Need to confirm reason why games force my GPU to restart
A couple weeks ago I asked which GPU, between RX 7900 XTX and RX 9070 XT, I should go with. I was gifted the 9070 by a family friend recently so that's done. However my PSU, while I plan to upgrade soon, is not 850W. It's 750W. Which is obviously below the recommendation. So really I'm just asking to confirm whether this is the problem or not. The error I get when, for example, Monster Hunter Wilds crashes, says "Fatal D3D error (24, DXGI_ERROR_DEVICE_REMOVED)".
< >
Showing 1-10 of 10 comments
Could be caused by power spike. 750 is the minimum, the safe recommendation is 850. Your PSU just need more headroom in case it happens
Originally posted by ArielNugie:
Could be caused by power spike. 750 is the minimum, the safe recommendation is 850. Your PSU just need more headroom in case it happens
That's what I'm thinking, I'm definitely leaning towards power spikes because it happens VERY abruptly. Sometimes it doesn't happen at all but that's less than half the time.
Last edited by xXconnor2006Xx; 9 hours ago
most likely psu

psu brand/model/age?
Originally posted by _I_:
most likely psu

psu brand/model/age?
I haven't changed my PSU since February 2022. Corsair CX750F.
Originally posted by xXconnor2006Xx:
Originally posted by _I_:
most likely psu

psu brand/model/age?
I haven't changed my PSU since February 2022. Corsair CX750F.

Some watts may have gone missing since then due to age.

Try underclocking/volting.
Does just the game crash or whole PC? If just the game crashes does it recover to Windows ok?
Last edited by wing0zero; 5 hours ago
The error suggests that DirectX is crashing for whatever reason. Possible driver problem casing that.
Using DDU in safe mode would be my first recommendation but in some cases Windows reinstallation is needed. Windows can randomly damage GPU drivers with updates where even DDU can’t help.

I would probably try DDU first.
Then I would try to reconnect GPU and avoiding daisy chain connection to the PSU if possible. (750w psu may not have enough individual connectors)

Then I would reinstall windows.
And after exhausting other options I would consider getting a different PSU. 1200W for a peace of mind and 0rpm mode at 50% utilisation.
Last edited by C1REX; 5 hours ago
Set up a FPS cap /V-Sync and try. Update Windows and the drivers. Check the cables and reseat the card .
Last edited by smallcat; 4 hours ago
the psu is 5 years old
should not be bad, but may not like newer gpus

gpu is more sensitive to 12v power since it only runs on 12v rail, not needing 5v/3.3 and other minor rails for bios or controls

amd recommends 850w to cover for older and overrated psus

are you using the chained pci-e connectors by chance?
use one cable per pci-e power plug
I had the same issue, but with the RTX 50XX series from Nvidia. The culprit was not the power supply but the motherboard BIOS, which locked the card to the PCIe 1.1/2.0. This caused the system to restart. Updating the BIOS finally allowed the card to run at PCIe 5.0 and the problem disappeared for good.

Download OCCT and run a stress test along with a separate power supply test. If the system restarts, then you know the power supply is the culprit.
< >
Showing 1-10 of 10 comments
Per page: 1530 50