• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Random reboots during gaming after upgrading to RX7900XT

Associate
Joined
5 Sep 2012
Posts
125
I've recently replaced my RTX3080 10GB with a RX7900XT Reference.

I ran DDU removed all drivers and shut down before swapping the cards for a clean install and installed the latest version of adrenalin.

I'm able to run Heaven benchmark's without issue but experiencing sudden shutdown/reboots during gaming I never experienced before? This is across multiple titles.

The 7900 is 2nd hand but temps seem okay 70 GPU temp and 90 hotspot.

I have a 850W EVGA G3 power supply.

I'm going to run memtest just to rule out any random RAM errors. CPU is a 5900X.

Any other suggestions?
 
What are the coolers like? Any differences in how they're dumping the heat into your case with temps and what not?

I'd agree that it is worth trying to set a power limit and/or underclock the card. Your PSU is a good one, but you never know.
 
I set the power limit to -10 started Atom heart and it crashed even quicker (10 mins maybe). PC boot looped 5-6 times and then finally got back into windows.

Used DDU to clean and restart again and reinstalled Adrenalin 24.8.1 leaving all settings stock including leaving my screen refresh rate at 60hz

Just played an hour of Atomic Heart without any issues so fingers crossed we might be okay now!

My RTX 3080 was a basic twin fan Dell OEM model but it would sit silently at 50% fans and a solid 71c with a slight OC via MSI afterburner. I have a Evolv X case with 3x140 Noctua's in front so plenty of airflow for the GPU. Since reinstalling Adrenalin the difference between GPU and Hotspot seems closer but that could be due to the frames locked at 60hz putting less stress on the card.

Will try again with the FPS set to 144hz (I have a 165hz OLED 1440p) and see what happens? I also have a Kryosheet on order as it doesn't look like the cards been opened to be repasted.
 
I set the power limit to -10 started Atom heart and it crashed even quicker (10 mins maybe). PC boot looped 5-6 times and then finally got back into windows.

Used DDU to clean and restart again and reinstalled Adrenalin 24.8.1 leaving all settings stock including leaving my screen refresh rate at 60hz

Just played an hour of Atomic Heart without any issues so fingers crossed we might be okay now!

My RTX 3080 was a basic twin fan Dell OEM model but it would sit silently at 50% fans and a solid 71c with a slight OC via MSI afterburner. I have a Evolv X case with 3x140 Noctua's in front so plenty of airflow for the GPU. Since reinstalling Adrenalin the difference between GPU and Hotspot seems closer but that could be due to the frames locked at 60hz putting less stress on the card.

Will try again with the FPS set to 144hz (I have a 165hz OLED 1440p) and see what happens? I also have a Kryosheet on order as it doesn't look like the cards been opened to be repasted.
I would want to check all is ok at 144hz before counting your chickens. I imagine at 60hz is not pulling much power.
 
I have a 850W EVGA G3 power supply.

Are you pigtailing the PCIe power cable or using two cables from the PSU to the GPU? You should opt for the latter. And have you checked the power connectors at the PSU end?

I also have a Kryosheet on order as it doesn't look like the cards been opened to be repasted.

It's far too young for repasting to be necessary.
 
To be honest I would not be repasting until I'd exhausted other means of telling if the card is faulty.

My usual troubleshooting method is to reduce everything (core clock, mem clock, power limit) to minimum and then increase it gradually until it becomes unstable. Another way to rule out temperatures/air flow could be to remove the side panel and blast a desk fan at it. Unfortunately you need to get it stable in some form before you can start to figure out what the actual problem is.
 
Posted this in a few other threads, I had the same issue with my XFX Merc 310 7900 XT.

I tried lots of different things under lock/volt, fresh windows etc and could run stress tests like Firmark for 2+ hours, no temp issues or crashes. But gaming was a different story, in the end all I could put it down to was the drivers causing issues.

I could run drivers from around May/June but ones after that including preview ones would cause crashing, black screens, timeouts even from cold boots, I would get the default setting message as well. So even though I could run earlier drivers I’m then missing out on game updates and things like AFMF2, in the end I’ve had to change card.

Sorry to be bearer of bad news but hopefully you can get it sorted.
 
Posted this in a few other threads, I had the same issue with my XFX Merc 310 7900 XT.

I tried lots of different things under lock/volt, fresh windows etc and could run stress tests like Firmark for 2+ hours, no temp issues or crashes. But gaming was a different story, in the end all I could put it down to was the drivers causing issues.

I could run drivers from around May/June but ones after that including preview ones would cause crashing, black screens, timeouts even from cold boots, I would get the default setting message as well. So even though I could run earlier drivers I’m then missing out on game updates and things like AFMF2, in the end I’ve had to change card.

Sorry to be bearer of bad news but hopefully you can get it sorted.

Yes, this is definitely just a stop gap until I can order a RTX5080.
 
The cards approx. 18 months old so long overdue a repasting

I'm using two separate 8 pins from the PSU

Another hour without any issues locked 144 FPS
86% utilisation
308w total board power
67c GPU
93c Hotspot

We seem to be getting somewhere which is good
Seems like it could well just have been the driver swap then. Maybe something didn't get quite cleaned. Hope it keeps up, all temps seem ok as well, although might be nicer if the hotspot stayed at around 90.
 
Last edited:
Which game? There's a long standing bug on World of Warcraft with the Radeon 7000 series that neither AMD nor Blizzard seem inclined to fix.
 
Okay 18 months might be extreme but 2-3 years certainly plus it’s only a 10-15 min job (CPU) and I normally see an improvement in temps.

My Dell RTX3080 had a significant improvement in temps when I repasted the die and added additional thermal paste to the Vapor chamber contact plate.

Other forums and various YouTube vids suggest switching to Kyrosheet on RX7900 GPU’s reduces the difference between the GPU temp and hotspot so why not give it a try?
 
Last edited:
I've had cards for 3 or 4 years that haven't required repasting so I don't know where you're getting that information. .
An OEM is going to install a thermal paste that lasts the length of the cards warranty, keeps the card within spec at a certain price point which is why as you say people can go many years without repasting.

But that doesn't mean there isn't a performance uplift to be gained from replacing the OEM paste with a TIM better thermal properties. It's the same reason people buy a GPU then slap a water block on it.
 
Back
Top Bottom