3080TI Founders Edition unexpected shutdown issue

[Kenpachi2k]

After recieving my 3080TI Founders Edition, I have come across an issues when running the GPU at load (playing a game + streaming) my computer just unexpectedly shut down - 4 shut downs within the span of 2hrs, event viewer doesn't really help me as much as I don't understand what things are but one that comes up the most HID-compliant headset, which I read on most reddit forums that headset drivers can create some issues with the GPU sound drivers - tried a fix and the PC worked for a solid 1-2hrs underload with 30-40 minutes being run on Heaven stress test at max. Still recieving unexpected shutdowns but less often. I am now trying an undervolt (started 9:45pm) as of to see if reducing temps and power consumption would help and has been running at load just gaming at max settings for 2hrs no unexpected shut downs. Would really like some help on this issue as im unsure where to look for whats making this issue.

Heres my current build:
CPU: AMD Ryzen 9 5900X 3.7 GHz 12-Core Processor
CPU Cooler: ARCTIC Liquid Freezer II 280 72.8 CFM Liquid CPU Cooler
Motherboard: Gigabyte X570 AORUS ELITE ATX AM4 Motherboard
Memory: Crucial Ballistix 32 GB (2 x 16 GB) DDR4-3600 CL16 Memory
Storage: Samsung 970 Evo 1 TB M.2-2280 NVME Solid State Drive
Storage: Samsung 860 QVO 1 TB 2.5" Solid State Drive
Storage: Seagate Barracuda Compute 2 TB 3.5" 7200RPM Internal Hard Drive
Storage: Seagate Barracuda Compute 2 TB 3.5" 7200RPM Internal Hard Drive
Video Card: NVIDIA GeForce RTX 3080TI 12 GB Founders Edition Video Card
Case: Phanteks Enthoo EVOLV ATX TG ATX Mid Tower Case
Power Supply: EVGA GA 850 W 80+ Gold Certified Fully Modular ATX Power Supply

 

[Kenpachi2k]

Now I’m confused.

sorry typo, its updated but I do have the 3080TI.

 

[Kenpachi2k]

should be ok on power supply side of things with 850w.
is cpu overclocked?

the cpu is undervolted.

 

[Kenpachi2k]

Try default everything maybe? Surely that's the first thing to do.

Alright I will try give that a go, so far still gaming now it's been going fine.

 

[Kenpachi2k]

Is it just switching off like some one pulled the power ? Does it auto restart (depends if your motherboard bios is set to recover from a failure) or stays totally off and then you have to wait a bit before you can power it back on ?

If so it's the PSU not able to run that setup. How old is the psu and have you made sure to use more than one rail on it for the PCIE connections to the gpu and are you using seperate cables from the PSU to the gpu not the one cable with two connectors ?

It's like someone pulled the power but with all the LEDs on the mobo and ram stay on and the system stays off until i hit the power button to turn it back on. It use to do this with my 3080 vision once in a while but that just restarts when it happened. The whole build is very recent, everything new since January 6th 2021. I have also made sure to use seperate cables to link into the 12pin splitter that Nvidia gave and done the same with the 3080 vision.

 

[Kenpachi2k]

It's sounding like a PSU issue then mate, if it was doing it with the 3080 and now the 3090ti. Maybe the over current protection on the PSU is doing it as these gpus power spike a lot.

Easy test set the motherboard bios to defaults and even xmp off for ram, then boot into windows and run cinebench multi core test in a loop and run heaven at same time to stress gpu, if it shuts down right away its PSU, you can also use the cpu-z stress test too for the cpu that works it hard too.

Alright will give this a go now.

 

[Kenpachi2k]

Easy test set the motherboard bios to defaults and even xmp off for ram, then boot into windows and run cinebench multi core test in a loop and run heaven at same time to stress gpu, if it shuts down right away its PSU, you can also use the cpu-z stress test too for the cpu that works it hard too.

Ran a full cycle of cinebench multi core looping for 10 minutes alongside heaven maxed out, system is at default and didn't shut down once.

The symptoms you're having sounds 99% like your PSU's over current protection kicking in to trip it to prevent 1) PSU from blowing up and 2) Starting a fire

You should try a newer PSU

Although I have my doubts about the XMP profile of my ram, I forgot to include this but when I first used the XMP profile for a week thats when I started seeing the restarts with the PC. Being advised by the manufacturer, I MemTest each stick individually overnight over 6 nights (3 runs per stick) both returning with no errors. I noticed with XMP off, the PC has almost next to no crashes/shutdown/restarts but I may be wrong here as I'm very new to this.

I also don't have the funds to purchase a new PSU as of the purchase of the 3080TI, I have been refunded by Amazon in full for the ram but they're expecting it to be shipped back to them until I find a new kit.

 

[Kenpachi2k]

Have you run memtest with xmp on to make sure you are not getting errors while the ram is overclocked ? Is the ram on the motherboards QVL to rule out not 100% compatible ram ?

I have ran XMP once on memtest during one of the overnight testing of each stick, no errors. Just checked the motherboard's QVL the only one listed with the same serial is the 16GB (2x8) 3600MHz version but I noticed that the dual rank version of my kit (32GB 2x16 3600MHz) is on the list and the one im currently using is the single rank version. I've looked at trying to get the dual rank but reading reddit and other forums on getting dual rank kits from crucial is like trying to get a 3080 for MSRP :3

Try and test the PC with your old graphics card if you have one, and see if the problem persists.

Tested a full 10 minute cycle with maxed heaven and multicore test on cinebench with my other 3080, no shut downs or errors. All system settings are at default.

 

[Kenpachi2k]

check the gpu temps when under load, the 3080 ti uses the 3080 cooler and does run warm when under load (in the high 70's into the low 80's) so its very important to have excellent airflow through your case, avoid overclocking on that cooler as at stock the 350w board power is pushing the max the air cooler can manage.

The temps ive seen for 3080 vision under load are just under 70c, 73c at most. The 3080ti fe hits in the high 70s usually stays around 77c mark but sometimes going into low 80s. I don't plan on overclocking the 3080TI.

One final thing to check is that you are using two separate 8-pin cables from the PSU, rather than a single cable with dual connectors.

I have confirmed I am using two separate 8-pin cables from the psu. As my PSU has 4x 8-pin inputs, would you say to try seat them with a space inbetween as they are both next to each other (e.g. 2 & 4 or 1 & 3 on the inputs) or would that not make any difference?

 

[Kenpachi2k]

have you checked ram temps under load if the core is getting into the low 80's, the ram will be running very close to its max temp, hows airflow in your case?

gpu-z will be your best bet to check, how are cpu temps too, when i overclocked my cpu i could see low 90's to its max of 105 degrees and then my pc would shutdown, so thats something to check too, just saying the extra heat my be pushing the cpu too far hense shutdowns, but as others have pointed out, the psu could be at fault also

Just done testing, all system settings are at default. Heaven 3D is set to extreme settings and Cinebench CPU multicore. Testing done for a 15 minute cycle and the results:

3080 VISION
GPU temps: 73c - gpu-z & hwinfo64
Ram temps: 48.5c (both dimms) - hwinfo64
CPU temps: 75c - hwinfo64 & ryzen master

3080TI FE
GPU temps: 79c - gpu-z & hwinfo64
Ram temps: 48c - 50c - hwinfo64
CPU temps: 76c hwinfo64 & 74.5c ryzen master

No shutdowns during the tests.

 

[Kenpachi2k]

@wookiee87 the 3080TI hit a max of 96c on th memory junction.

 

[Kenpachi2k]

@stooeh Nope I don't have any other games/programs more demanding than heaven. I currently run at 2K atm.

 

[Kenpachi2k]

Just tried out playing a game, streaming, cinembench and heaven all at the same time and i've just shut down.

 

[Kenpachi2k]

I do have a spare Tier B, Gold Aorus 750w PSU that you could borrow to test (collection only).
Also check yours on there to see how it fairs.

No worries, a friend might possibly have a 1200w for me to borrow to test.

On the list my PSU is on A tier under single rail and is 11th down but markings show that the 850w is not recommended.

 

[Kenpachi2k]

Assuming you have a GA unit as stated in the OP, it's actually in Tier B under the 'low priority units' section.

Ahh is see, now I know aswell as the PSU tier list. I got given a tier list from someone on the OCUK discord when I was choosing parts and was linked the one I am currently using who said it is good, but also may well be a PSU fault.

I think it's liokely established already but is it simply shutting down or going into a reboot cycle? Shut down would indicate it's doing something to protect components but then that could be linked to the CPU also. The PSU should be the easiest thing to rule out and if you have access to another great or if you must, purchase from a retailer where you know you can easily return and claim it's incompatible should that not be the issue...

I have stated that with the 3080 vision it reboots like once in a while and with the 3080ti it completely shuts down.

I can't actually purchase a new PSU to test and esily return as I don't own a credit card nor enough funds to buy one of more power or equivalent wattage. For me to get a new one would be in the next 2 months.

 

[Kenpachi2k]

UPDATE:
Sending my current PSU for RMA and get a replacement 48hrs-72hrs from them recievinig my PSU.
I have managed to scrap some money up to get a decent PSU to use in the meantime > Seasonic Focus GX 850 80+ GOLD PSU (LTT Tier list - A Tier single rail) which I can return within 14 days of purchase.
Although I haven't tried testing out removing connected drives from the power, which I'll try before I send out my PSU for RMA.
I have also read on many reddit threads that my specific PSU (EVGA GA supernova 850w) has this same exact issues with others using it to power their systems with 3080s and 3090s - assuming its a fault by design across the board?

 

[Kenpachi2k]

After testing with the new PSU running heaven maxed, multicore cinebench, broadcasting a stream and a game all at the same time finishing a full 10 minute cycle. The system didn't shutdown once where as it would shutdown 1-2 minutes after launching a game whilst running this test setup that I have done.

Issue stemmed from EVGA Supernova GA 850w PSU. Many thanks all!

 

[Kenpachi2k]

Great to hear you identified the root cause and now have a stable system.
I have the EVGA Supernova G2 850W PSU. Who'd have thought a single letter make all the difference.

Reading other threads on EVGA Supernova PSUs, GA is apparently the newer production line so i'd assume there is a change in materials used to make it cheaper is my guess.
the G2 ive seen is solid as a rock where as it's sucessor the G3 (the one I had pre-ordered but changed by the supplier due to very late shipment delivery) is not as great - see a trend here.

Which model did you replace it with? Still got RMA open with EVGA?

I replaced it with a Seasonic Focus GX 850W gold. I still have yet to send the EVGA GA for RMA, which will be dropped off tomorrow.