troubleshooting gpu GPU crashing - Gigabyte 5700xt OC crashes a lot

[Leskit]

Hi all,
I bought a Gigabyte 5700 Xt OC (revision 1) back in January 2020 but I've had nothing but constant yet intermittent crashes when playing games: cod warzone, Death Stranding, Ghost recon games, Subnautica 1 & 2. Warzone crashes usually only crash the game, whereas crashes in Death Stranding and Ghost recon games sometimes do a full BSOD. I've tried all of the available drivers for it and haven't found one that's more stable than another. I did undervolt + a slight overclock a few weeks ago, which was as stable as anything else I've had. I reset to default settings this morning and cod crashed after 3 minutes of run time. This problem has been going on since I got it and I'm completely fed up. I've tried running everything from max to minimum graphical settings and no luck. Let me know what info you need and I'll include it. I haven't had any overheating problems. It's intermittent too - sometimes cod or DS will crash in less than 3 minutes, sometimes it crashes after 2 hours, sometimes it doesn't crash during a plays session. I use a 1440p 75HZ acer monitor with a second 1080p 60hz on the side.

I'm getting my doctorate and use my PC for work every day so I haven't been able to spend a lot of time on the problem, including time-intensive tasks like a fresh windows install.

Setup:

Ryzen 2600 (no overclock)

Asus B-350-F gaming

Gskill 3200 16Gb using XMP settings

Samsung 960 evo m.2 ssd boot drive + other hard drives
EVGA supernova nex650g PSU

Windows 10

 

Things I haven't tried:

updating mobo bios
resetting ram to default non xmp settings
fresh install of windows
?

Thanks!

[Leskit]

bump (I'm allowed to do that, right?)

 

[viking_power]

I have an ASROCK 5700xt Phantom gaming D that was doing the same thing. Crashes and black screens constantly. I ended up RMA the card 2 times before I got a stable card. My second card was only stable in an overclock. I would look into an RMA as this might just be bad hardware.

[viking_power]

1 hour ago, Leskit said:

bump (I'm allowed to do that, right?)

 

Quoting due to forgetting too with earlier post.

[BTGbullseye]

4 minutes ago, viking_power said:

Quoting due to forgetting too with earlier post.

Only need to quote if it isn't the OP.

[Wheresmehammer]

I have the Aorus version of your card and had the same problems.

 

I reflashed the bios on my card on the OC bios and reinstalled the Windows 10 OS and installed 20.4.1 fresh and have not had an issue ever again, but i am scared to change drivers in case...

 

I play at 1080p 144hz but do use VSR too for down-sampling in some games.

[DoctorNick]

You sure your ram is stable? Maybe check bsod code / event viewer. Should be evident what is causing this.

[Leskit]

5 hours ago, DoctorNick said:

You sure your ram is stable? Maybe check bsod code / event viewer. Should be evident what is causing this.

Ran a quick 1-pass memtest and no errors were present. I found one of the death stranding crashes and I'm attaching it. One of my rare BSOD crashes from cod was a driver_irql_not_less_or_equal error from amdkmdag.sys
Would updating my mobo bios be likely to make a meaningful difference?

[Leskit]

7 hours ago, Wheresmehammer said:

I have the Aorus version of your card and had the same problems.

 

I reflashed the bios on my card on the OC bios and reinstalled the Windows 10 OS and installed 20.4.1 fresh and have not had an issue ever again, but i am scared to change drivers in case...

 

I play at 1080p 144hz but do use VSR too for down-sampling in some games.

I'm hesitant to flash the vbios, having never done it before. It is running the latest bios already as far as I can tell though. Unfortunately, I can't afford the downtime of a fresh reinstall right now - I use my pc for work/school on a daily basis.

[Wheresmehammer]

19 minutes ago, Leskit said:

I'm hesitant to flash the vbios, having never done it before. It is running the latest bios already as far as I can tell though. Unfortunately, I can't afford the downtime of a fresh reinstall right now - I use my pc for work/school on a daily basis.

Do you have integrated graphics or a spare card at all? I would just use whatever is reliable and leave gaming for a while, or game at considerably lowered settings on another card.. assuming a person would have a lower end one? For example i have a spare HD 4850 & GTX 650.

[Leskit]

9 minutes ago, Wheresmehammer said:

Do you have integrated graphics or a spare card at all? I would just use whatever is reliable and leave gaming for a while, or game at considerably lowered settings on another card.. assuming a person would have a lower end one? For example i have a spare HD 4850 & GTX 650.

Unfortunately I don't, I sold my old 660ti a few months ago. note to self: always keep a spare if you can!

[DoctorNick]

11 hours ago, Wheresmehammer said:

I have the Aorus version of your card and had the same problems.

 

I reflashed the bios on my card on the OC bios and reinstalled the Windows 10 OS and installed 20.4.1 fresh and have not had an issue ever again, but i am scared to change drivers in case...

 

I play at 1080p 144hz but do use VSR too for down-sampling in some games.

Did you ever update the bios to the version with higher fan speed?

https://www.gigabyte.com/Graphics-Card/GV-R57XTGAMING-OC-8GD-rev-10/support#support-dl

[Leskit]

9 hours ago, DoctorNick said:

Did you ever update the bios to the version with higher fan speed?

https://www.gigabyte.com/Graphics-Card/GV-R57XTGAMING-OC-8GD-rev-10/support#support-dl

I did not, but I run a more aggressive custom fan curve anyway. my bios version is 017.001.000.049.000000. GPU at 64, gpu mem at 78c, and hot spot at 86c during heaven engine 4.0 benchmarking, so temps are good. 

[DoctorNick]

4 minutes ago, Leskit said:

I did not, but I run a more aggressive custom fan curve anyway. my bios version is 017.001.000.049.000000. GPU at 64, gpu mem at 78c, and hot spot at 86c during heaven engine 4.0 benchmarking, so temps are good. 

Maybe try it anyway and without 3rd party program for fan curve/overclocked

[Wheresmehammer]

10 hours ago, DoctorNick said:

Did you ever update the bios to the version with higher fan speed?

https://www.gigabyte.com/Graphics-Card/GV-R57XTGAMING-OC-8GD-rev-10/support#support-dl

My reflash was the bios from the TPU database.

[Leskit]

From another forum, I tried the following: 

1) Reset WattMan settings and lower the GPU power limit value to -30%. If this prevents crashes, your PSU is probably not enough. (still crashed as before)

 

2) Reset WattMan settings and lower the GPU maximum voltage to 1162mV. If this prevents crashes, your GPU voltage is probably too high for your system. (still crashed as before)

3) Reset WattMan settings and lower the GPU maximum frequency by 100Mhz with 1162mV max. voltage. If this prevents crashes, your PSU is enough but cannot operate the GPU voltage stably. (Modern Warfare/Warzone is much more stable, only 1 crash in 5 days, however, Death Stranding is still crashing after less than hour of playtime.)

my PSU passed OCCT tests without any issues, but I think I've narrowed it down to the GPU  and PSU. I'm going to do a burn-in session with a bunch of different stress tests with the lowered frequency and voltage.

Worth noting that I actually had this undervolted to 1100 mV, but with +45% power, and it was crashing a ton. So I think my PSU is riding the line of what it can supply (EVGA nex650G) which is really disappointing.