• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

3d instability, CPU vs GPU

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Zerileous

Senior Member
Joined
Jun 21, 2002
I've attempted to return to gaming after a significant hiatus, and the instability that I previously assumed was due to my mining undervolt is persistent. System specs in sig.

What happens: Any time the PC is subjected to a moderate 3D load, it crashes. This is not a normal crash, there is a clicking noise and everything goes dark, including any motherboard lighting etc. The PC will not respond until the PSU power switch is toggled for about 5 seconds. Even if it sits overnight.

What is a "moderate 3D load? Superposition 1080p high or extreme will crash as soon as the scene loads. Doom will crash as soon as you get into action with multiple bad guys, firing guns, etc, but will load and you can run around. Superposition 1080p medium and heaven run fine.

What have I tried? DDU and fresh driver re-install. Bios defaults on motherboard. Switched to factory "low power" BIOS using the bios switch on the card. Even in the low power (200W mode) it does the same thing.

What as the card been through? Since I bought it about 4 years ago, I put a FC water block and had it in a custom loop. Even hotspot temperatures were always reasonable. I did flash a liquid cooling bios which raises both the power limit and core voltage. I mined with the card for about one winter (i.e. 6-8 months) using an undervolt and memory overclock. The card was drawing less than 200w and temps were always great. It was not stable to do other things such as game/youtube, but ran the miner fine and was okay for normal browsing/forums.

I think something is happening to cause the PSU to trigger one of its protections. I don't know if its damage to the power regulation of the card or if its the PSU itself having issues with the current the card is trying to draw. I am having a hard time monitoring during testing because the built in logger for the driver doesn't run in full screen mode, :(, and the logger for HWiNFO64 has so much in it I can't make sense of the CSV, since even the header line runs multiple lines on my monitor. Since this seems like a power issue, I am pretty hesitant to ask a friend to borrow their parts for testing and either subject their PSU (which is probably not as robust) to a card that has faulty power draw or vice versa subject their card to a PSU that is at fault.

Can anyone make heads or tails of this?
 

caddi daddi

Godzilla to ant hills
Joined
Jan 10, 2012
and run these commands to run system file checker from an elevated command prompt.

DISM.exe /Online /Cleanup-image /Restorehealth

sfc /scannow
 
OP
Zerileous

Zerileous

Senior Member
Joined
Jun 21, 2002
An update, even running a light load long enough (CS:GO for 20 mins) caused the same to happen.
 

EarthDog

Gulper Nozzle Co-Owner
Joined
Dec 15, 2008
Location
Buckeyes!
Vega64... oldschool... maybe bad gpu.. ?

Since the card is under water and a pita to swap into another system, see if a new psu does it. A quality 550w unit will easily do it.
 
OP
Zerileous

Zerileous

Senior Member
Joined
Jun 21, 2002
Sounds good. Thanks for the suggestions everyone. My plan is to give the case a thorough cleaning, remove the cable extensions and re seat all the power cables. If the issue persists I'll try to get a budget PSU or just accept the downtime and send it to Seasonic for testing.
 

don256us

Uber Folding Senior
Joined
Jul 17, 2003
My first thought was bad GPU but... low voltage BIOS? PSU can definitely be the issue if it is not giving the GPU enough power or "POWA!" if you prefer. I don't.

I would make sure that the GPU is getting all of the power that it wants before calling it dead. Try a non low voltage BIOS?

Reasons for my thought is that 3D requires voltage and your issue happens when loading 3D programs. I don't think that it would be something like direct x or some other API would it? APIs are generally NOT something that I look at but it could.

How old is the PSU? Is it the Seasonic 750 you list in your sig? If it's old or been through a few power outages, it may be done. So PSU could be the issue but swapping in another PSU wihtout installing it could easily serve as a test.
 
OP
Zerileous

Zerileous

Senior Member
Joined
Jun 21, 2002
I appreciate all the suggestions guys. I'm pretty sure the problem is fixed, although I haven't done any overclocking or tuning or extensive testing. I removed my cable extenders and plugged everything directly into the PSU, and I haven't had one of these "trip" crashes since. So far I've just been playing CS:GO and DOOM. I did have one more typical GPU crash where the screen went blank but fans and lights continued. I wouldn't be surprised if the GPU has degraded somewhat between being run at nearly 300W gaming and the mining, plus being 4 years old. I'll have to do more experimenting and testing, though I'm limited to a couple hours a day right now.

Regarding the BIOS, the card shipped with a dual BIOS switch, one was a stock 220W BIOS, the other a "quiet" 200W BIOS. I flashed the stock 220W BIOS with a 260W "liquid cooling" BIOS that was used on other reference cards that shipped with a hybrid/AIO cooler. It has been this way since 2018.

Regarding component age, everything was purchased and assembled in November/December 2018, except for a few upgrades (CPU 2600x ->3700x, doubled memory, added an SSD).