Question Games crash after overclocking the GPU but I can't find a solid reason why ?

[Tankvsdps]

AMD Ryzen 9 5900X 12-Core Processor

Ram: 32,0 GB 3600hz

GPU: EVGA RTX 3070 8GB

NVME SSD: 1.82TB

MOBO: gigabyte a520i ac

Windows: 11 24H2

850 gold

BF6 was on beta, and I decided to probably OC my GPU was all good until I tried playing other games. They seem to crash when hitting OC setting higher than the settings below --VVVV

I used MSI Afterburner

settings: Core voltage +100

Core clock (mhz) +110

Memory clock (mhz) +1200

power limit (%) 106

Temp limit ( c ) 87

Fan speed: auto

TEMPS are not higher than 73 when playing at this settings, even did a run of Heaven benchmark with no problem

Errors: Game crashing errors

Games: The finals

an unreal process has crashed: EU-Discovery

CurrentQueue.Fence.D3DFence->GetCompletedValue() failed at .\Runtime/D3D12RHI/Private/D3D12Submission.cpp:994 with error DXGI_ERROR_DEVICE_REMOVED with Reason: DXGI_ERROR_DEVICE_HUNG

Game throne and liberty

An unreal process has crashed: UE4-TL-1.363.22.3983

GAME: CYBER punk 2077

random error no code was given.

I am somewhat aware of the limit on the GPU/PC, but not enough, because the temperatures seem good. so that is not the reason I am posting. It's more like I want to find the reason, and IF a solution is out there. I have looked around and tried multiple things that have not given a real result besides re-installing Windows, which allowed me to land in a lower OC setting, but stable, good fps with the same results.

Things I have tried to solve this problem, but have not worked.

deleting programs such as game bar, razer app, xbox apps etc etc

disabling Ram profile

sfc /scannow

Re- installing game/s

updating all drives

I have done some other regit and service stuff too but also no luck. Any advice are welcome

 

[MEMOFLEX]

AMD Ryzen 9 5900X 12-Core Processor

Ram: 32,0 GB 3600hz

GPU: EVGA RTX 3070 8GB

NVME SSD: 1.82TB

MOBO: gigabyte a520i ac

Windows: 11 24H2

BF6 was on beta, and I decided to probably OC my GPU was all good until I tried playing other games. They seem to crash when hitting OC setting higher than the settings below --VVVV

I used MSI Afterburner

settings: Core voltage +100

Core clock (mhz) +110

Memory clock (mhz) +1200

power limit (%) 106

Temp limit ( c ) 87

Fan speed: auto

TEMPS are not higher than 73 when playing at this settings, even did a run of Heaven benchmark with no problem

Errors: Game crashing errors

Games: The finals

an unreal process has crashed: EU-Discovery

CurrentQueue.Fence.D3DFence->GetCompletedValue() failed at .\Runtime/D3D12RHI/Private/D3D12Submission.cpp:994 with error DXGI_ERROR_DEVICE_REMOVED with Reason: DXGI_ERROR_DEVICE_HUNG

Game throne and liberty

An unreal process has crashed: UE4-TL-1.363.22.3983

GAME: CYBER punk 2077

random error no code was given.

I am somewhat aware of the limit on the GPU/PC, but not enough, because the temperatures seem good. so that is not the reason I am posting. It's more like I want to find the reason, and IF a solution is out there. I have looked around and tried multiple things that have not given a real result besides re-installing Windows, which allowed me to land in a lower OC setting, but stable, good fps with the same results.

Things I have tried to solve this problem, but have not worked.

deleting programs such as game bar, razer app, xbox apps etc etc

disabling Ram profile

sfc /scannow

Re- installing game/s

updating all drives

I have done some other regit and service stuff too but also no luck. Any advice are welcome

If the games are crashing after an attempted overclock of thr GPU then that is the first place I would start! It sounds like an unstable OC.

Return the GPU to its original values and retest

 

[LordVile]

AMD Ryzen 9 5900X 12-Core Processor

Ram: 32,0 GB 3600hz

GPU: EVGA RTX 3070 8GB

NVME SSD: 1.82TB

MOBO: gigabyte a520i ac

Windows: 11 24H2

BF6 was on beta, and I decided to probably OC my GPU was all good until I tried playing other games. They seem to crash when hitting OC setting higher than the settings below --VVVV

I used MSI Afterburner

settings: Core voltage +100

Core clock (mhz) +110

Memory clock (mhz) +1200

power limit (%) 106

Temp limit ( c ) 87

Fan speed: auto

TEMPS are not higher than 73 when playing at this settings, even did a run of Heaven benchmark with no problem

Errors: Game crashing errors

Games: The finals

an unreal process has crashed: EU-Discovery

CurrentQueue.Fence.D3DFence->GetCompletedValue() failed at .\Runtime/D3D12RHI/Private/D3D12Submission.cpp:994 with error DXGI_ERROR_DEVICE_REMOVED with Reason: DXGI_ERROR_DEVICE_HUNG

Game throne and liberty

An unreal process has crashed: UE4-TL-1.363.22.3983

GAME: CYBER punk 2077

random error no code was given.

I am somewhat aware of the limit on the GPU/PC, but not enough, because the temperatures seem good. so that is not the reason I am posting. It's more like I want to find the reason, and IF a solution is out there. I have looked around and tried multiple things that have not given a real result besides re-installing Windows, which allowed me to land in a lower OC setting, but stable, good fps with the same results.

Things I have tried to solve this problem, but have not worked.

deleting programs such as game bar, razer app, xbox apps etc etc

disabling Ram profile

sfc /scannow

Re- installing game/s

updating all drives

I have done some other regit and service stuff too but also no luck. Any advice are welcome

It’s your OC then? GPUs don’t really OC well anymore anyway. Kinda died around pascal. The boost Algorithm's are so good now that a manual overlock will only have a tiny performance boost over stock if you can even get a stable one.

You’re now better off undervolting the GPU to unlock more thermal headroom for the algorithm to dynamically boost the card.

 

[SkyNetRising]

AMD Ryzen 9 5900X 12-Core Processor
Ram: 32,0 GB 3600hz
GPU: EVGA RTX 3070 8GB
NVME SSD: 1.82TB
MOBO: gigabyte a520i ac
Windows: 11 24H2

I decided to probably OC my GPU
was all good until I tried playing other games. They seem to crash

What PSU?

 

[Tankvsdps]

If the games are crashing after an attempted overclock of thr GPU then that is the first place I would start! It sounds like an unstable OC.

Return the GPU to its original values and retest

Ah sorry maybe I didn't word it correctly. I did that and landed on the listed OC settings I listed. However I had it running on higher OC sittings when I was playing BF6 beta. I am just trying to figure out why it's unstable with the other game when the temperature are max than 70c - 75c with current settings but if I OC a little more it will pass some benchmark but when running game it will not work.

 

[Tankvsdps]

It’s your OC then? GPUs don’t really OC well anymore anyway. Kinda died around pascal. The boost Algorithm's are so good now that a manual overlock will only have a tiny performance boost over stock if you can even get a stable one.

You’re now better off undervolting the GPU to unlock more thermal headroom for the algorithm to dynamically boost the card.

It made games go from 60fps to 85fps avg I think it still works well. I am not big in OC but isn't undervolting for temperatures going down? Current not even hitting 80c and I had read max this gpu goes is 87. So I should had room to boost it a bit mor3

 

[Tankvsdps]

What PSU?

Oh sorry forgot to add that. It's a 850 gold evga I think. But that make me think maybe I should see if I still have the box with maybe extra cables for the gpu

 

[MEMOFLEX]

Ah sorry maybe I didn't word it correctly. I did that and landed on the listed OC settings I listed. However I had it running on higher OC sittings when I was playing BF6 beta. I am just trying to figure out why it's unstable with the other game when the temperature are max than 70c - 75c with current settings but if I OC a little more it will pass some benchmark but when running game it will not work.

Temperatures are only one thing that need to be kept an eye on. Instability can present itself in many different ways and under different circumstances. A test like Heaven might be fine but then it crashes in a game. This is not stable and either the speed needs dialing back or voltage is insufficient or the card is exceeding its limits.

Most gpus and cpus are binned at the manufacturer nowadays with only minimal additional headroom available in a lot of circumstances. They are often pre overclocked to some extent and this may well be the limit of the silicone.

If your card was fine without any crashes prior to you attempting to overclock then this is what you need to return to and just accept that there is not any scope for pushing the card any further. If this is no longer stable then there may be something else that is an issue as well.

How are you recording GPU temp? I would download something like HWINFO and see what all the temps are saying including hotspot and memory temperature.

 

[naczelnik5]

If you want to be sure your card's overclocking or UV is rock-solid, test any changes to its parameters in Metro Exodus Enhanced Edition. In-game, set all graphics settings to maximum and unlock the frame rate. You don't even have to play the game; just open the game menu and leave it for a few hours, preferably overnight, and monitor your settings using something like MSI Afterburner. You don't need benchmarks or tests in several other games. If Metro doesn't throw an error, you shouldn't make any other game either. This game is the biggest graphics card killer.
And yes , classic overclocking is no longer used in current cards. Today, cards perform best when UV-transmitted. Lower temperatures allow for higher card performance and clock speeds while maintaining the graphics card's power consumption.

 

[Tankvsdps]

Temperatures are only one thing that need to be kept an eye on. Instability can present itself in many different ways and under different circumstances. A test like Heaven might be fine but then it crashes in a game. This is not stable and either the speed needs dialing back or voltage is insufficient or the card is exceeding its limits.

Most gpus and cpus are binned at the manufacturer nowadays with only minimal additional headroom available in a lot of circumstances. They are often pre overclocked to some extent and this may well be the limit of the silicone.

If your card was fine without any crashes prior to you attempting to overclock then this is what you need to return to and just accept that there is not any scope for pushing the card any further. If this is no longer stable then there may be something else that is an issue as well.

How are you recording GPU temp? I would download something like HWINFO and see what all the temps are saying including hotspot and memory temperature.

Ah okay thank you I was only aware of temperature being the reason, but now I see that other stuff might be cause the problem. I use openharware monitor for temps and the max temp I have seen is 72. so temperatures are the the problem and that's why I was confused.

 

[MEMOFLEX]

Oh sorry forgot to add that. It's a 850 gold evga I think. But that make me think maybe I should see if I still have the box with maybe extra cables for the gpu

We would need to know the exact model number and age of the PSU as there are good and bad PSUs. Unfortunately Gold and 850w is not a uniform good or bad

If you dont have the box the psu will have a label on the side which will provide the full model number.

 

[MEMOFLEX]

Ah okay thank you I was only aware of temperature being the reason, but now I see that other stuff might be cause the problem. I use openharware monitor for temps and the max temp I have seen is 72. so temperatures are the the problem and that's why I was confused.

Temp can be an issue but there are multiple temps. Does your software give you memory temp and hotspot temp for the gpu?

Does the gpu crash when you return it to stock?

 

[Tankvsdps]

Temperatures are only one thing that need to be kept an eye on. Instability can present itself in many different ways and under different circumstances. A test like Heaven might be fine but then it crashes in a game. This is not stable and either the speed needs dialing back or voltage is insufficient or the card is exceeding its limits.

Most gpus and cpus are binned at the manufacturer nowadays with only minimal additional headroom available in a lot of circumstances. They are often pre overclocked to some extent and this may well be the limit of the silicone.

If your card was fine without any crashes prior to you attempting to overclock then this is what you need to return to and just accept that there is not any scope for pushing the card any further. If this is no longer stable then there may be something else that is an issue as well.

How are you recording GPU temp? I would download something like HWINFO and see what all the temps are saying including hotspot and memory temperature.

Ah okay thank you I was only aware of temperature being the reason, but now I see that other stuff might be cause the problem. I use openharware monitor for temps and the max temp I have seen is 72. so temperatures are the the problem and that's why I was confused

We would need to know the exact model number and age of the PSU as there are good and bad PSUs. Unfortunately Gold and 850w is not a uniform good or bad

If you dont have the box the psu will have a label on the side which will provide the full

I will post the details of it when I get home

edit: https://tweakers.net/pricewatch/172..._rDKvuATnRUGO6W_NV_LlYLlIv6dUE6YaAraAEALw_wcB

 

[Tankvsdps]

Temp can be an issue but there are multiple temps. Does your software give you memory temp and hotspot temp for the gpu?

Does the gpu crash when you return it to stock?

I think for the GPU it's only 1 temp I will need to double check this.
edit: i have checked and it only shows me 1 temperature for the GPU.

The GPU itself it's unaffected negatively by the OC as far as I can tell . Just the game crashing. Right now its running the lower OC settings with no crashing or other problems.

 

[LordVile]

It made games go from 60fps to 85fps avg I think it still works well. I am not big in OC but isn't undervolting for temperatures going down? Current not even hitting 80c and I had read max this gpu goes is 87. So I should had room to boost it a bit mor3

cards now auto OC based on temperature and fan curves. If you undervolt the card to run cooler while leaving the fan curve the same then the algorithm has more thermal headroom to boost,

 

[Ralston18]

Any error codes, warnings, or even informational events being captured in Reliability Histoyr/Monitor and/or Event Viewer?

Look for entries that occur just before or at the time of the crashes.

 

[Tankvsdps]

cards now auto OC based on temperature and fan curves. If you undervolt the card to run cooler while leaving the fan curve the same then the algorithm has more thermal headroom to boost,

i will try to look up how to undervolt it and hope it works better.

 

[Tankvsdps]

Any error codes, warnings, or even informational events being captured in Reliability Histoyr/Monitor and/or Event Viewer?

Look for entries that occur just before or at the time of the crashes.

i have no errors from my PC itself only the game errors

an unreal process has crashed: EU-Discovery

CurrentQueue.Fence.D3DFence->GetCompletedValue() failed at .\Runtime/D3D12RHI/Private/D3D12Submission.cpp:994 with error DXGI_ERROR_DEVICE_REMOVED with Reason: DXGI_ERROR_DEVICE_HUNG


besides that nothing else

 

[Ralston18]

Re: "errors". Just to be sure - no warnings or informational entries of any sort referencing the PC?

In Device Manager:

Any devices flagged for errors?

= = = =

Reading back I saw that "sfc /scannow" was run. Run "dism" if that was not done, then run "sfc /scannow" again.



= = = =

Referencing the DXGI errors. I googled a bit to see if anything relevant would be found:

https://www.nvidia.com/en-us/geforce/forums/geforce-experience/14/150018/dxgi-error-device-removed/

https://www.makeuseof.com/dxgi-error-device-removed-error-windows/

There were other links as well.

My suggestion is to look at the two links and others that you find.

Read carefully and study the offered solutions - key is to not do anything until the problem is narrowed down or perhaps specifically identified.

Notes :

1.) Do not be tempted into downloading any utilities or tools claiming to fix the problem.

2.) Do not run any scripts or other code that may have been used by someone to fix the problem.

3.) Stay out of the Registry. Registry edits are a last resort and require a full system backup before attempting.

 

[Tankvsdps]

Re: "errors". Just to be sure - no warnings or informational entries of any sort referencing the PC?

In Device Manager:

Any devices flagged for errors?

= = = =

Reading back I saw that "sfc /scannow" was run. Run "dism" if that was not done, then run "sfc /scannow" again.



= = = =

Referencing the DXGI errors. I googled a bit to see if anything relevant would be found:

https://www.nvidia.com/en-us/geforce/forums/geforce-experience/14/150018/dxgi-error-device-removed/

https://www.makeuseof.com/dxgi-error-device-removed-error-windows/

There were other links as well.

My suggestion is to look at the two links and others that you find.

Read carefully and study the offered solutions - key is to not do anything until the problem is narrowed down or perhaps specifically identified.

Notes :

1.) Do not be tempted into downloading any utilities or tools claiming to fix the problem.

2.) Do not run any scripts or other code that may have been used by someone to fix the problem.

3.) Stay out of the Registry. Registry edits are a last resort and require a full system backup before attempting.

i ran the " dism " and after the "sfc /scannow " and it said nothing was founded
I will look into those links and see if any of those work. Thank you very much