GPU crashes in every game unless severely underclocked.

[Jack Rollins]

I have a Zotac GTX 1080 Ti AMP Extreme edition.

It crashes in every game even when cool and even when low usage.

 

I bought it used a while ago and it always had this issue however it wasn't as bad and I only had to underclock a tiny bit, and I got it for a good price so didn't see the point in returning it.

However ever since then it has been gradually getting worse, I now have to do a -300 core and memory underclock.

 

Both game crashes and whole PC crashes take place, and I also get artifacts before crashing

 

Reinstalled Windows and Drivers.

Changed PSU.

Reapplied thermal paste.

Problem persists.

 

I know that graphics card do degrade overtime like most PC components, but I would really appreciate if anyone had any ideas on how to fix this. I was considering baking the GPU but have only been told to do this as a very last resort.

[Moonzy]

13 minutes ago, Jack Rollins said:

graphics card do degrade overtime like most PC components

Generally for stock clocks, it's not an issue

It's definitely considered defective, and that die is literally on the edge of stability to begin with for a slight degradation to affect stock clock speed

 

That said, that card is probably overclocked from the factory, so there's that

And low quality PSU can kill GPU from poor ripple quality, there's that too

[Brooksie359]

6 minutes ago, Moonzy said:

Generally for stock clocks, it's not an issue

It's definitely considered defective, and that die is literally on the edge of stability to begin with for a slight degradation to affect stock clock speed

 

That said, that card is probably overclocked from the factory, so there's that

And low quality PSU can kill GPU from poor ripple quality, there's that too

Yeah that would be my guess as well. A bad psu can stress the vrm and cause damage to it in return making the gpu stability finicky as it can't regulate the voltage as well as it should. Granted it could be a completely different issue but that is the biggest one I can think of. 

[MadGoatHaz]

Two things here:

 

First, let's get the specs of the PSU you are using. These cards can spike 370 watts. If you don't have 35a available to the card alone, it will trip OCP and reset the system. A lot of the time people will buy a multi rail psu with plenty of wattage for their system, but not realize that the rail providing power to the GPU is insufficient. A way to test this is to use the power slider in afterburner, set it down to 60% and game/ test. If it does not reset or crash, it's your psu. 

 

Second, the VRMs on those cards are known to get very hot. I would check the thermal pads under the VRM heatsink and make sure they are still in contact and the springs are still applying enough pressure. 

 

 

[Jack Rollins]

3 hours ago, MadGoatHaz said:

Two things here:

 

First, let's get the specs of the PSU you are using. These cards can spike 370 watts. If you don't have 35a available to the card alone, it will trip OCP and reset the system. A lot of the time people will buy a multi rail psu with plenty of wattage for their system, but not realize that the rail providing power to the GPU is insufficient. A way to test this is to use the power slider in afterburner, set it down to 60% and game/ test. If it does not reset or crash, it's your psu. 

 

Second, the VRMs on those cards are known to get very hot. I would check the thermal pads under the VRM heatsink and make sure they are still in contact and the springs are still applying enough pressure. 

 

 

Thank you for your message.

Currently using a 1600W EVGA PSU, and calling it overkill is an understatement. Before I had a 750W and had the same issue, got my current PSU cheap and thought buying it was one way to troubleshoot the problem. 

I've heard that about the VRM's too in a Gamer Nexus video, I have very good airflow so I doubt that is the issue, but it might be worth adding some more thermal pads.

 

Are there any other things that you could think of to why the card isn't working well?

And what are your opinions on the baking idea?

[MadGoatHaz]

Well I'm not a fan of evga's power supplies, but that should not be the issue then. (Especially if it's been replicated between two PSUs now)

 

I would still check under the VRM sink. Even with adequate air flow, if the pads are dried up or the sink itself doesn't have enough pressure it will still cause problems. In doing this I would also clean and inspect for cold or broken solder connections, as well as damaged components on the board. Specifically around power delivery (6&8 pin Pcie power plugs mainly) and all caps. Reasoning being, you'd want to repair individual solder joints before trying to reflow to board completely. (By baking) 

 

Now my thoughts on reflow via baking. I have done it many times for GPUs and an assortment of troublesome PCBs with many successes. It's a delicate procedure however and as you stated, very much a last resort. I would much rather take the time to inspect the board with a can of electro-kleen, magnification and repair it the "right away" (with a soldering iron) before ever jumping to home reflow. 

 

Furthermore, if you have a dead cap for instance that is causing dirty power, you simply have to replace that component to repair the card. 

 

 

[Jack Rollins]

22 hours ago, MadGoatHaz said:

Well I'm not a fan of evga's power supplies, but that should not be the issue then. (Especially if it's been replicated between two PSUs now)

 

I would still check under the VRM sink. Even with adequate air flow, if the pads are dried up or the sink itself doesn't have enough pressure it will still cause problems. In doing this I would also clean and inspect for cold or broken solder connections, as well as damaged components on the board. Specifically around power delivery (6&8 pin Pcie power plugs mainly) and all caps. Reasoning being, you'd want to repair individual solder joints before trying to reflow to board completely. (By baking) 

 

Now my thoughts on reflow via baking. I have done it many times for GPUs and an assortment of troublesome PCBs with many successes. It's a delicate procedure however and as you stated, very much a last resort. I would much rather take the time to inspect the board with a can of electro-kleen, magnification and repair it the "right away" (with a soldering iron) before ever jumping to home reflow. 

 

Furthermore, if you have a dead cap for instance that is causing dirty power, you simply have to replace that component to repair the card. 

I checked the GPU and there doesn't seem to be any damage, VRM's are making good contact with the heat sink and all solder connections look to be good.

 

Before I try the baking method, I would just like to add these errors that always come up in event viewer when I have a crash.

"

The description for Event ID 13 from source nvlddmkm cannot be found. Either the component that raises this event is not installed on your local computer or the installation is corrupted. You can install or repair the component on the local computer.

If the event originated on another computer, the display information had to be saved with the event.

The following information was included with the event: 

\Device\Video3
Graphics Exception: ESR 0x516648=0x3d0009 0x516650=0x0 0x516644=0xfb3ffa 0x51664c=0xe000010f

The message resource is present but the message was not found in the message table
"

and

"

Display driver nvlddmkm stopped responding and has successfully recovered.

"

Does this help at all?