Major consistant BSOD after changing GPU

[SniperRu]

So I just bought a new RTX 3060 Ti 8Gb to replace my old GTX 970. After installing it, there is now a consistent complete freeze of my computer (I suspect a BSOD but I can't see it) when playing DCS and DCS only. The only thing I can then do is a hard reset using the power button. Seeing as I mainly bought the GPU for DCS, this is really frustrating. I am running the latest open beta and disabled all mods.

Things I've tried so far :

• Changing pagefile (all possible configurations)
• Reinstalling the GPU drivers (using DDU)
• Changing the graphics preset to medium
• Stress testing/benchmarking my GPU using a specialized app (no crash happened)
• I've tried running Arma 3 and Escape from Tarkov but they both run fine. However only DCS runs the GPU at full capacity.

Other details on the crash :

• It seems to be happing after about 10 minutes in game
• The map, mission, planes, player in-cockpit or out, only AI on map - nothing seems to affect the BSOD and its timing
• When it happens, these is a screeching sound and everything freezes on both my screens

dcs.log DxDiag.txt

[Rudel_chw]

7 minutes ago, SniperRu said:

Things I've tried so far :

• Changing pagefile (all possible configurations)
• Reinstalling the GPU drivers (using DDU)
• Changing the graphics preset to medium
• Stress testing/benchmarking my GPU using a specialized app (no crash happened)
• I've tried running Arma 3 and Escape from Tarkov but they both run fine. However only DCS runs the GPU at full capacity.

 

Try this:

 

Rename the folder "C:\Users\User\Saved Games\DCS\"  to DCS.bak

Restart DCS, it will re-create a default "C:\Users\User\Saved Games\DCS\" folder, you will be asked for your user credentials and you may need to setup your graphics options anew.

Does DCS still crashes 10 minutes later?

 

 

[SkateZilla]

Winkey + R, "Eventvwr.exe"

Under Logs, System, Look for Hardware Errors.

I guarantee you, that the GPU is Overheating or there's a Power Problem.

As I've told others, Stress Testing a GPU doesnt really prove anything, because it stresses only the GPU, that's the only heat source, and the only power draw.

Playing DCS Taxes the GPU, CPU, RAM and Disk Drives.

Now, Stress test that GPU, while Running a CPU and Memory Stress Test, and see if your system remains stable.

Edited July 21, 2023 by SkateZilla

[SniperRu]

46 minutes ago, Rudel_chw said:

 

Try this:

 

Rename the folder "C:\Users\User\Saved Games\DCS\"  to DCS.bak

Restart DCS, it will re-create a default "C:\Users\User\Saved Games\DCS\" folder, you will be asked for your user credentials and you may need to setup your graphics options anew.

Does DCS still crashes 10 minutes later?

 

 

Seems to have helped a bit. The game kept running for about 30-40min (in mission time), but then the crash happened when I exited back to the editor. Same manner as usual - screeching sound and everything frozen 

41 minutes ago, SkateZilla said:

Winkey + R, "Eventvwr.exe"

Under Logs, System, Look for Hardware Errors.

I guarantee you, that the GPU is Overheating or there's a Power Problem.

As I've told other, Stress Testing a GPU doesnt really prove anything, because it stresses only the GPU, that's the only heat source, and the only power draw.

Play DCS Taxes the GPU, CPU, RAM and Disk Drives.

Now, Stress test that GPU, while Running a CPU and Memory Stress Test, and see if your system remains stable.

Went through the logs, in the "System" logs, the only "critical" ones are due to 

The system has rebooted without cleanly shutting down first. This error could be caused if the system stopped responding, crashed, or lost power unexpectedly.

Source : Kernel-Power

Nothing in there about the GPU. Additionally I usually monitor the GPU temperature and it never goes above 74ºC in DCS. I also checked the required power supply on Newegg using my gear and it says 699W and my PSU is rated at 750W

[SkateZilla]

21 minutes ago, SniperRu said:

Seems to have helped a bit. The game kept running for about 30-40min (in mission time), but then the crash happened when I exited back to the editor. Same manner as usual - screeching sound and everything frozen 

Went through the logs, in the "System" logs, the only "critical" ones are due to 

The system has rebooted without cleanly shutting down first. This error could be caused if the system stopped responding, crashed, or lost power unexpectedly.

Source : Kernel-Power

Nothing in there about the GPU. Additionally I usually monitor the GPU temperature and it never goes above 74ºC in DCS. I also checked the required power supply on Newegg using my gear and it says 699W and my PSU is rated at 750W

you need to log temps and power draw.

what you're experiencing is a hardware reset at the bios level.

[SniperRu]

4 minutes ago, SkateZilla said:

you need to log temps and power draw.

what you're experiencing is a hardware reset at the bios level.

How would you do that with the game running?

Does this mean I have to choose between this GPU and playing DCS?

[SkateZilla]

Run a a Hardware Monitor App and have it overlay on your screen.

nVidia/AMD's Drivers do this natively already.

[Nevyn]

Try setting your nvidia drivers to debug mode and see what happens.

 

 

 

 

[diego999]

As Skatelliza says, you're most likely running into power problems.

Your estimated power consumption (keyword: estimated) is close to your PSU max rating. A spike could send it above the limit and trigger a freeze or a restart.

[SniperRu]

Downloaded GPU-Z, started logging 

30 minutes later and 2 missions down still no crash.... murphy's law

The whole time the power draw does not exceed 200W total from the PSU ports. Usually one is 100-120W and the other 70-80W.

Temperature stayed below 80ºC

 

Will try again tomorrow

[SkateZilla]

14 hours ago, SniperRu said:

The system has rebooted without cleanly shutting down first. This error could be caused if the system stopped responding, crashed, or lost power unexpectedly.

Source : Kernel-Power

Nothing in there about the GPU. Additionally I usually monitor the GPU temperature and it never goes above 74ºC in DCS. I also checked the required power supply on Newegg using my gear and it says 699W and my PSU is rated at 750W

 

Here's the Rub, 
How old is your PSU, (if you can supply the model# as well).

As PSU's age, their ability to convert and supply power diminishes, 
PSU's degrade quicker w/ cheaper components (ie off brand, 80+ bronze and lower).

So what was 750 watts 8 years ago, after general everyday use and wear on a cheaper PSU, is prolly 675watts now, and the rails would be really sensitive to spikes.

Even with your current config, you shouldn't be pulling more than ~550watts.

 

Another thing you'll have to look at is mainboard age, you just slammed a 16 Lane PCIe4 Card in a mainboard that's prolly PCIe 2.0, 3.0 depending on age.

Edited July 21, 2023 by SkateZilla

[SniperRu]

ok ran the test again, crashed maybe 5 minutes in. Logging did not catch any spikes, but since the whole computer crashes then it probably is a power issue

6 hours ago, SkateZilla said:

 

Here's the Rub, 
How old is your PSU, (if you can supply the model# as well).

As PSU's age, their ability to convert and supply power diminishes, 
PSU's degrade quicker w/ cheaper components (ie off brand, 80+ bronze and lower).

So what was 750 watts 8 years ago, after general everyday use and wear on a cheaper PSU, is prolly 675watts now, and the rails would be really sensitive to spikes.

Even with your current config, you shouldn't be pulling more than ~550watts.

 

Another thing you'll have to look at is mainboard age, you just slammed a 16 Lane PCIe4 Card in a mainboard that's prolly PCIe 2.0, 3.0 depending on age.

 

Power Supply is an Antec HCG-750M bought new in 2015.

Just spent 500$ on a GPU and now have to buy a new PSU as well 

Any recommendations are welcome

[SkateZilla]

6 minutes ago, SniperRu said:

ok ran the test again, crashed maybe 5 minutes in. Logging did not catch any spikes, but since the whole computer crashes then it probably is a power issue

Power Supply is an Antec HCG-750M bought new in 2015.

Just spent 500$ on a GPU and now have to buy a new PSU as well 

Any recommendations are welcome

Best case buy a new 750w PSU, and if it doesnt solve the mainboard hard crashing take it back.

[corn322]

Source: https://www.tweaktown.com/reviews/5792/antec-hcg-750m-750-watt-80-plus-bronze-power-supply-review/index.html

Are your power connectors only connected to one of the +12V rail, or are you split between both of them?

Each side can only supply so much power.