0.17 + Radeon GPU = constant random CTD's

"GPU" not responding I tried clean install on drivers, different drivers, re-installing the game, deleting all mods, safe mode, verify, clearing cache, and about everything else I can think of and nothing works. I can only use the "work around" with one mod. Any other mods at all it crashes. Also, my computer is running wide open. i7 4720HQ, 980M GTX 4GB, 32GB RAM, 256 SSD.

 

OK folks. I want everyone to go out and try something here. It's not really risky, but you should give it a shot.
I've noticed something here - my video card - only on Windows 10 - is going just 0.5~1.2% over it's set 100% speed. Say you set the clock to 1200mhz core on the GPU. You then have it going 1% faster automatically - 1212mhz core on the GPU. That's plenty enough to cause a crash of these sorts. I don't know why this is happening across different brands, or what the case may be, but imperfect PCI-express bus clock speeds (which can differ by 1~2% either way per vendor) can cause this among other things.
My XFX Radeon RX 480 8GB 'XXX' edition card is factory-clocked at 1288mhz. Normal stock reference boards I believe are 1250mhz. This one won't do 1300mhz without a small power limit or core voltage bump to maintain it's voltage during long sessions under load, as I had tested it when I first got it in late 2016. For some odd-ball reason, an issue is cropping up, where the card is now trying to do 1300mhz (give or take 3~5mhz as it jumps around a bit), which it CAN NOT do. See attached picture of what happens when I under-clock my card to 1200mhz, using MSI AFTERBURNER.
The link in the name above will allow you to download it from Guru3D, which is safe-for-work. I am only attaching a link to this utility so that you can lower the max clock of the card a few % (15~25mhz) and see if it helps. WARNING: Overclocking your card could murder it and cause lots more issues, don't do it unless you know the risks if your computer lets out some - or all - of the 'magic smoke'. Don't come crying if you overclock it and it casts FIRE 3 spell in your mum's basement. You have been warned!
So as you can see, there's about 0.5~1.2% of it's original clock-speed - this can and will cause crashes on some but not all - cards, as not all are created equal. Some cards are a little better than others, some will barely do their rated speed from the factory.

So if your card is 1000mhz core clock, lower it by 15mhz. If it's 1500mhz core-clock, lower it by 20~25mhz, if it's 1800mhz core-clock, lower it by 30~40mhz. Use a calculator to find 5% and lower it by that and slowly bring it back up so that your MAX core speed does not go over it's factory rated speed (even if by accident, which is the case). If you're experienced, you can mess with voltages, but this causes heat, and can bypass the safeties built into the components, to protect them (and you) from overheating and breaking or causing other somewhat catastrophic issues like the above-mentioned magic smoke coming out.

I don't know WHY it's doing this yet, or what the root cause is. It could even be the Windows power plan settings of all things (it's happening on intel cpu's with nvidia GPU's, and on AMD offerings for CPU and/or GPU as well - so it's not 'just AMD' or 'just intel' or 'just nVidia' or any combination there-of. I just know what's causing the error & that I am sorry I didn't install MSI Afterburner sooner on the new machine.

Has this 100% fixed the issue - NO. It's worked-around the issue. The issue is the card is boosting past it's ability and anyone without the top 20% card-yields (best GPU silicon of said product run) might have an issue the same as me (my card is pretty 'meh' for overclocking). However, this *WILL* cut down - and may entirely stop - your crash-to-desktop issues. Surprisingly, no other game has showed this crashing issue. My game has run for a few hours without crashing when it would crash within a minute or two - or up to an hour before. A bit early to rule a final verdict yet.

Stupid technology!

I am otherwise at a loss for words & I am completely clueless as to who to "threaten to sit-on" over this whole debacle. Can't blame the BeamNG developers for this one, honestly.

Attached screenshot showing 'set 1200mhz core clock' yet '1211mhz' core-clock is held. Factory clock = 1288mhz, it's set lower for testing reasons, I could go to 1250~1270 with the current issue.
You can set the exact clock speed you want for your card by pulling the slider for GPU CORE speed to the left, and using the LEFT or RIGHT arrow keys to 'nudge' it in 1mhz increments.

AGAIN, do *NOT* overclock it, we're trying to underclock it slightly by a few % to stop the crashing when it goes over it's rated speed here, NOT blow it up.

FINE PRINT: Neither BeamNG nor myself is be held liable for you blowing up your computer or anything else with this software or methods described here-in, use only as intended. If you get a funny rash, fever, or lay square eggs: talk to your doctor.

Derp? *head banging on desk*
EDIT: YES THIS WORKS! (not the head banging on desk, underclocking the GPU stops the crashing)
I edited my map all day after this and nary an issue.

TODO: Find out how many people this helps. Let me know if it helps you by responding to this forum.

 

Maybe helpful maybe not, but I finally got the R7 1700, overclocked too.
Traffic spawned, italy map, half hour gameplay, no crashes or errors on my end. I hope you all sort out whats causing your issues!