RTX 3090 Ti issue

[Rosco]

Have a client who is a big gamer. He has a cyberpower gaming pc. I upgraded the graphics card last summer to an rtx 3090 ti right before the new one was released. The client called the other day and said the screen blacked out after about 11 hours of continuous gaming at graphics maxed out. They could get it back online after holding the power button and restarting it. They said it had had an error message of something like “video card not recognized” the first time, but after letting it sit for a few hours were able to boot it up. When I picked up the pc, the client showed me how it still worked. He booted up the pc and played Forspoken at max graphics. Nothing active, just moving around. It was running at 200fps and did look incredible. Since it had been unstable, he still wanted me to go over it.

When I got it back to the bench, I checked the ram and hard drives. I also went through the log files, which had nothing helpful. I also reseated the ram and video card. I made sure the drivers were up to date. Resting temp were all good. I give it a test using FurMark at 75 percent power of the graphics card. It ran for about 30 minutes and cut out as the client had described. I had been monitoring the CPU temp, which was hovering at about 85C to 87C at the time of the black screen. Given all this, it’s got to be the video card. Anything I am missing? Also, any way to prevent this from happening again?

Hardware is

i9-10850K CPU 3.60GHz

Asus PRIME Z490-V

32GB Ram

Water cooled

GeForce RTX 3090 Ti

Running windows 10 home

 

[GTP]

Imho its a heat issue where the card overheats after a certain time.
To test, can you aim a extra fan at the back of the card? Or setup a pedestal fan aimed at the card?

I had a similar issue a few weeks ago with an AMD RX 6700 XT.
I added another 300 mm case fan that was pulling heat away from the back of the card (as opposed to how the cards own fans pull cooler air in and over the heat spreaders.)
Its been stable so far...

Might be worth investing in a GPU water cooler kit for it.

 

[frase]

What is the PSU you have installed?

 

[Rosco]

What is the PSU you have installed?

I will open up the back panel and check. Not sure whats

What is the PSU you have installed?

Its a 1000w psu

 

[Rosco]

Update:

I turned it on this afternoon to do some more tests. The video card will not work only onboard graphics. So I am ruling it the video card.

 

[Markverhyden]

I agree it's most likely the video card. Did you try a different slot? Do you have an IR temp sensor? It would be interesting to what kind of temps it gives over time.

 

[nlinecomputers]

Sounds like it is already too late. If it did overheat then it already got cooked.

 

[britechguy]

If it did overheat then it already got cooked.

Doesn't that depend on whether it intentionally throttles and shuts down, just like most processors do, before the "point of no return" is reached? I thought that some of these high-end cards were equipped with their own independent temperature monitoring to prevent reaching the point of no return.

 

[nlinecomputers]

Doesn't that depend on whether it intentionally throttles and shuts down, just like most processors do, before the "point of no return" is reached? I thought that some of these high-end cards were equipped with their own independent temperature monitoring to prevent reaching the point of no return.

Yes but they also have overrides so you can overclock.

 

[Blues]

Also on some of these cards they lack sensors on some parts or even if they have the sensors the thermal safety don't exist or aren't sufficient for those parts.

I would as a final test let it sit for some time then reboot with some added active cooling on the card and if you can check temps with an IR temp sensor.