Graphics Cards

[PetK]

Hi all,

So I'm one of the lucky few without heat issues. Unfortunately, I'm dealing with a different one.

I've seen a few posts with other's also dealing with hard crashes. However, none seem to point out a definite power issue. 

My setup:

• 5950x 
• 32gb Ram at 3600mhz
• Asrock Taichi x570 mobo
• Bequiet 1000w PSU
• 7900xtx reference card

When my hard crashes occur, I need to switch off my PSU and turn it back on again before I can restart my desktop. To me, this indicates a power surge. I was wondering if the design of the reference card is the flaw here? You have 2x 8-pin connectors good for 300W+75W from the PCIE lane. However, at full load the 7900xtx pulls 400W and that's excluding any possible spikes.

Are any of you experiencing it like this?

[Justifier]

nah. thats most likely overheat protection kicking in.

check your Junction Temperatures, bet it's hitting 110c (if it spikes, and hits 118 it will cut power to the system to protect itself)

[PetK]

With excessive stressing my junction temperature hits 91C, so heat is really not an issue. It crashed while loading a WoW screen, so not even GPU demanding.

[MADZyren]

How was your CPU temp?

[Moist]

My hard crash symptoms led me to believe it was a power issue as well but it ended up being a monitor issue of all things. After getting the Alienware 2721D drivers from dell it resolved the crash issues. Not sure if it's other 240hz monitors or because I always turn off hardware drivers from windows 11 because windows is horrible. It's something you might want to check. I will also add when I hard crashed this way I always had to reinstall adrenaline.

[uncyler825]

If you have an AMD Radeon RX 7900 Series graphics product, and the BSOD (error code: DPC_Watchdog_Violation) problem （caused by amdkmdag.sys）occurs for unknown reasons, you can refer to this article:
https://forums.guru3d.com/threads/for-rx-7900-series-driver-with-refresh-rates-above-60-hz-may-cause...

You can try adjust refresh rates to 60Hz and turn off variable refresh rate feature, disabled AMD FressSync from Radeon Settings.

Notice: Connect 1 monitor only.

This ways maybe can solve your problem.

[PetK]

You could be onto something. However, Samsung is currently not rolling out any drivers for this. The crash happened when WoW was loading a map, so neither GPU nor CPU intensive. But, I still have the feeling the system spiked somehow causing my PSU to say ''Nope''.

[uncyler825]

If you have an AMD Radeon RX 7900 Series graphics product, and the BSOD (error code: DPC_Watchdog_Violation) problem （caused by amdkmdag.sys）occurs for unknown reasons, you can refer to this article:
https://forums.guru3d.com/threads/for-rx-7900-series-driver-with-refresh-rates-above-60-hz-may-cause...

You can try adjust refresh rates to 60Hz and turn off variable refresh rate feature, disabled AMD FressSync from Radeon Settings.

Notice: Connect 1 monitor only.

This ways maybe can solve your problem.

[PetK]

Thanks for the advice, but I don't get a BSOD nor any error code. It's my PSU saying no, at a given moment. Not even during stress tests or so. When I swap back my 6900xt I do not have this issue. I there a way to measure/register power spikes in your system? Considering I don't have any logs showing me info why my desktop shuts down.

[PetK]

So, I managed to recreate the problem while logging with HWInfo. I simply start up World of Warcraft and enter a server. When it's almost done loading my pc hard crashes. So, no power. Forced to switch the PSU off and on again in order to restart. Now, looking at the logs I don't see any temperature spikes in the GPU nor any other abnormalities. When I look at the CPU I see an almost maxed out usage, but still no temperature or power issues. No power spikes in the system either. Usually, when the CPU is too overclocked or so the system will reboot and event viewer will show which core is unstable. But, this is not the case. And this problem didn't occur with my 6900XT and does with the 7900XTX. So, the question remains... What is causing my PSU to say ''nope''. Cause if the PSU were faulty shouldn't it occur with the 6900XT also?

[ThreeDee]

Some GPU's just don't play nice with some PSU's .. with both Nvidia and AMD. Seems like your PSU's OCP is getting tripped

I'd try a different power supply ... or at least run separate power cables from your PSU to each power input on your GPU if you are not already .. 

[MADZyren]

With previous machine, I did a lot of memory oc experimenting and if I went too far, I had to switch PSU off to get it to boot. Sometimes clear bios too.

For testing, I would lower memory speed to 3200 to see if problem persists.

There is a chance power usage on CPU spikes so fast, your PSU can't keep up or maybe your PSU->GPU power cable is too thin or some of the copper wires broken (sorry about limited English - basically why you also would not get strong, precise bass from speakers, if speaker cable is too thin - cable starts to behave a bit like resistor). So make sure you are using two cables from PSU to GPU and not just one that splits into two at GPU's end.

If you have PBO enabled, would disable it. 

[PetK · Accepted solution]

In the end it wasn't PSU related, but I suppose heat related. In September last year I completely watercooled my system, GPU block, CPU block etc. Never had a hard crash since. Now, I'm only left with the software issues between Windows and AMD... ... :')