PCI-E slot. As the stronger card, I figured it'd benefit from that more, so that's where it went. Especially since I intend to use that for everything non-rFactor related. And that puts the R7 360 in the PCI-E 4x slot.
My cable management and colour scheme rules me out of winning any "sexiest rig" competitions. The "coffee & cream" Noctua fan, red lighting on the motherboard and white light on the 980 don't exactly go together, and my cable management isn't exactly aesthetic either. But it works and it circulates just fine.

The fan on the CPU heatsink pushes air away from the graphics card and out through a vent on the right side of the case (which also exists on the left). But thinking about it more, it should probably be pointed at the vent at the back, maybe with a fan installed there to help exhaust hot air and avoid dust entering through it. Oh well, I'll turn it around the next time it's due a clean.
The R7 360 on the left is a pretty humble card which I picked up the morning of GEARS Spa at my local Maplin for £100. I could easily have gotten it cheaper if I went online, but I wanted it for that evening. I'd been backed into a corner by my own short sightedness, as usual.

It's not blowing anything away in benchmarks, and it doesn't handle GTA5 on max settings with dignity. But it does what I wanted it for just fine; not freezing into a black screen!

It's not the smartest card ever, I think. You can see here the 980's fans aren't spinning because it's quite cool, just idling. They turn on automatically if it gets hot (and it turns them on one at a time). The 360 meanwhile just seems to have its fans on all the time... It's just idling in that picture, and had been for about 30 minutes. But it still wants its fans on.
Inevitably, there's more cables going around. The 980 is hungry and takes
two 8-pin PCI-E connectors, which is both of the ones I have on my PSU. So for the 6-pin connector the 360 wants, I had to resort to an adapter. Power consumption didn't take too big a hit though, because these cards don't use much when idling. I've no intention of using them both at once, so I'll always have at least one idling. Right now my PC pulls around 300w during the GTA 5 benchmark through a 500w PSU. So I feel pretty safe there.

Having two graphics cards from two manufacturers running at once turned out to be quite easy in Windows 10 (not tried under Ubuntu yet). OSes seem to have been progressively sandboxing programs more and more over the years, drivers included. So we're now at a point where you don't even need to reboot for updated graphics drivers. I guess that same thing has kept the two separated, because making them both work was as simple as installing the drivers for both and making sure they're both plugged into a monitor. I have HDMI cables going from each one to the same monitor, and I swap the source on the monitor depending on which card I want. It's kind of annoying having the PC 'pretend' to have 3 monitors. But mirroring the screen across both stopped me from being able to choose different cards in the graphics config for rF2/SCE/AMS and other games, which defeats the reason I did this in the first place.

60fps in a field of 20 AI cars. It's hardly a 'next gen' look, but it's a lot prettier than driver crashes!
Bring it on, Sebring!
"April... You absolute bitch" - Neil 15/09/2017