Re: Game Keeps Crashing don't know what to do HELP!

Well, now I can't get BF5 working with the 980ti either. These people really destroyed this game for some.

I noticed that the first time I play the game, it tends to last a lot longer before it crashes. I get like 60 mins in or more sometimes. Then when I try to persist in playing it after the first crash, the subsequent crashes get closer and closer together until it crashes once ever 2 or 3 minutes.

The fact that it gets worse lends itself to my theory that it may be the tensor cores don't have a well placed temp diode (or none at all), and they're overheating. Reminder: the temps reported by the stock thermometer(s) are great even under a stress test (75c max).

I will repeat (since people keep offering ideas / solutions that don't make sense): This is the ONLY game that crashes at all. I have many other games that use a ton of RAM (Cities Skylines is a supreme example), so the "my min  RAM voltage isn't enough" almost certainly doesn't apply to me. I have a ton of other games that max out my GPU usage and use a similar amount of vRAM. The only wildcards here are BF5 and ray tracing.

There are no crash logs generated; my BF5/Temp folder exists, but is empty.

Maybe there's more common hardware that exists between us than we realize. Here's a list:

7700k OCed to 4.9ghz
RTX 2080 OCed to +65mhz GPU / +675mhz vRAM
Gigabyte z170 Gaming 5
32GB RAM @ 3200mhz
Windows 10 Pro 1809
417.71 driver

Edit 6:40pm MST: I played about 5 mins before it crashed. I got about 2 or 3 micro-stutters per minute. I feels like it goes into a microstutter that just doesn't recover.

I noticed that the first time I play the game, it tends to last a lot longer before it crashes. I get like 60 mins in or more sometimes. Then when I try to persist in playing it after the first crash, the subsequent crashes get closer and closer together until it crashes once ever 2 or 3 minutes.

The fact that it gets worse lends itself to my theory that it may be the tensor cores don't have a well placed temp diode (or none at all), and they're overheating. Reminder: the temps reported by the stock thermometer(s) are great even under a stress test (75c max).

I will repeat (since people keep offering ideas / solutions that don't make sense): This is the ONLY game that crashes at all. I have many other games that use a ton of RAM (Cities Skylines is a supreme example), so the "my min  RAM voltage isn't enough" almost certainly doesn't apply to me. I have a ton of other games that max out my GPU usage and use a similar amount of vRAM. The only wildcards here are BF5 and ray tracing.

There are no crash logs generated; my BF5/Temp folder exists, but is empty.

Maybe there's more common hardware that exists between us than we realize. Here's a list:

7700k OCed to 4.9ghz
RTX 2080 OCed to +65mhz GPU / +675mhz vRAM
Gigabyte z170 Gaming 5
32GB RAM @ 3200mhz
Windows 10 Pro 1809
417.71 driver

I played about 5 mins before it crashed. I got about 2 or 3 micro-stutters per minute. I feels like it goes into a microstutter that just doesn't recover.

RTX 2070 OC for me... so slightly different enough that I cant discount your theory... however, of note is that everything in my case is air cooled, I am unsure of your config

AMD 2700x, stock speed, air cooled only

zotac RTX 2070 AMP extreme core edition, air cooled only

Asus ROG strix b450-f

16 gig RAM @ 3000 (capable of 3200 but I had stability issues, not in just bf5)

win pro 10 1809

417.71 driver

Basic 650 watt power supply (evga) with single 12v rail

Im always up for discussion in origin as well if anyone wants to chat offline...

other items of note: (please don't disable these if you are not familiar with the ramifications)

win firewall disabled

win defender disabled

UAC disabled

Further notes on my system (see my previous post):

I tried rolling back the overclock on EVERYTHING (including the factory overclock on my 2080). It made no difference - I still crashed just as often with or without the OC.

Since I have this AIDA64 widget visible on my second monitor constantly, I'm 100% certain my temps never spiked above 80 on my CPU, sitting at 60 - 75 typically while in BF5. My GPU sits at a steady 72 - 75c while in BF5 for long enough. I water cool my CPU and air cool my GPU in an extremely well ventilated case that I regularly dust (including making sure there's negligible dust on the radiator / heatsinks). I was just showing off my computer to co-worker less than 2 weeks ago, and I very thoroughly dusted it off then.

I also used a second widget GPU-Z a few times to confirm the temps. I made sure the temp poll rate is fairly high to ensure it's not spiking crazy high when it's not polling (to get accurate numbers).

Here's my idle temps:



Here's my temps after running BF5 for about 5 minutes (capped at 75fps due to 75hz display with vsync on, so load was at like 85%):



On my CPU, throttling begins at 100c. According to GPU-Z, throttling begins at 83c on my 2080. So I'm not hitting the thermal limit on anything ever.

So all this being said, I can safely eliminate the possibility of my computer overheating being why the game is crashing. I know that's a very real possibility with all the computer novices that surely play this game (nothing against any of you), but it's not for me.

Incidentally, I also have my CPU "unparked". Basically this just disables a default feature that down-clocks your CPU when it thinks the computer is idle - I had problems with microstuttering in BF4 and BF1, which this resolved.

Two other possibilities that can be eliminated:

I had a 7 year old, but fully functional power supply I decided to replace since it was nearing it's end of lifespan. I replaced it with a newer modular one. I had general computer crashing problems shortly after, so I had that new one replaced under warranty. The replacement did the same thing. So I went to Best Buy and replaced my video card with an identical one, and the computer crashing problem ceased. Nvidia replaced the defective one under warranty. The computer crashing problem has not returned since then. So that's 3 different cards, if you're keeping track.

All 3 cards had the BF5 crashing problem described in this thread. I'm including the original 2080 since it was working fine for a couple months, except BF5 would crash. So the possibility of defective card (the best buy temp one plus this replacement) is all but eliminated. The only possibility that remains for that to be defective is if there's an extremely large batch of defective 2080's, or there's a major design flaw with the 2080's. Since people are reporting the problem on other cards, this seems very unlikely (unless my tensor core theory holds true).

All 3 power supplies have the BF5 crashing problem present. It's astronomically unlikely any of the 3 power supplies are defective since the BF5 crashing problem was present on all 3 equally (and AIDA 64 reports very healthy voltages on all rails). So that possibility can safely be eliminated.

Sorry for the TL;DR, but I want to provide as much data possible to prove to the devs this is not an isolated issue with people with crappy computers and viruses and whatnot. I work for level 3 tech support, so I know what it's like to escalate an issue to the devs and have them bounce it back because they can't recreate the issue. This issue is 100% on BF5's side and / or Nvidia's side. Regardless, since BF5 is the only game that crashes, DICE needs to fix the issue, or work with Nvidia to identify a design flaw with their RTX cards.

With as many people with different configurations and different video cards and different rig set up’s, I am absolutely convinced that it is a software issue with dice. The preposterous part of this whole scenario is that Dice will not even respond and I have sent them no less than 100 crash reports. I have consulted an attorney and we are proceeding with putting together a class action lawsuit. During the discovery phase of the lawsuit, Dice will be required to supply the name and email address of every single person that has ever purchased the game to include them in the class-action suit. At that point I would think that losing millions of dollars might give them the kick in the * to get off their lazy * and FINALLY ADDRESS THE PROBLEM!