vulcZ

LocutusHUN

vulcZ
 
...blah...
NVCP -> Power Management Mode "Optimal Power" -- setting to High Performance causes GPU clock speed bug that locks GPU clock at 1800mhz even while idle at desktop.
...blah...
 

 
And this is exactly the workaround, that prevents these black screens, until nvidia fixes their drivers.

You might right. I got an NVLDDMKM error last night. So I'm going to try that next. I don't like trying a lot of things at once because then it's harder for me to determine which one is responsible for improving things.

Baasha
I started having this issue since last night.
 
RTX 3090 FTW3 Ultra - black screened during gameplay. Hard rebooted several times and still no video signal. Power cycled three or four times and then randomly got signal back.
 
Then today, black screened during web browsing (no load on GPU) and had to hard reboot. Then black screened again during live stream - had to reboot since no video signal.
 
This is really getting on my nerves - an $1800 GPU that has these issues? I have 4x 3090 FE cards and none of them have any issue - using the same cables/monitors with those and they all work fine.
 
What is the solution?

Baasha

Nereus
Try plugging your DP cable in another port from your 3090 card. Also you might need to buy a certified DP1.4 cable, seems a few people have found a sub-standard DP cable to be the issue (or HDMI if you're using that).
 

 
Tried that and it didn't work. Now, the GPU has a red LED and no RGB - it seems completely dead. This is really infuriating. I swapped my RTX 3090 FE into the system with the same cables/monitors etc. and it works flawlessly. Just live streamed for over 1.5 hours with no issues at all - and I had the GPU OC'd as well. A side note, the FE card was ~ 10C cooler than the FTW3 Ultra with the same setup - that is quite surprising.
 
Anyway, I'm not sure if I have to RMA the card now or if there's a way to recover it - there's just a red LED when the PC starts and no video signal in any of the ports - tried 3 different ports (DP 1.4) and a HDMI connection - on 3 different monitors - zero video signal from the FTW3 Ultra.
 
My DP 1.4 cables are certified for 4K 120Hz.
 

Baasha
I started having this issue since last night.
 
RTX 3090 FTW3 Ultra - black screened during gameplay. Hard rebooted several times and still no video signal. Power cycled three or four times and then randomly got signal back.
 
Then today, black screened during web browsing (no load on GPU) and had to hard reboot. Then black screened again during live stream - had to reboot since no video signal.
 
This is really getting on my nerves - an $1800 GPU that has these issues? I have 4x 3090 FE cards and none of them have any issue - using the same cables/monitors with those and they all work fine.
 
What is the solution?

LocutusHUN
New driver is out.
457.51
They claim to have fixed this.
 
What i found is, that even tough i set the new driver back to optimal performance, the 3080 still does not want to downclock in idle. As if their fix is exactly this... Can anyone else try this?
It seems that there is no difference now between optimal and max performance mode in the drivers energy option.

ronatar2
I dont believe i thought i am the only one who having blackout.
I have 3080 xc ulta and i tried everything and nothing work

soundwave404

LocutusHUN
For me the same. NVLDDMKM error. The display driver stopped responding, and has recovered. Except that recovering did not gave the display signal back, so i always had to restart the PC. Even tough sound seemed to play in the backround, so not the entiere pc was frozen.

This is exactly what I had been experiencing and it seems to have started for me since installing v457.51. It's happened twice so far, both times when the PC has been idle.
When I have some time I will try and downgrade to the previous version and see if it stops.
(3090 FTW3 Ultra)
 
EDIT to mention: When this happens, the GPU fans all go to what sounds like 100% until I reboot.
 

Manuel@NVIDIA19m
 Close
 


Our telemetry shows that the outstanding bug was pretty much fixed in the field. You are likely seeing a different bug which we are not seeing. Previously users mentioned that the TDR did not create a dmp file so you will need to trigger a bugcheck as described below:

1. Enable full dmp files by running the registry file fulldump.reg from the URL below: https://nvidia.s3.us-east-1.amazonaws.com/fulldump.reg
2. Download and then run the registry file TdrLevel1.reg from the URL below to change the level of recovery: https://nvidia.s3.us-east-1.amazonaws.com/TDRLevel1.reg
3. Reproduce the browser monitor reset issue again.
4. On your keyboard, press the Windows key and the letter "R" at the same time to bring up the Run command window. Type the following into the Open field: %LOCALAPPDATA%\CrashDumps Then press the OK button to proceed.
5. This will open the folder location where the dmp file should be located (if you see multiple files, it will be the file with the most recent date.
6. Right-click over the file -> select "Send to" from the context menu -> choose "Compressed (zipped) folder" to zip the file to a smaller file size.
7. Upload the zipped file to a file sharing service such as Google Drive or Microsoft OneDrive and then email me a link to download the file to driverfeedback@nvidia.com.
8. Last but not least, in your email, please also email me your MSINFO32 file as described in the FAQ below: http://nvidia.custhelp.com/app/answers/detail/a_id/2507/