Strix 3080 OC crashes on google Chrome?

[Vovin]

So I just got the Strix the other day and it's not working as well as I expected. It is doing some weird crashes while just doing stuff on Chrome, like scrolling through Youtube. It freezes for a few seconds, then HDMI connection cuts out from TV. PC itself is still running, just no visual or audio connection to GPU. I have to reboot the PC from the power button then it starts up normally. Event logs don't have any obvious explanation, just stuff about dump file creation, WindowsApp Model not loading, etc.. Not too sure if any of that is related.

This crash strongly resembles a driver error that I encountered years ago where a new GPU driver was causing the same weird crash during Netflix/Youtube videos. I remember having to roll back the driver at the time. I did that this time: rolled back to studio driver, which is supposed to be stable, but still doing it. It has crashed three times since yesterday, ALWAYS while viewing Chrome browser. I heard some rumours about 3080s crashing on Chrome browser, but I can't find any more information about it. This is what I would like to know more about, because Chrome is at newest version, GPU should be stable driver, not sure what else to do.

This is the Strix 3080 OC version and my CPU is only i7-8700, but apparently that is not supposed to malfunction beyond mere bottlenecking (until I get a better CPU down the line). But I'm not sure if this is true. Keep in mind that this is just happening in Chrome when I'm not really doing much, and I've been too busy to try this out on a game (though I tried it for RE2R briefly just to max graphics, and it did just fine with no stress).

PSU is 850w. I'm probably going to order a better one soon, but this one is doing ok for now. It could potentially be a random power spike in the GPU but I doubt it is a power supply issue- the people who suggested this to me also claim that the Strix 3080 has 500w power spikes, which doesn't make much sense considering that it has 3 PCIE slots at 150w each. Also I was reminded of what actual PSU failure looks like when the Strix PSU that came combo'd with the GPU was tweaking out and nearly caught fire when I turned it on for the first time.

Any ideas?

[[☥] - CJ -]

display driver crashing

Have you tried DDU in safe mode?

What GPU did you previously have and did you uninstall that driver beforehand?

It may have 3 PCIE's power but dont forget the PCI-E slot itself provides 75w as well

[Vovin]

Originally posted by ☥ - CJ -:

display driver crashing

Have you tried DDU in safe mode?

What GPU did you previously have and did you uninstall that driver beforehand?

It may have 3 PCIE's power but dont forget the PCI-E slot itself provides 75w as well

I used to have a GTX 1060 in there. I just updated to the newest Game Ready driver like 3 days ago, then got new GPU in mail the next day. When I installed new GPU, I performed a clean install of same new GR driver, then eventually did another clean installed of Studio Driver (apparently studio driver is just an older but stable GR driver).

I strongly feel like it is a driver issue as well. The other funny thing is that Rainmeter isn't picking up the 3080 at all, while that used to be my main way to observe 1060's haptics. It also doesn't send any signal from GPU's HDMI 2.1 output to TV's HDMI 2.1 input like with 1060, though this is most certainly an issue with the uncertified cable. So just regular HDMI for now.

I have not tried DDU in safe mode. Should look into it.

[[☥] - CJ -]

As i said, use DDU in safe mode.

For the record, there is no "clean install" without using DDU

The clean install option when installing the driver only resets the NVIDIA CP settings, it doesnt actually do a clean install.

[Vovin]

Originally posted by ☥ - CJ -:

As i said, use DDU in safe mode.

For the record, there is no "clean install" without using DDU

The clean install option when installing the driver only resets the NVIDIA CP settings, it doesnt actually do a clean install.

Actually that makes sense. Furthermore, something else I noticed is that while Nvidia settings display GPU as 3080 with 10 VRAM, Intel Extreme Tuning Utility says it is 3080 with 4 VRAM. The GTX 1060 had 4 VRAM. Would you say that's definitely a conflicting driver issue?

[[☥] - CJ -]

Naw, its a software issue not reading it correctly, likely hasnt been updated in a while.
Has nothing to do with the driver.

[Vovin]

Originally posted by ☥ - CJ -:

Naw, its a software issue not reading it correctly, likely hasnt been updated in a while.
Has nothing to do with the driver.

Correct. I had the same thing with HWinFO, so I updated it and it was then able to read the GPU correctly.

So I did the DDU safemode thing and it has appeared to work. There was one moment where it did freeze for like two seconds as before, but didn't crash. Again, no event log.

So now I suspect that this was a combination of a poor PSU/daisy-chaining (2 PCIE cables only with PSU) where the GPU was spiking, which bumped off graphical connection somehow due to conflicting drivers. I ordered a Corsair RM1000i, which was only like a $100CAD more than other PSU, so it's a worthwhile investment. Said screw it and ordered a new case too, pretty much constructing a whole new PC at this point lol.

I did a lot of looking around on daisy-chaining before/during this installation, and official sources said that having two PCIE cables (where one splits into two connections) is fine. However, I suspect that some PSU models are cheaply made and not really up to snuff. It's also strange how most PSUs seem to come with only two PCIE cables, including the Strix PSU that came combo'd with my GPU (of which this PSU was dangerously defective).

So for anyone getting a new GPU, keep in mind that you NEED a quality PSU with 3 PCIE cables. You can't mix and match the cables from different PSUs either.

[GOD RAYS ON ULTRA™]

Try debug mode in nvidia cp. I'm not at all surprised that an overclock is crashing with basic inputs. If you enable debug mode, your gpu will be returned to nvidia reference values. If debug mode works then for sure the overclock is unstable. At that point you can choose to RMA the card because it was basically ruined from the factory.

[Bad 💀 Motha]

Should just disable Hardware Acceleration in browsers like Chrome; otherwise just using this browser will force the GPU into 3D Clock Mode.

[[☥] - CJ -]

Originally posted by Bad 💀 Motha:

Should just disable Hardware Acceleration in browsers like Chrome; otherwise just using this browser will force the GPU into 3D Clock Mode.

If Power Performance mode is enabled it'll do that regardless of Accel being enabled or not, but 3D Clocks might not engage as long as the power setting isnt set to Prefer Max Performance.

[Monk]

Try reseting the card, was having the sane issue with my vega 64 after I sold my 2080ti, figured something had died, when I put my 3090 in I (embarrassingly) noticed the card wasn't seated properly, having messed with pc's as long as I have, I never even considered I'd messed up something so simple.

[Bad 💀 Motha]

Originally posted by ☥ - CJ -:

Originally posted by Bad 💀 Motha:

Should just disable Hardware Acceleration in browsers like Chrome; otherwise just using this browser will force the GPU into 3D Clock Mode.

If Power Performance mode is enabled it'll do that regardless of Accel being enabled or not, but 3D Clocks might not engage as long as the power setting isnt set to Prefer Max Performance.

Hmm no I always have my PC set to High Performance and set NVIDIA CP to Prefer Max Performance and Web browsers never use my gpu since I turned off Hardware Acceleration in them. As its simply not needed to have it use your gpu. This option is better suited for lower end systems so it can use a mix of cpu and gpu rather then just cpu.

[[☥] - CJ -]

icic..
hmmm, spose it depends on the content on the browser as well.

[Illusion of Progress]

Maybe this is related to many of the RTX 3080s that have insufficient capacitors leading to instability? The card might not be getting the voltage it needs, and yes, this is possible to happen at idle/low demand while being more stable under higher demand. I've experienced it.

[dOBER]

Originally posted by Illusion of Progress:

Maybe this is related to many of the RTX 3080s that have insufficient capacitors leading to instability? The card might not be getting the voltage it needs, and yes, this is possible to happen at idle/low demand while being more stable under higher demand. I've experienced it.

This affect only a handfull of cards on release which is 3 months ago. They didnt build these cards anymore and it never affect asus cards because all asus cards are build with right capacitors even on launch.

[Illusion of Progress]

I see. Didn't know all the details of which it affected and which it didn't, but wanted to bring it up to rule it out.