LocutusHUN
460.79 is out...
...and not a single word about if they fixed the idle tdr/black screen issues 

 
Not to be bad news bears but my first 3090 FTW had a weird fan issue as well as the black screen issue so I got a replacement from my local store. Second one had the black screens too and died within hours so, I had to do an RMA. Third one with the black lips have absolutely 0 issues. I don't want to declare this a hardware issue but just wanted to share my experience.

Hello here is my experience:

Only in World of Warcraft, I get this mad fan revving when changing maps and then at some point the system just goes full tilt and black screens with the fans revving mad.

Changing the game power to maximum efficiency instead of optimal seemed to fix the problem + capping the FPS way below my refresh rate of 165.

However I still get one such event every day. I ran fsc /scannow which said it found and fixed corrupt registry entries. Will see how that one goes.

Note: Absolutely no problem with games like Division 2/Cyberpunk/GTAV.

Really driving me crazy.

P.S. No red lights from the card when this happens.

Ive just read a post on the nv forums.
 
They claim, if you disable RGB on your card, the browser TDR's will stop :D
This would be hilarious if it proves true... i have my RGB also set to temperature changing in PX1... :D

jamesleroux
well heck.. been a couple days now and no crashes while browsing. I think this might be the fix (until they fix the root cause anyhow). (the TDR timeout setting, i have not turned off the rgb)

My tought about this: IF the problem is, that sudden hardware acceleration in browsers (or other programs too) causes the card to momentarily spike the boost frequency above the allowed, and that is causing the crash, isnt there a risk, that extending TDR timeout will kill the card?

LocutusHUN

jamesleroux
well heck.. been a couple days now and no crashes while browsing. I think this might be the fix (until they fix the root cause anyhow). (the TDR timeout setting, i have not turned off the rgb)

My tought about this: IF the problem is, that sudden hardware acceleration in browsers (or other programs too) causes the card to momentarily spike the boost frequency above the allowed, and that is causing the crash, isnt there a risk, that extending TDR timeout will kill the card?

Argh, spoke too soon, had a browser crash today. Whole system reset.

Maybe its time to put this in the 1st post here.
 
Currenty there are only 2 confirmed workarounds:
 
1. Disable Hardware Acceleration in your browser. This causes performance loss with demanding sites, and videos, streams, so it comes at a price. TDR's still might happen in other applications that are graphics demanding, but not games (graphics programs, wallpaper engine etc)
2. Set the nvidia driver to Prefer Maximum Performance (instead of Optimal). This wont let your card downclock in idle, so it will use more power, and cause more heat than usual. But this entirely prevents TDR's.
 
+ You can try a hybrid solution, and set the Prefer Maximum Performance on a per application base in the nvidia driver. For example for your browser's exe. Has to be yet confirmed, if this is a reliable solution.

jamesleroux
I just tried it doesnt seem to be effective, plus you cant set it on firefox and that also TDRs a lot..

Tried what?

nice

this is happening recently to me as well after updating to precision x1 1.1.3 and updating the card's firmware (3080). screen would go black for a few seconds and apps like chrome would crash (i've noticed it's mainly just chrome?) and then afterburner would say that connection to GPU1 is not found. event viewer says its either a dwm.exe issue or paging file too small error. super irritating. chrome's hardware acceleration option is disabled and my power setting is set to optimal power

Asrock 570 phantom gaming, Evga 850 platinum power supply with all 8 pins from VGA 1,2&3 so no daisy chain cable 32Gb ram Corsair AMD 3600 and 3090 ftw3 ultra, Well after about 15hours the card is dead black screen disconnected it removed and reseated (6/7 attempt) got it working tried to load cyber punk and it crashed again with fans ramping 100% then after that never got the Windows screen again, I know the computer has booted into windows as the hue plus lighting is doing what I set it to in window, normally at power loss the hue plus results to constant lighting until it boots into windows. Not a happy chappy 

got a link?

It seems that setting prefer max performance setting fixes even the gsync flickering for me.
 
Optimal: 4K/120/Gsync: Flicker in windows
Performance: 4K/120/Gsync: No problems at all
 
Maybe its time for nvidia and/or evga to look into their cards idle power profiles, because it seems that it isnt enough for running high bandwidth HDMI.

pretty sure chrome is involved somehow. when i'm using version 87 after a while chrome will crash and the GPU will disconnect, black screen flicker etc for a bit then I'd need to restart the PC. when i tried running version 86 i didn't have any issues for hours. the stupid thing is no matter what i disable, chrome will auto update itself and even crash itself to force you to update to 87. 

As i got my 3080, i have started with 457.30. But it showed these black screen problems immediately. Dont know about prior ones, but after that ever driver does the same.

so chrome isn't the issue. just got the same black screen, GPU connection lost issue while using the latest firefox. event viewer says a bunch of error dwm.exe error happened. super annoying. i think it might be the latest BIOS that came with precision version 1.1.4. is there a way to rollback on this?