0.17 + Radeon GPU = constant random CTD's

The CRASH simulator HAS CRASHED. Color me surprised. Or just mad.

Doesn't matter if playing, editing, or loading a level and leaving it sit.
AMD KERNEL driver crashes with (from event viewer in Windows)
"Display driver amdkmdap stopped responding and has successfully recovered."
The game crashes with 0x0000001 error and 'your GPU stopped responding' even though the GPU works in EVERYTHING else. I can't afford to put out any more money to play this game properly in the near future. 0.16.x was working well once I rooted out the PNG skybox issue, but this is the second time in a row with major updates that make the game unusable. Doesn't need to be my map, doesn't need mods, it just crashes hard-core constantly. Sometimes I can get 2 minutes, sometimes an hour, and everything in-between.

Either that or the day before you push a public update to Steam *LET ME KNOW* and I'll do my best to bust it (normally takes me <5 minutes) ASAP. Not begging for a private beta here (of all people), but I seem to be good at crashing the crash simulator.

Posted my problems about crashing during editing to the steam forum, it's been there under bug reports for days. Whatever changes you've made to the renderer, it's CRASHING the AMD Kernel-mode display driver, especially when I am close to the terrain in the editor either adjusting decal roads or especially in the F1 object editor as soon as the icons come into view (like when you want to copy-paste a decal road to use a similar one). It DOES NOT require editor use at all, though, and I haven't used the F10 editor.

Tested on the July 28th 2019 driver and the driver from within the past few days (just installed).
Windows 10 Pro, RX 480 8gb, Ryzen 3700x stock, 32GB 3000mhz CL15 memory, NVME SSD, Soundblaster Audigy RX 7.1 sound-card, and an external Blu-ray / DVD-RW. That's the complete computer (besides PSU and Case). Machine is otherwise STABLE, it's ALL NEW parts except the GPU. The GPU will run Fallout 4 all night and that's rather demanding of both of the CPU and GPU. It will run RIMWORLD all night and day (good lord knows it does, too!). Does fine browsing all day and night. DOES NOT RUN BEAMNG WITHOUT CRASHING. GRRRRR...
Original Steam post made before I realized the entire game (not just the F11 editor) was unstable...
https://steamcommunity.com/app/284160/discussions/2/3185654583887106683/

At one point, even the crash submitter had crashed. No other applications have crashed on this PC except THIS one. Not a heat issue (nothing goes over 80~85c here). Not a power issue (I just checked it to be sure before I readied this new machine for production use), and it's all-new parts except the video card, which is 2~3 years old and kept very clean. I tested the heck out of this thing, too.
It's not the editor, needn't use the editor, just spawn in and leave it sit. In the time it takes to take a dump, it 100% ALWAYS takes a dump, too (the screen does not idle off during this time, that's set for 45 minutes).
*sigh* Back to 0.16 I go, I go. Again.
Don't take it personally, but I'm starting to feel like a Porcupine in a Nudist Colony with this game this summer.

EDIT: Still crashing in 17.0.2

 

Im having the same exact issue, gives vague 0x00000001 crash error, im updating the game now to see if it works better, did you try submitting a support ticket?

 

Yes I did. I managed to get one through before locking up the crash-reporter here the rest of the times. Everything but this game works beautifully. I used this card in my old computer for a little over two years, and 0.16.x (not to mention EVERY OTHER GAME I have) works just fine. It's just the rendering engine updates are doing something at random that the AMD driver just can't handle.
The crash report ticket number is on the steam post I linked in the OP (original post).
I've tried updating the video driver. I have tried updating chipset drivers. It's a fairly fresh install of Windows. Did check-disk stuff on command prompt (always do after system locks up, which this game sometimes does if the video driver doesn't restart). I haven't tried updating the BIOS yet, but I will as soon as I back-up my map online later, due to the RISK involved with updating BIOS.
Interesting that you have a GTX 1070.

That means it's Nvidia cards, too, but it's a RYZEN processor. I have a Ryzen processor too, and Windows 10 Pro, as you do. I swear if this is because of Windows 10...
*cusses a blue streak about Windows 10, as no one hates Windows 10 more than THE BOB* *lots of really spicy but CENSORED words*

So maybe I should call @tdev @synsol or @Nadeox1 to that discovery. It's not just Radeon cards...
Might be some instruction passed to the Display Drivers in the game-engine that works fine with intel's (more liberal / vulnerable) CPU's and gets shut-down by AMD processors. This would not be the first time I've heard of it (as there are plenty of fundamental differences in the architecture) but I wouldn't know off the top of my head what it's doing, that's causing it. It's random. It happens ANY time a level is loaded. I can load into ECA or my map and without touching anything, BOOM. Only thing I pulled out of the new ECA is a few road textures for edging and wear, but it was crashing even while I was playing ECA.

If anyone - ANYONE - knows a debugger I can use for this, to catch the instruction or op-code doing this, give me a shout.

I spoke with @Gamergull about this earlier today as he was concerned not having heard too much of it - but there were other reports of it on Steam forums by a few people.

Tried with no mods - makes no difference in frequency - tried 'safe mode', that crashes too just the same. It makes no difference if mods are there at all on this one. I wish mods caused it as at-least I could edit and finish my map, it's been absolute [censored] trying to get my map update out with this going on.

If it is Ryzen processors causing it, Nadeox has a Ryzen (if I remember correctly), and they're also on the best-seller list for a while now, so I'd be quite concerned about it.

Re-installing directX using the redistributable file inside the BEAMNG install folder did not fix it or change frequency of the crashes.

Created another ticket.
57974

 

That would be great. I'll check for it on Steam when I wake up tomorrow (it's 4 AM here my time).
Debug branch will help me give YOU more information. This is ALWAYS good.

I have tried everything, I have NO CLUE what's causing it at all. The machine doesn't seem to crash or hang otherwise, and it's not on fire / throwing heavy smoke out... so I am out of ideas here.

Debug branch will be great though, awesome. I am uploading my map shortly to my personal backup server.* Whatever is crashing out the game happens when I edit or happens even playing East Coast with no editing.
Just for clarification, I was not using the F10 editor, it was F11 editor. F10 editor scared me away with 'warning, this will break things' message.
*I can link you to a copy of the map tomorrow if that helps, but it's not needed to cause the issue. It's my Los Injurus Map that I've been working on for a year. Even just running ECA will crash the same way.

Gracias Luis

 

Before changing GPU two days ago, I was using that exact RX480 8GB with an intel processor, no issues for me.
Sorry if I missed this anywhere in your post, but have you done a full clean install of your drivers?
I'll be getting a Ryzen 7 processor sometime next week and I will come back to this post with any info that might be helpful if I do have problems.

 

Awesome, can always use more help. Don't forget to post your new machine's prowess on the Banana-Bench forums.
This is a fresh OS install, so yes, technically the drivers were clean-installed. I installed my own drivers before connecting it to the internet. I also re-installed directX twice now, too.
I know about using DDU (Display Driver Uninstaller) so that program comes in super-duper handy. I don't know what's going on with this so hopefully the debugger works out OK.
It just HATES me.
Off-topic:

Spoiler

Enjoy your Ryzen, I can do between 15~21 vehicles here on the 3700x depending on how fast you want to run it (it gets close to 20fps sometimes with 21 vehicles). This is with only 3000mhz memory in dual channel, cas latency of 15. The infinity fabric on the processor is running at 1500mhz (get low latency 3200~3600mhz memory if you're in the market, infinity fabric will increase with RAM speed!). Don't go over 3733mhz however.
Or haul 66 pianos in a T-series trailer, at full simulation speed. My 4790k could barely manage 8~11 vehicles before it got too slow, and mine was configured pretty well.
Using Micron-based memory here, they had stuff cheap at Micro-Center. It was on sale. Was not on the QVL but you may wish to mind your QVL for best results. Also purchased cheapest full-ATX board (Asrock Phantom Gaming 4) I could get in the x570 chip-set. Would do it again!
Go to Micro-Center if you can get to one if in the US, it's worth it. I drove 4 hours each way that day to make that trip.
Color me Amazed. Now if I can just get BeamNG 0.17.x to stop committing suicide... how dare I have fun?

--Cheers!

 

I have a ryzen 7-1700x, 24gb ram, gtx1070 with 8gb ram. Getting same crashes and/or pc shuts down completely! I even recently installed windows on a new ssd, etc etc. I disabled the MS xbox gaming crap, but that didn’t solve the issues.

If/when the pc gives me the option to send report, i send it to you guys (beamng devs). Hope that helps.

Also, seems more prevalent when I am recording (either shadowplay or obs)... weird.

 

Just for reference with a 4690K@4.5GHz and a R9 Fury I've got no crashing issues, running on latest AMD drivers

 

Follow-up: My Ryzen 7-1700X was set for "Gaming Mode" in BIOS so I turned that off. So far, that seems to have settled things down, though the problem was random, so I will keep an eye on it. No idea if this info helps or not, but I figure anything that might lead to a clue is worth mentioning.

 

That may only be a placebo affect of there just being less cores to do work. Sometimes I get 60~90 minutes out of BeamNG, sometimes I get one minute. It's 100% random.
This is also some bit of the old 'the faster the PC, the faster it crashes' adage from the Pentium days (or there-about).

Super-technical info that might help the staff or other hardware guru's.
To the best of my knowledge, only 14nm (1xxx all, 2xxx APU only) and 12nm (2xxx, 3xxx APU only) Glo-Flo sourced Ryzen chips have 'Game Mode', and only for 'more than 4 cores/8 threads' I believe. The 7NM TSMC process chips (3xxx series) DO NOT feature 2 dies on the CPU until you get up over 8 cores, as the core-count per inner CCX module chip is doubled (this is also why there's up to 16 cores/32 threads on high-end 3xxx Ryzen, with up to 12 core / 24 threads currently on the market). So these 3xxx chips do not have game mode. This is due to 'Non-uniform memory access' (NUMA) which gives some cores less latency to system memory than others - less latency = higher fps - so less latency cores stay enabled in game-mode while the rest get shut off. Due to fundamental design difference with the 3xxx series (non-apu 3xxx), the 7nm TSMC chips actually use a 12nm or 14nm Glo-Flo sourced IO die that's separate from the main CPU cores which are on a 7nm TSMC-sourced cpu die, all of these within the CPU package itself. This way, with Ryzen, through Threadripper, and up to Epyc server processors, the entire line of chips can consist of 1 to 8 7nm 8-core cpu-dies, with one 12nm / 14nm controller IO die. This also leaves ALL processor cores with the same decent latency and number of hops to the system memory, and cache memory on other cores. The latency is also masked somewhat by the addition of much, much more cache (35mb on 8-cores, 70mb on 12 and 16-core dies), it's also possible to fuse off processing cores while leaving the (shared between all cores on the same 7nm die) level-3 (L3) cache memory intact in whole or in part.

To recap:
8 core 1700/1700x/1800x - 2 cpu dies on one cpu package 14nm - 4 cores 8 threads per 'CCX' - 6-core chips are the same with 1 cpu core fused off per CCX
8 core 2700/2700x - same as above with 12nm - same with CCX specifics - again 6 core chips are the same here - 4-core 2xxxG APU's are 14nm and single CCX is used + iGPU
8 core 3700x/3800x - one 12 / 14nm IO die, one 7nm CPU die - 8 cores 16 threads per 'CCX' (this also applies to 6-core 12-thread chips as they're currently one CCX and one IO die)
12 / 16 core Ryzen 3900x/3950x (respective) - one 12 / 14nm IO die, *TWO* 7nm CPU dies - 6/8 cores 12/16 threads per die (respective). 3900x just has 2 core / 4 threads disabled per CCX vs 3950x
4-core 3200G/3400G and similar 'G' chips are the only ones to be 12nm yet, 4-cores and 4 to 8 threads. One CCX is used + iGPU.
It's also worth noting that 3xxx Ryzen series (but not the APU series to best knowledge) have de-coupled the Infinity Fabric clock (which cores use to communicate) from RAM speed. It can now be 1:1, or 2:1 ratio of Ram speed to Infinity Fabric speed. Again this only pertains to the 3600,3600x,3700x,3800x,3900x,3950x(unreleased). Whether this regards EPYC or Threadripper is unknown.

So, that BOOK aside, there's no game mode (or reason for it) on 3xxx Ryzen, while it *DOES* exist on 1xxx and 2xxx chips. NUMA characteristics (hard to program for) almost necessitated the option for game mode for lightly threaded applications to get proper performance on Threadripper, and to a lesser extent, Ryzen desktop too.

Worth noting also, Ryzen when it hits the thermal limits, down-clocks in both clock-speed *AND* instructions-per-clock (IPC). This is important to know from a programming perspective. While I can't seem to get mine to crest 80C no matter what I do, HWINFO type software (like Afterburner, or CPU-Z / Core-temp) that shows clock-speed can be misleading if you manage to get your Ryzen toasty warm, and it seems to maintain it's speed. Also, keep in mind, that temperature / voltage logging software may be keeping Ryzen (and some intel) CPU's at boost-clocks, as the periodic check of the software is interpreted as a load by the CPU.

The BOB does LOTS of reading. LOTS and LOTS of reading. Cache numbers on 3xxx Ryzen are close but might be off by a meg or two.
I hope this wall of info helps someone.

Oh, and @davidinark when you get the money burning a hole in your pocket to buy a 3700x (8 core) or 3900x (12 core), you won't be disappointed at all. Wait for one on sale or grab a used one when 4xxx / 5xxx series hits and people sell theirs, you won't need a new board, just a BIOS update (req'd).

 

I would also like to talk about my experiences with this specific update:

My specs are:

• Ryzen 5 3600 @ Stock Voltage (1.4V) bordering 4.0 GHz most of the time
• AMD Radeon RX 5700 XT (Reference model, made by Sapphire, utilizing latest drivers, which are 19.8.1)
• 16GB (8x2) of HyperX Predator DDR4 RAM at 3200 MHz (CL16) utilizing the primary XMP made available by my motherboard
• Gigabyte Aorus Pro B450
• EVGA 600W 80+ White Certified PSU
• Kingston A1000 240GB NVMe
• Seagate Barracuda 7200RPM 2TB HDD
• Latest BIOS and chipset driver versions (F42a for the BIOS)

Not only is my system completely stable (minus Gigabyte software, of course) in other applications, but I stress tested it in multiple ways, such as benchmarks and stress tests (Furmark, CPU-Z) and the system remained stable and working, albeit toasty (never went above 90 degrees celsius with stock cooler, even with a few minutes of stress on the GPU, idling at around 45 degrees celsius). A few of those benchmarks can be found here (CPU-Z) and here (3DMark).

Deus Ex: Mankind Divided runs perfectly fine, other sim games are ok (Project Cars and Dirt: Rally if you consider those sims). But what happens when I play BeamNG is that moments after spawning, possibly seconds, the computer shuts down without a blue screen and I have to power cycle it to return it to normal function. I'm talking drop-dead crashing, where all fans stop spinning, peripherals shut down and only the RGB lights on the motherboard stay on. Once (thankfully) I got a page_fault_in_nonpaged_area blue screen after exiting BeamNG and attempting to install Logitech G Software, but the RAM is perfectly fine according to Thaiphoon Burner, along with the HDD and SSD themselves being perfectly healthy, with no faulty sectors.

I am afraid of playing BeamNG now as continuous power cycling is not exactly healthy. Removing peripherals did not help, and I might try setting the RAM to 2400 MHz tomorrow just to be sure, but for now, this situation has gotten really weird.

This game killed my GTX 770 (although it was not BeamNGs fault, mostly nVidias) and there is no way in hell I'm going to give it a chance to do the same to my brand new RX 5700 XT. The PSU is fine and voltages are stable, with minimal variation due to ripple voltages, and it should be enough to support the hardware I've currently got going, but I'm not taking any chances.

I'm willing to disclose more information regarding the conditions of how the crashes happened, but I cannot give logs as my computer gets orbit-nuked 80% of the time I try to spawn a car, especially on the East Coast map, where the "default" spawn guarantees my PC will kill itself. I really, really wanted to play BeamNG again and it's a shame this is happening, so I can try to help with all information I've got.

I could understand hardware-level shutdowns from extremely low level applications, but I've never seen a game cause this much commotion among my feeble components. Something really weird is happening and I hope it's on my side.

 

Steam forums are FULL of 'GPU NOT RESPONDING' errors. I linked this thread in all the forum threads on Steam that I could find, to send them here.

 

I also keep getting
Fatal Error - The GPU Is Not Responding
crashes since the 0.17 update. I've tried clearing the cache (which takes ages!), and deleting all mods. Still happening, though.

WIndows 7 64
nVidia GTX 1060 6GB with the latest drivers (I actually updated the drivers to try to fix this problem, but it's made no difference)
8GB system RAM

 

Do you have an AMD processor?
It's sounding like this is CPU related rather than GPU despite the error.

 

I'm using an Intel i5 2500.

 

Yep, "GPU not responding" here too I am way over spec, I deleted all mods, cleared cache, reinstalled game, deleted all the files in in the "Documents" folder, verified integrity, and now using the "work around" which works ONLY with one mod. No other mods work anymore. My computer goes nuts too and is running way to hard for this.

 

Im getting a gpu not responding issue even though before the update everything ran fine with no issues

 

Work-around uses 0.16.3.x which works fine here. All 0.16.x versions work fine here once I rooted out a lag-causing PNG skybox issue (old format graphics = mega lag) with it in July.
'Computer goes nuts and is trying way too hard to run this' - well part of that is as they make the engine more efficient to run the game, the computer can do more, so fans kick up and it puts out more heat, but that shouldn't be causing all these errors.

Yes, entirely. There's something in this new version of the game that's making the graphics driver panic. It's happening on nVidia as well as AMD Radeon. It's happening on older intel systems and laptops in addition to the newer Ryzen line of systems. Haven't seen any Bulldozer/Piledriver systems (FX series AMD) affected yet, but that's because most people using them for this game have upgraded to newer intel or Ryzen processors, for reasons which will be obvious (FX stinks!). Windows 7 is affected as well as Windows 10 so it's not version-dependent.
Something with the updates to the rendering engine have caused the graphics driver to time out or lock up, causing it to get reset. I noticed some odd hitching before it locks up, the graphics driver is obviously having trouble with some instruction.
Everything I read on this error is caused by a malformed line of rendering code that's crashing some systems but not others.

 

"the crash submitter had crashed"
that's how you know something's gone really wrong.