Hello,

By End August 2021, I had purchased a new EVGA RTX 3080 Ti FTW3 Ultra Gaming from a German retail Hardware-Shop. However, I couldn't insert/use my new-bought EVGA GPU for over a quarter year because around the same time, my motherboard (ASUS Maximus XI Hero, Z390) suddenly died for no apparent reason and it took that said quarter year waiting game of sending it (the motherboard) to ASUS, then ASUS rejecting RMA (an absolute d***-move!!) and then finally buying a used examplar of the same motherboard in January 2022 - and shortly afterwards (End January 2022), I could finally insert and start operating my then-new EVGA RTX 3080 Ti. Immediately after insertion, I installed the then-latest Nvidia driver for Windows 7 (472.12 WHQL) and then flashed the latest firmware/vBIOS via Precision X1 software.

In the first months, everything was fine, because I almost exclusively played very old games (Q3A-Engine based games from the early 2000's) during that time period, which didn't really tax my GPU at all (very light load).

Then, however, I started playing newer games (from March 2022 onwards) that quite caused real GPU load and Battlefield 4 was the first game where this event occured: everytime I entered the player customization/loadout menu during a multiplayer session, just after a couple of seconds, my entire computer suddenly shut off - at the same moment/second, when the screen went suddenly black in front of me, I could hear the fans firing up to full-speed VERY LOUD for a moment - though, I couldn't exactly pinpoint, whether it was the GPU's fans or if it was my Seasonic PSU's fans or both. And immediately (within 1-2 seconds) after shutdown, the PC automatically started up again with literally zero error messages - no POST or BIOS error messages and also no Windows error messages (except the pre-boot notification before reaching Windows desktop, that my system wasn't shut down properly the last time...).

And this phenomenon also occured in almost every newer game that I played during that time: Crysis 3, Deus Ex: Mankind Divided and the original Crysis (from 2007, not the Remaster - but with my own custom ExtremeQuality config that is extremely stressing both the GPU and particularly the CPU because of the exceptionally high polycount). The only exception from this was Battlefield 3: no matter how long I played or where in the game/menus I was, there was no sudden emergency shutoff not even a single time. Crysis (2007), on the other hand, caused this phenomenon to occur particularly rapidly (but not instantly like BF4 Loadout Menu); usually within 30 minutes of starting the game.

For all games, the FrameRateLimiter was set to 120fps at maximum either ingame or in Nvidia Inspector software, but in some cases (Crysis 2007) I set it to just 60fps.

Then, in July 2022, I had the idea to undervolt (but NOT underclock) my RTX 3080 Ti in order to fix this problem - and indeed! It seemed to be solved: no more sudden shutoffs during high-load gameplays!

...until about 1-2 months ago, when the exact same problem started re-occur again. Moreover, now even very old games (the Q3A-Engine based games mentioned above) cause the exact same phenomenon: sudden emergency shutoff followed by immediate automatic restart of the computer! And it gets worse: from now on, neither undervolting and even additionally underclocking (fixing GPU clock around 1,8 GHz in V/F-Editor, reducing memory clock from OC'ed 10000 MHz to 9000 MHz - that's 500 MHz below the default clock!) my 3080 Ti do not solve this problem anymore!

And I'm finally out of options (ideas) now...

My PSU is a Seasonic Prime TX-850 (850W, 80 PLUS Titanium, Single Rail) purchased new just a few months before my EVGA RTX 3080 Ti.
 
Looking forward for any help, support or advise...

T.S.O.M.

The Spirit of Morpheus
Hello,

By End August 2021, I had purchased a new EVGA RTX 3080 Ti FTW3 Ultra Gaming from a German retail Hardware-Shop. However, I couldn't insert/use my new-bought EVGA GPU for over a quarter year because around the same time, my motherboard (ASUS Maximus XI Hero, Z390) suddenly died for no apparent reason and it took that said quarter year waiting game of sending it (the motherboard) to ASUS, then ASUS rejecting RMA (an absolute d***-move!!) and then finally buying a used examplar of the same motherboard in January 2022 - and shortly afterwards (End January 2022), I could finally insert and start operating my then-new EVGA RTX 3080 Ti. Immediately after insertion, I installed the then-latest Nvidia driver for Windows 7 (472.12 WHQL) and then flashed the latest firmware/vBIOS via Precision X1 software.

In the first months, everything was fine, because I almost exclusively played very old games (Q3A-Engine based games from the early 2000's) during that time period, which didn't really tax my GPU at all (very light load).

Then, however, I started playing newer games (from March 2022 onwards) that quite caused real GPU load and Battlefield 4 was the first game where this event occured: everytime I entered the player customization/loadout menu during a multiplayer session, just after a couple of seconds, my entire computer suddenly shut off - at the same moment/second, when the screen went suddenly black in front of me, I could hear the fans firing up to full-speed VERY LOUD for a moment - though, I couldn't exactly pinpoint, whether it was the GPU's fans or if it was my Seasonic PSU's fans or both. And immediately (within 1-2 seconds) after shutdown, the PC automatically started up again with literally zero error messages - no POST or BIOS error messages and also no Windows error messages (except the pre-boot notification before reaching Windows desktop, that my system wasn't shut down properly the last time...).

And this phenomenon also occured in almost every newer game that I played during that time: Crysis 3, Deus Ex: Mankind Divided and the original Crysis (from 2007, not the Remaster - but with my own custom ExtremeQuality config that is extremely stressing both the GPU and particularly the CPU because of the exceptionally high polycount). The only exception from this was Battlefield 3: no matter how long I played or where in the game/menus I was, there was no sudden emergency shutoff not even a single time. Crysis (2007), on the other hand, caused this phenomenon to occur particularly rapidly (but not instantly like BF4 Loadout Menu); usually within 30 minutes of starting the game.

For all games, the FrameRateLimiter was set to 120fps at maximum either ingame or in Nvidia Inspector software, but in some cases (Crysis 2007) I set it to just 60fps.

Then, in July 2022, I had the idea to undervolt (but NOT underclock) my RTX 3080 Ti in order to fix this problem - and indeed! It seemed to be solved: no more sudden shutoffs during high-load gameplays!

...until about 1-2 months ago, when the exact same problem started re-occur again. Moreover, now even very old games (the Q3A-Engine based games mentioned above) cause the exact same phenomenon: sudden emergency shutoff followed by immediate automatic restart of the computer! And it gets worse: from now on, neither undervolting and even additionally underclocking (fixing GPU clock around 1,8 GHz in V/F-Editor, reducing memory clock from OC'ed 10000 MHz to 9000 MHz - that's 500 MHz below the default clock!) my 3080 Ti do not solve this problem anymore!

And I'm finally out of options (ideas) now...

My PSU is a Seasonic Prime TX-850 (850W, 80 PLUS Titanium, Single Rail) purchased new just a few months before my EVGA RTX 3080 Ti.
 
Looking forward for any help, support or advise...

T.S.O.M.

Generally a spontaneous reboot or shutdown is caused by a defective PSU.

what is your CPU and other system specs? Any overclocking?
 
From what information you've provided, either your PSU is failing or your GPU is and either problem was exacerbated initially by the current spikes from a a full powered GPU and eased by dropping those spikes (reducing the power the card draws). There were definitely earlier power supplies from Seasonic (and just about everyone else) that couldn't handle the spikes from 3080ti/3090(and some 3080's). Most offered RMAs and almost every company has updated their lineup now to be more tolerant of the spikes. 

If your using a riser cable, plug the GPU directly into the MB, for troubleshooting
 
if GPU is in MB already, check that it is fully seated & locked in place.  Confirm your three PCIe power cables from the PSU to the GPU are fully installed & locked
 
Did your PSU came with additional PCIe power cables, you could try swapping them (Only use cables that came with the PSU or that you have confirmed are wired for that PSU)
 
I was also thinking it could be the power spike issue mentioned in the post above
 
Do you have a spare PSU to test with? (Only use cables that came with the PSU)

HeavyHemi
Generally a spontaneous reboot or shutdown is caused by a defective PSU.

But now it also happens on older games, which only cause negligible GPU load (newer & new games caused serious load with power spikes waaay over 400W for the GPU alone! Undervolting fixed that... until 1-2 months ago). And this event clearly happens ONLY when the GPU is engaged! (= when playing games)
 

B0baganoosh
what is your CPU and other system specs? Any overclocking?

 
CPU: Intel Core i9-9900K @ 4.8 GHz (All-Core & Cache)
RAM: Dual Channel 4x 8GB (all memory slots occupied) G.Skill Trident-Z RGB DDR4-4000CL17 CR2 overclocked @ DDR4-4266CL16 CR2 & tweaked sub-timings
Storage: 1TB NVMe-SSD (PCIe 3.0 x4) and 8TB SATA3-HDD
 
I removed any form of overclocking (CPU, RAM) once this problem started occuring again - so right now, the CPU is @ 3.6 GHz All-Core & Cache and the memory is even below it's spec @ DDR4-3200 CL16 CR2 & default sub-timings. In accordance with the removed overclocking, I also significantly/massively reduced CPU & RAM voltages.
 
But that didn't fix the problem - it's still present!
 

B0baganoosh
From what information you've provided, either your PSU is failing or your GPU is and either problem was exacerbated initially by the current spikes from a a full powered GPU and eased by dropping those spikes (reducing the power the card draws). There were definitely earlier power supplies from Seasonic (and just about everyone else) that couldn't handle the spikes from 3080ti/3090(and some 3080's). Most offered RMAs and almost every company has updated their lineup now to be more tolerant of the spikes. 

So which one is it? Should I RMA the GPU or the PSU? Or both?
 
T.S.O.M.

If you take a look at HWiNFO64, can you put a slight load on the gpu and tell what the +12v rail voltage (and also the 3.3v and 5v) are showing for minimums? You can look at the number reported in the gpu section for pcie +12v rail and the numbers reported in the motherboard section (launch the app with sensors only). It will help to have the idle voltages and then with any load that hopefully doesn't cause a shutdown.

Could be either the psu or gpu. Best thing to do would be to test the gpu in another known good working system.

Sajin
Could be either the psu or gpu. Best thing to do would be to test the gpu in another known good working system.

...or take the video card to a local PC repair store and ask them if they can check it, but they might charge you a fee

B0baganoosh
If you take a look at HWiNFO64, can you put a slight load on the gpu and tell what the +12v rail voltage (and also the 3.3v and 5v) are showing for minimums? You can look at the number reported in the gpu section for pcie +12v rail and the numbers reported in the motherboard section (launch the app with sensors only). It will help to have the idle voltages and then with any load that hopefully doesn't cause a shutdown.

I was abroad for over 6 weeks, so I didn't really have the opportunity to respond to this topic - Sorry for that! I had done the inspection runs You've requested (playing games while AIDA64 logs everything) already before being abroad but had not the time/opportunity anymore to post them here because of the hassle of traveling preparations in the last few days before departure...
 
Is it possible to attach FILES? So, I can upload the entire AIDA64 log files ("Generic Log Viewer" tool necessary to view these files! You can download it here:  https : // www . hwinfo . com /forum/threads/logviewer-for-hwinfo-is-available.802/) instead of posting a screenshot of each entry one by one...


Otherwise, here is a link that contains all 4 log-files in .csv format: https : // filebin . net / obbufrh3bfb4d7o3
 
T.S.O.M.

post above restored from spam filter
 
 
Limits on new & low post count members
 
Learn your way around the EVGA Forums, Rules & limits on new accounts Ultimate Self-Starter Thread For New Members