Crashes, Crashes, Crashes, and weirdness

[Velikanka]

OS : Windows 11 Home Edition
MB : ASUS TUF Gaming B550 + WiFi
CPU : Ryzen 5 3600, then Ryzen 7 5800X
AIO MSI Watercooling
SSD : Samsung Pro 980 M2
RAM : Corsair Vebgeabce Pro RGB (2x8)
GPU : ZOTAC 3070 Ti Amp Holo
PSU : 650W Gold 80+
VR : Oculus Quest 2

Hi all,

After 2 years of flawless service, my computer started to crash in different ways while gaming, mostly in VR.
Some games, like Assetto Corsa Competizione, would crash in under half a second after launching a session.
Other, like Automobilista 2, would hold for 10/15 minutes before crashing.
A few “lighter” games, such as Rez Infinte, would not crash.
I’ve also had crashes in non VR games (Two Point Campus and others.)
It turns out that one of the cores of my Ryzen 5 3600 was severely overheating under stress (96 degress while the other 6 wouldn’t reach 40.)
A that point in time, it seemed like it was a faulty temp sensor that was the culprit, showing a temp over 100 degress hence inducing shutdwon. How could one core overheat so dramatically and not affect its neighbors?

But…

As the Ryzen 5 was being replaced by the Ryzen 7, it turned out that the cooling paste had turned into a chewing gum like substance, indicating that there was overhating indeed.

Initially, the new CPU changed nothing. Crashes would appear at the same rate, in spite of the temperature of every core being nominal.

The real weirdness starts here : after a clean install of W11 (GPU drivers + Steam installed + Oculus Software), the first day wasn’t so good. More crashes.

But on the second day, everything was flawless. No crashes in spite of several hours of VR gaming.

On the third day… It’s the worst it’s ever been. For instance, I can’t even get past the menus in Assetto Corsa Competizione. A flahing blue screen before the main menu, and reboot.

I’ve done everything I could, all the way up to updating the BIOS. No dice.

I previously thought my GPU was at fault. The repair guy found that overhating core, which I never diagnosed, as I never thought such an issue was possible and only checked CPU package temp.

But since crashes now happen outside of gaming, I’m thinking it’s anything but.

Sorry for the long post, but the weirdness of the situation has engulfed me.

Thoughts?

Cheers.

[veeg]

Hello

Download and post.

https://www.ccleaner.com/speccy/download

To post.

To publish a Speccy profile to the Web:

In Speccy, click File, and then click Publish Snapshot.

In the Publish Snapshot dialog box, click Yes to enable Speccy to proceed.

Speccy publishes the profile and displays a second Publish Snapshot. You can open the URL in your default browser, copy it to the clipboard, or close the dialog box.

[Velikanka]

https://speccy.piriform.com/results/ffzZVkdGWtdddZRglrnJ0xZ

[Rustys]

GPU : ZOTAC 3070 Ti Amp Holo
PSU : 650W Gold 80+

The GPU requires a 750W PSU

https://www.zotac.com/us/product/graphics_card/zotac-gaming-geforce-rtx-3070-ti-amp-holo#spec

https://support.lumion.com/hc/en-us/articles/360003476773-Which-power-supply-do-you-need-

@phillpower2

Disable XMP

Set your Power Profile to Balanced

[Velikanka]

Thanks for the replies. I shall disable XMP later and let you know how it turnedout.
I would however dispute the idea that I need a 750W PSU. Maybe it’s the case now with a more powerful CPU (65W->105W), but with the Ryzen 5, my multimeter plug was telling me I would never get past around 500W. Makes sense as my 3070 TI pulls 315W at full blast, add 65W for the CPU.
Add whatever the rest needs, and I believe I end up well within the range of my PSU.

Am I missing something?

[Velikanka]

Balanced mode enabled, XMP disabled. No Change.
What’s more, my mechanical hard drive is no longer detected. That has never happened before.
I am at a complete loss.

[Rustys]

I would however dispute the idea that I need a 750W PSU.

[ATTACH type=“full”]13054[/ATTACH]

https://www.zotac.com/us/product/graphics_card/zotac-gaming-geforce-rtx-3070-ti-amp-holo#spec

Not seeing an HDD (mechanical) all showing are SSDs.

Which one are you missing?

[ATTACH type=“full”]13055[/ATTACH]

since the HDD is missing if you were to completely disconnect power and data form it and boot the system does that make any difference?

[Velikanka]

I’ll take my PC to the repair shop on tuesday with a new MoBo.
I’ve sort of given up on trying to fix anything at this point.
Would you all be interested in hearing how it turns out? MoBo replacement was the tech’s idea.

[phillpower2]

I would however dispute the idea that I need a 750W PSU. Maybe it’s the case now with a more powerful CPU (65W->105W), but with the Ryzen 5, my multimeter plug was telling me I would never get past around 500W. Makes sense as my 3070 TI pulls 315W at full blast, add 65W for the CPU.
Add whatever the rest needs, and I believe I end up well within the range of my PSU.

Am I missing something?

Something to keep in mind, it is the GPU manufacturer that is expected to honour any warranty should the card fail because it was underpowered, not a PSU calculator or some faceless person on the internet that says that it is ok to ignore any minimum PSU requirements specified by the GPU manufacturer.

GPU manufacturer are not legally obliged to and nor will they replace a product or pay a refund for damage caused by the end user.

PSU : 650W Gold 80+

Not including the brand and model name or number is not normally a good sign + in 100% in agreement that the PSU regardless of the brand and quality allowing for the additional storage devices and the extra screen this PC needs a minimum of an 850W Gold or above efficiency rated PSU from one of the following, Corsair RM range, EVGA, Seasonic or Super Flower, seven year warranty minimum but ten years ideally.

Balanced mode enabled, XMP disabled. No Change.

Balanced mode is good but having to disable XMP is bad, Ryzen CPUs are notorious for not liking slow RAM and purchasing the maximum speed of 3200MT/s RAM so that you could have enabled XMP would have been more appropriate.

I’ll take my PC to the repair shop on tuesday with a new MoBo.

Would you all be interested in hearing how it turns out? MoBo replacement was the tech’s idea.

There is nothing to suggest that a new MB is required but there is an awful lot of evidence that an appropriate PSU is required and the RAM needs to be manually OCd to 3200MT/s.

The worst part about you purchasing a new MB is that it will not change the weak PSU and incorrect RAM speed issue/s.

[Velikanka]

Thank you very much for that, I can return my MoBo if it turns out it’s not the solution.
But regarding the PSU, 2 quick things :

• I’ve tried with one screen unplugged, no effect.
• I downgraded my GPU to get 70% power max (capping at 210W instead of 315W), no effect.

As for the RAM, I’m surprised to find out the speed was incorrect. Thanks for pointing it out, I’ll set it at a fixed 3.2GHz.

[phillpower2]

The load of the additional drives and second screen on an appropriate quality brand and output 750W PSU would be too much so dropping down to just the one screen will do zero with such a weak PSU I`m afraid.

Same as above regarding the GPU but for a GPU it is worse because being the most power hungry device in a computer the video card tends to suffer first and the most.

You are welcome and good luck.

[Velikanka]

Quick upate : after re-enabling XMP, my computer is again able to go trhough benchmarks (3DMark stress test passed without issue with very nice temps.)
I plugged it in a separate outlet, one screen on. My multimeter plug tells me it caps at 475W.

Does it mean my PSU is limited to this value? (650-12% equals 472W). GPU runs at 100% @ 73 degrees (Throttling temp is factory set at 83.)

CPu runs at 71-74, but the load is low (20%). I yet have to find a benchmark that stress both at 100%.
Any tips?

[Rustys]

Thank you @phillpower2 for the explanation and clarification.

since the HDD is missing if you were to completely disconnect power and data form it and boot the system does that make any difference?

Did you ever attempt this or did the drive decide that it want to be there now?
Some times a fail to failed drive can show these issues.

[phillpower2]

I plugged it in a separate outlet, one screen on. My multimeter plug tells me it caps at 475W.

Multimeters and cheap PSU testers are not capable of properly testing a PSU and will only give you the most basic of information.

The attached videos are the ones that I use as an example to show the sort of equipment that is required to conclusively test a PSU + what can happen when testing cheap and nasty PSUs

What the equipment does is simulate a computer under load and so the PSU behaves as it would when it is powering your PC, multi-meters and the basic testing devices that you can purchase from some merchants only test the basic voltage output from the PSU rails and so cannot be relied on as 100% conclusive.

Cheap PSUs exploding when being properly tested

Jonny Guru video

In addition to the above asked by Russ, what is the brand and model name or number of the present PSU and how old is it.

[Velikanka]

Thank you so much again.
Very interesting to know that a multimeter is no reliable way to test a PSU, I would never have guessed.
I would however like to point out that my PSU is not a cheap no-name one. I simply do not remember the brand, but I was careful to buy a proper one. I believe it’s a Corsair. (Gold cerifaction).

But you all have convienced me that I need a new one. After all, I had no issues before adding another screen (but the problems didn’t start right away, it took a couple of weeks.)

I’m currently browsing to check whch one to get. I’ve settled on 850W.

I’ll let you know how it turns out once it’s installed.

P.S : the missing hard drive was simply due to me screwing something up in the BIOS after upgrading it. It instructed me to do something regard’ing RAID which I did not quite understand. But it’s back on and works fine.

P.P.S : just finished a 2 hour gaming session in VR with nothing going wrong at all. One screen off, nothing more.