PC Graphics

[Ragnarkar]

Hello AMD community!

I'm looking for some assistance with a transient power issue I have been experiencing with my system. Below are the details of my setup and the problem I'm encountering:

System Specifications:

• CPU: Ryzen 7 7700X
• Motherboard: Gigabyte B650 AORUS ELITE AX
• RAM: Kingston Fury 32GB (4800MHz)
• PSU: Corsair RM1000X
• GPU: AMD Radeon RX 7900 XTX
• Operating System: Windows 11
• Drivers: 24.10.24 (is installed at the moment for testing the issue)

Problem

The system experiences what appears to be transient power issues, resulting in crashes during gameplay. These crashes are accompanied by an unexpected shutdown , and there are no specific error messages displayed on-screen. The system does not appear to overheat or overload the cpu during these events.

Background:

I built this system around December 2022 and have not experienced any issues until about three weeks ago. The problem only occurs while gaming, most frequently in Delta Force, but it has also happened in other games such as Rocksmith, Black Desert Online, Hell Let Loose, and Rainbow Six Siege.

When it does happen it can either be twice in a row, or hours between shutdowns, its random. 

Troubleshooting Steps Taken So Far:

1. Temperature Monitoring: I have monitored the temperatures of the CPU, GPU, motherboard, and case. All components are operating within normal temperature ranges.
2. Stress Testing:
   • CPU Overload Test: No issues detected, and no power spikes were observed.
   • FurMark Test: No issues detected, and no power spikes were observed.
   • 3DMark Test: No issues detected, and no power spikes were observed.
3. Undervolting: I attempted to undervolt the GPU by 50mV, but the crashes persisted. Further undervolting led to instability.
4. Driver Reinstallation: I reinstalled older drivers, suspecting the issue might be driver-related, but the problem persists.
5. Hardware Reseating:
   • Reseated the GPU and all connected cables.
   • Reseated all cables connected to the motherboard, including the CPU power cable.
6. Event Viewer Logs: I reviewed the Windows Event Viewer logs, and the only relevant entry is an "unexpected shutdown" error, which does not provide further insight into the root cause of the issue.

Additional Considerations:

• Power Supply Unit (PSU): I do not currently have a spare PSU to test with
• GPU Testing: Similarly, I do not have a spare GPU to test with

Given the troubleshooting steps I've already taken, I am uncertain whether the issue lies and what to do, I do not have a lot of disposable income to buy new part, however, all my components are under warranty. 

[cpurpe91]

Have you ensured the GPU is running on separate cables coming from the PSU? It has become nearly mandatory to use a different cable for each 8 pin or 6 plus 2 pin connection on GPUs since the time around the RX 6000 series release. Many have issues when using multiple connectors on one cable.

Without a new PSU you could try a program called OCCT. It has a test that will stress the PSU. If it is the PSU this test will likely trip the PSU protection and cause a shutdown or reboot. 

I would try another PSU if OCCT trips the PSU protection and shuts down. Understanding that you don't have one can make things difficult. Try getting a different 1000w PSU, or 1200w if you can afford it. In my experience PSUs can be tricky to diagnose. 

If the new PSU fixes the issue, I would RMA the old one as it is still under warranty and keep it for testing in the future.

[cpurpe91]

I misread the disposable income situation. I would use OCCT to test with each GPU cable being individually plugged in at the PSU, and RMA the PSU if it powers off or restarts the PC.

[Ragnarkar]

Hi,

Thank you for your reply. 

It is currently hooked up with the provided 16-pin single cable (it splits, but only 1 connector to the PSU). I will try that first as it is a bit more cost-effective than buying a new PSU and will use OCCT to test the PSU.)

I'd assume these would be suitable: https://www.scan.co.uk/products/45cm-cablemod-modflex-8-pin-pcie-extension-cable-black-modflex-sleev...

There are the Corsair ones, but they are double the price, would they be a better option or would the cable mod one be of similar quality?

I will update once I've received the cables and done the testing.

[johnnyenglish]

Hi, instead of undervolting, power limit the card, like 80% and keep it well under stock clocks, no boosting.

If its a power spike, this should troubleshoot it.

Good Luck

[Ragnarkar]

Hi,

By changing the power limit, do you mean changing the below to -20% (as it only goes to -10%) : 

[johnnyenglish]

Try it either way.

Some other third party GPU apps do go bellow that I believe, like Afterburner or ASUS Tweak. But the intent is to lower the clock and draw much less power.

If it still crashes, it shouldn't be a power issue unless there is something really wrong going with the PSU or GPU.

[Ragnarkar]

Thank you, I will give that a go tonight and test it.

[Ullsokk]

Hello.

Did this fix your issue?

I'm having the same problem except the PC shutting down.

[Ragnarkar]

Hi,

No it hasn't fixed it. 

I'm waiting for my new cables to arrive - ordered 2x 8pin PCIe cables from corsair - they arrive on the 24th. So will see then.