Question Help - AMD Ryzen 7 9800x3d & GTX 5090 pc constantly shutting off

[goot66]

Have a brand new PC with a AMD Ryzen 7 9800x3d and GTX 5090 card and 1500W PSU

When gaming after about 10 - 20 mins, the computer randomly just suddenly shuts off

as if someone just disconnected power to the computer

I don't get any error logs or anything in Event Viewer

Is this a faulty PSU ? Faulty GPU?

GPU and CPU temps are fine

I have very good cooling in the PC

 

[mikeymikec]

If it's an ASRock board, I understand they've had some trouble with the 9800X3D that I think the latest BIOS update should fix. It might be worth reading up about it first though. The trouble involved some scorching of the CPU's contacts.

In any case, checking for a BIOS update for such a new CPU might not be a bad idea, but by the sounds of things you should give the PC a rest to ensure that it can get through the update without shutting down.

Does the PC keep running if you're not gaming?

 

[goot66]

it happens when gaming at max graphics and max resolution and only on certain games - e.g. happens on Motogp25, Kingdom Come Deliverance II, Avatar

other games running fine

when it happens, its instantaneous and pc suddenly shuts off and loses power

 

[goot66]

specs below :
CPU: AMD Ryzen 7 9800X3D (8-Core) 5.2 GHz Turbo (3D V-Cache) (Zen 5)
CPU cooler:
Motherboard: MSI MPG X870E CARBON (Wi-Fi) (AMD X870) - (Up to 3x PCI-E Devices) (DDR5) - Latest BIOS
Ram: 96GB DDR5 6400MT/s Kingston FURY RGB
SSD/HDD: 1x SSD M.2 (2TB Crucial T700 PRO) (Gen5 NVMe)
GPU: GeForce RTX 5090 32GB
PSU: 1500W BeQuiet Straight Power (Modular) (80 Plus Platinum)
Chassis:
OS: Win 11 Pro
Monitor: Samsung G9 Odyssey

 

[Hail The Brain Slug]

When it happens, are you able to just press the power button to turn it back on?

Or do you have to switch the PSU off and back on/unplug and replug the power cable?

 

[goot66]

Yes i can press power button on and computer turns back on fine

 

[Hail The Brain Slug]

Are there any errors reported in Windows event viewer?

 

[igor_kavinski]

Check and make sure the power connectors for the 5090 are properly connected all the way in.

Are you using the cable that came with the 5090 or the one with the PSU? If it's the one that came with the 5090, personally I wouldn't use it. The PSU maker knows and cares more about their PSU so they would include a decent cable with it. The constant shutting off seems like a preventive measure as something bad is being detected by the PSU.

 

[igor_kavinski]

CPU cooler:

That's important info you left off.

If it's an AIO, did you take off the plastic sheet on the block before fitting it over the CPU? Are the various cables connected to the proper motherboard headers?

If it's an air cooler, did you tighten the screws properly?

 

[igor_kavinski]

it happens when gaming at max graphics and max resolution

Points to a GPU power related issue. So only happening when the GPU is stressed hard. Check the cables!

Also, you said it's Geforce 5090. Is it the founder edition or some other AIB card? Gigabyte cards have leaky thermal putty which may cause issues.

 

[DAPUNISHER]

Pretty iffy ram config. Also, if you are using frame generation or DLSS they can cause the PC to shutdown. Nvidia drivers are the worst I think I've ever seen them.

 

[igor_kavinski]

Also, if you are using frame generation or DLSS they can cause the PC to shutdown.

That's crazy! Isn't that supposed to be their whole selling point???

Has any of the popular outlets shown this happening on video?

 

[DAPUNISHER]

That's crazy! Isn't that supposed to be their whole selling point???

Has any of the popular outlets shown this happening on video?

Steve mentions it in his rant about Nvidia's dirtbag tricks. Time stamped -

 

[gdansk]

Over Current Protection?
Years ago I had a 1000W Titantium SeaSonic power supply which would shut down much like this (with a measly "320W" RTX 3080).

 

[Hail The Brain Slug]

Over Current Protection?
Years ago I had a 1000W Titantium SeaSonic power supply which would shut down much like this (with a measly "320W" RTX 3080).

Your Seasonic wasn't hitting OCP. Seasonic has a dedicated 12v sense line to the motherboard on the ATX 24 pin to adjust 12V output voltage to maintain super tight regulation.

Ampere was such an electrical disaster it would induce off the charts amounts of electrical noise into the 12v line of the motherboard that the PSU would believe something was catastrophically wrong and power off.

 

[gdansk]

Your Seasonic wasn't hitting OCP. Seasonic has a dedicated 12v sense line to the motherboard on the ATX 24 pin to adjust 12V output voltage to maintain super tight regulation.

Ampere was such an electrical disaster it would induce off the charts amounts of electrical noise into the 12v line of the motherboard that the PSU would believe something was catastrophically wrong and power off.

So power supply is not a likely suspect in this case?

The description, however, reminded me of the power supply shutting off (for good reason or not).

 

[Hail The Brain Slug]

So power supply is not a likely suspect in this case?

The description, however, reminded me of the power supply shutting off (for good reason or not).

The power supply could be related in this case, but I don't think I've seen any evidence the 5090 has high enough power spikes to trigger protections on a modern 1500 watt power supply. Maybe if the PSU is defective.

The probable causes I can think of off the top of my head is:

• 5090 drivers
• bad ram config
• defective power supply
• some other software defect
• some other hardware defect

If we could get some details from OP on if there's anything in the event viewer that would point toward a WHEA error, driver error, something else, or nothing at all would be useful.

If there's no smoking gun in the event error logs, I'd reset the bios/cmos and try to reproduce it with a completely unaltered, stock bios config to rule out EXPO/XMP/RAM, PBO/CO, etc. causing system instability.

 

[burninatortech4]

I would try process of elimination. Try each step below and see if you can recreate the crash. Change only one thing at a time so you can attempt to isolate it.

1) Turn off XMP (run RAM at its default JEDEC profile).
2) Run Memtest
3) Remove any overclock or undervolt
4) If you have 4 sticks, remove 2 and run 2 sticks (make sure you follow MOBO manual for correct DIMM topology for 4, 2, and 1 stick configurations)
5) If you have 2 sticks, remove 1 and run 1 stick
6) Chipset driver reinstall
7) Windows reinstall
8) Use a different dGPU (or the built in 610M)

If none of those narrow it down I'd suspect your PSU.

 

[maxemery]

I couldn't see anyone recommending trying to eliminate the variables of the CPU GPU by running separate stress tests.
Furmark would be ideal as it barely touches the CPU then maybe prime 95 for the CPU. Hopefully this would replicate the issue.
Just be careful with furmark as people like the point out that it "unnaturally" stresses the GPU and can lead to higher than normal heat output.

 

[DaaQ]

I would also look at the gen 5 ssd temps. Plus 96GB of DDR RAM at 6400, might be worth investigating.