• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

4090 randomly loses signal to display

DG-87

DG-87

DG-87

Associate
Joined
4 Jun 2016
Posts
48
Been having this issue for a week or two now.

PC randomly loses video signal and screen goes dark. PC is still running but there's no way to get the display to work and there's no user input anymore, unless I reboot. Can happen weather playing a game or watching YT or just idling at desktop.

MB spits out codes: 38, 40, 41, 42 iirc but this happens after screen goes dark and there a different sound from the pumps ect.

I haven't really changed anything recently, the only thing I do is update the display driver.

System is all stock.

I did have XMP enabled (without issue all this time) but turned it off but makes no difference.
Updated display driver and tired another display and it doesn't fix it.

One thing that HAS been an issue with this 4090 for a year now is an occasional green screen flicker that I could never figure out, again totally random and not all the time, a reboot would fix it temporarily.

I know the C2 doesn't always play well with sleep mode (display won't wake, forcing a restart) but this feels like it's a 4090 issue...
 
PS looking at the event viewer, just before an "unexpected shutdown", some things I see:

WHEA-Logger:
A corrected hardware error has occurred.
Component: PCI Express Root Port
Error Source: Advanced Error Reporting (PCI Express)

Primary Bus:Device:Function: 0x0:0x1:0x0
Secondary Bus:Device:Function: 0x0:0x0:0x0
Primary Device Name:PCI\VEN_8086&DEV_A70D&SUBSYS_88821043&REV_01
Secondary Device Name:

Display:
Display driver nvlddmkm stopped responding and has successfully recovered.

Dwminit:
The Desktop Window Manager process has exited. (Process exit code: 0xe0464645, Restart count: 1, Primary display device ID: NVIDIA GeForce RTX 4090)

This just keeps "counting" until reboot.
 
Hello mate, just to clear up some things. What do you mean by "no user input anymore"?

When the pump makes a different noise etc, do you recognise the noise, for instance does it sound like the system is restarting or is the cpu spiking/getting hotter?

Can you explain about the codes, is there a general theme?

On the face of it does sound like dodgy 4090 or unstable driver.
 
check your 4090 power connections and try a different pci slot if available?
 
How much sag does your 4090 have, i recently posted check your gpu for sag, i damaged a x670e extreme due to a 4090/7900xtx, as time went on i noticed my gpu would sit lower to one side, had gpu removed errors in windows on the 4090 and a 7900xtx i moved to, deduced my x16 slot was damaged.

I changed my motherboard to a hero x670 and bought a support bracket and have had no further problems.
 
Had a very similar "black screen" issue and it turned out that the Graphics card had somehow damaged the top PCIe slot. I have a thread on the "General Hardware" forum titled "black screen".
There was no noticeable sag on the graphics card. But definitely the PCIe socket, as running the graphics card in the lower PCIe slot and it ran just fine. Put it back in the top one and FUBAR.
Sadly I had thought it was the Graphics card, so raised an RMA (not OCuk) and they came back and said "no fault found". Originally did not try the card in the lower slot, as I was convinced the graphics card was FUBAR.

Looks like your mobo has a secondary PCIe slot. I would recommend trying it in this.

In the end I replaced my Z690 mobo with a new Z790 and all now appears well.
And "yes" I've bought a graphics card anti-sag bracket to provided some more support. Even though there is no obvious sag.

Good luck.
 
Last edited:
I was getting that exact same issue before my ZOTAC 4090 bit the bullet, got worse and worse over a period of 2 months to the stage where i couldnt start anything 3d related without an instant black screen,
RMA'd and replaced.
 
DG-87 said:
Been having this issue for a week or two now.

PC randomly loses video signal and screen goes dark. PC is still running but there's no way to get the display to work and there's no user input anymore, unless I reboot. Can happen weather playing a game or watching YT or just idling at desktop.
Click to expand...

I've been gettting this for the last week on my 3080ti. I thought it was an nvidia driver, but i rolled back today and just did it again.
Loses connection to monitor and also for some reason, turns off the lights on my keyboard... and only way to use pc again is a hard reset :/
 
sWiZzLe said:
I've been gettting this for the last week on my 3080ti. I thought it was an nvidia driver, but i rolled back today and just did it again.
Loses connection to monitor and also for some reason, turns off the lights on my keyboard... and only way to use pc again is a hard reset :/
Click to expand...
If you have not already... try the graphics card in a lower slot. You basically have nothing to loose trying this.
Good luck.
 
wookiee87 said:
How much sag does your 4090 have, i recently posted check your gpu for sag, i damaged a x670e extreme due to a 4090/7900xtx, as time went on i noticed my gpu would sit lower to one side, had gpu removed errors in windows on the 4090 and a 7900xtx i moved to, deduced my x16 slot was damaged.

I changed my motherboard to a hero x670 and bought a support bracket and have had no further problems.
Click to expand...
I always use a graphics card anti sag device of some sort these days due to their weight.

The asus one anti sag if it fits is amazing
 
I see there's been many replies and good suggestions so I will provide more info on my set up as it was:

I am using one of the first CableMod 12VHPWR to 3x8Pin cables, PSU is a 1000W. I know the PSU is more than enough as I confirmed it with a power meter and I don't really OC my PC these days. The CableMod cable has been on and off many times and it's never been a burden like with some and it's never tried to melt on me.

The 4090 Strix is indeed a dual bios and has always been set to silent.

My setup is an unusual one, Upright w/Phanteks G40 WB in a Lian Li EVO and so it requires a riser cable, which is a Lian Li PCIE Gen 4, plugged in to the very top slot for best performance...I cannot really reach the lower PCIE slot with this riser cable, nor can I simply flip in the GPU in the horizontal and there's a 360mm rad in the way. Again it's been on and off many times and the 4090 performs really well with it. I'd rather not use one at all as it's just adding a point of failure but this is the orientation I chose. When the 5090 releases I might play around with the orientation.

My PCIE lanes are set to Auto in UEFI. I did have an issues with an older system once, where there was no signal on boot up and setting it to Gen 3 specifically fixed it. It was a 3090 or a 4090 icr now.

Running Win 10 21H1 AME, it's stripped down and a personal choice and it's never caused me any issues when it comes to system stability. The only downside is the NVCP can't be installed but I can manually install the NV drivers. I'd DDU the drivers but I can't really update Windows itself, even if I wanted to but I will soon be moving on to Win 11 23H2.

I don't usually update firmware/bios unless I absolutely have to or there's a new feature I need but again these never fixed anything in the past for me so it's really a last resort. But maybe when I do my Win 11 update I will check out the firmware/bios too.

I always monitor my clocks/sensors with MSI AB and there's nothing weird there. The 13900K and 4090 both have their own EK FLT120/240 pumps/res and 360 rads and use KryoSheets. My 4090 VRAM temps are hotter than I'd like at 80-100C, depending on games but it's still in spec. I tried three times to reseat the pads and try different sizes but that's the best I can get.

And my user input I mean M&K.

TLDR/Update:

I tried setting PCIE native power management to off, turning XMP off, setting PCIE to Gen 3 and even updated my Win 10 AME to 22H2 and NVCP to latest, "sfc /scannow" kinda forced this I believe...but these crashes still happen, they have become less frequent but nvlddmkm seems to be the culprit as it's always in event viewer before a crash...it seems to be something affecting many 40 and even 30 series.

I'm quite busy to be trouble shooting this 24/7 but I'll keep trying things that I think might help but for such an expensive product we shouldn't have to do this, it's been an issue for years now. I swear I never had a high end PC be this much of a pita in the past and would cost a fraction of the price too.
 
DG-87 said:
I tried setting PCIE native power management to off, turning XMP off, setting PCIE to Gen 3 and even updated my Win 10 AME to 22H2 and NVCP to latest, "sfc /scannow" kinda forced this I believe...but these crashes still happen, they have become less frequent but nvlddmkm seems to be the culprit as it's always in event viewer before a crash...it seems to be something affecting many 40 and even 30 series.
Click to expand...

It used to happen to me, but only when a second monitor (in my case a TV) was plugged in. Even if the TV was off, just the presence of a cable connected from my 4090 to the TV could cause my main display to randomly lose signal in games.

The problem for me fixed it's self when I upgraded my main monitor. Now both my monitor and TV are 4K with GSYNC support.
Before hand they were different resolutions and refresh rates etc so I guess it was just a conflict in the Nvidia drivers
 
Last edited:
Slight update.

The black screen got worse again and was outta control, PC was unusable...before messing around with more settings (that I doubt would do much anyway) I dug out my GTX1650 too see if it would lose signal.

After using the PC as I normally would for a couple of days I noticed no black screens at all. Plugging the 4090 back in caused a black screen not long after boot.

So it's the 4090 or things attached to it, it seems. I re seated all the cables and noticed my CableMod 12VHPWR cable now HAS what looks like very minor heat damage on the plug, around the sense wires...it looks like when you put super glue on plastic and get that chemical reaction.

Looking up posts from a year ago regarding "4090+black screen" and the CableMod 12VHPWR cable, I saw a lot of similarities to my 4090's behaviour.

So maybe it was the cable all along.
 
You must log in or register to reply here.
Back
Top Bottom