Hmm that is interesting, but I reinstalled nvidia driver since these updates with DDU.This is what Google AI overview says, for whatever that's worth
GPU losing signal
- Thread starter metaton
- Start date
I do wonder if there's still an issue. There's nothing dodgy in the add/remove programs window?
How many of those apps you mentioned can be disabled temporarily (at system start-up)? Most of them have an option somewhere, I'd imagine. Though, that doesn't stop services starting.
Soldato
- Joined
- 26 Aug 2004
- Posts
- 3,312
Since you are using the Hyte Y70, your GPU is vertically mounted using a PCIe riser cable. Riser cables are notorious for causing exactly what you are describing. A cold system might have slight physical contraction that affects the pins, or the motherboard might be failing to negotiate the PCIe 5.0/4.0 link speed on a cold boot.
The Fix: Pull the motherboard out of the case (or carefully re-orient things) and plug the RTX 5090 directly into the motherboard’s top PCIe slot, bypassing the riser cable entirely. If the issue disappears, your Hyte riser cable is faulty or struggling with the 5090's bandwidth.
Temporary Bios Tweak: Go into your MSI BIOS and manually change the PCIe slot configuration from "Auto" or "Gen 5" to "Gen 4". This reduces signaling strain and often fixes riser issues.
The Fix: Pull the motherboard out of the case (or carefully re-orient things) and plug the RTX 5090 directly into the motherboard’s top PCIe slot, bypassing the riser cable entirely. If the issue disappears, your Hyte riser cable is faulty or struggling with the 5090's bandwidth.
Temporary Bios Tweak: Go into your MSI BIOS and manually change the PCIe slot configuration from "Auto" or "Gen 5" to "Gen 4". This reduces signaling strain and often fixes riser issues.
I went back to that day when the errors started I am not sure if it related but I had bunc of errors for Event ID 14 from source nvlddmkm -then asus card got swapped in and these errors went away with the new card. Not the gpu signal lost stuff though.
I was thinking the riser could be the problem. Taking out the mobo would be quite a job and as I can see I cant really fit the gpu in because of the back panel. I will try the bios setting, worse case scenario I can order a new riser from hyte and test it with that. Or do a windows reinstall. Or both.
New test, reseated the gpu and done a cmos reset. Pc worked fine for a day 10 hours of intensive gaming. (007 first light) Next day cold boot same thing is happening again, in couple of minutes in after logging into windows either gpu fans spinning 100%, or just normal fan speeds and system restarts. Happens reliably. If it is a hardware issue how I can game for 10 hours without this happening? On that day I restarted the pc multiple times as well.I never seen anything like this...I removed the latest nvidia driver again and it is doing it with MS driver too. Can this be hardware?
The fact you have used two different cards now and the problem repeats, would suggest that it is not the card at fault.
Could it be other hardware? Yes, though I find the other suggestions in this thread (software/driver or a riser issue) to be more likely than a hardware fault.
You said it does not happen with integrated graphics, which is a good sign that most of your PC is operating normally, but it could still be (once a card is in-place) software/driver related, a PCIE slot, or a PSU problem.
I would go with knocking the PCI-E gen down in the BIOS for the primary slot (if you haven't already), since that's the easiest solution other than removing the riser itself. We're focused on this first, because risers are notorious for all kinds of shenanigans and they're relatively fragile.
Hmm, that's interesting, but I'd still recommend you get as many of those softwares you listed disabled on start-up or uninstalled completely as possible.
I think a fresh install with a bare minimum of apps is the best way to go, to rule software out completely.
It is possible that the card spinning up is just a red herring, e.g. we had black screens awhile back caused by a grounding issue between the monitor and the socket. So, I would rule nothing out.
Thank you for helping me. New development: I was curious to see what would happen if I leave the pc on with bios on the screen. It went dark three times in a row gpu signal gone, however fans did not spin up to 100%. So that would exclude windows and software related problems, right? I done a cmos reset too, it did not change anything. I dont have any spare parts at home, I am trying to pinpoint what this could be. I ordered a new 5.0 riser that was the least painful option. But if that does not help I guess it is either the psu or the motherboard, right? The fact that whenever I reseated the gpu it worked for a while I think it might be the pcie slot and yesterday I was again able to play 10+ hours with demanding games without issues after reseating the gpu I doubt it would be a psu problem, that would not let me play all day if there is a problem with power delivery?
Last edited:
Yeah, it would seem so.
Did you try setting the PCIE gen to 3 or 4 in the BIOS yet? That's a free way of testing the riser (to some extent).
Likely, but I'd suspect the power connector / seating of the connector more than the PSU itself. Problems with these connectors (and their adapters, of all kinds) leading to black screens are very common.
Not necessarily. I would not want to guarantee a new motherboard will fix the problem, because it is just trial and error at this point.
By the way, you mentioned trying a different HDMI cable, did you try a DP cable?
How is the monitor plugged in to the (wall) socket?
What USB devices do you have? Is there a chance you have something that wasn't plugged in for those 10+ hours?
Last edited:
I tried to set it to gen 4 and it did not make a difference. I havent tried dp cable yet, But it was working fine for a long time with hdmi and tried two different hdmi cables with the same result with two different displays (my monitor and my tv) so I thought the monitor issue can be excluded too. Same usb devices.
When you mention power connector you mean the asus adapter or the pcie cables?
Last edited:
The power connector on the graphics card. There are often issues with the 16-pin connector, particularly the mating between the card's power connector and the PSU's connector (potentially on the PSU-end too, if it also uses a 16-pin).
In some cases, using a native cable from the PSU works better than the card's box adapter and sometimes the other way around.
There are also adapters, like 90 degrees, from third parties (e.g. cablemod) and those can be a source of problems too.
The issues started with the msi 5090, it came with an adapter. I sent that back with the card (RMA) at this point I suspect there is nothing wrong with it. The new asus card came with the asus adapter, so I think we can exclude both the gpu and the adapter. If the issue with the cables it must be on the psu side or something entirely else. If bios settings also exluded (cmos reset) the only thing remains the psu, the psu cables, the memory cpu(? Is it less likely?) or the motherboard, right?
Unless the adapter itself the problem, but it was working for more than 6 month before all this started, so according to this my psu should be okay with the adapter?
Last edited:
Memory causing black screens after 2 minutes isn't something I've seen. Bad memory usually just means BSODs. EXPO/XMP can cause problems, though again unlikely in the circumstances and I assume you have these turned off after the CMOS reset anyway (memory probably running at 4800?).
CPU: you never know, but normally with CPUs they either work or they don't work. Boot problems are bent pins 99% of the time, but that wouldn't address your issue because sometimes it works fine. The only other common reason for boot problems is a bad cooler mount (too tight, too loose, etc), which can also be (partially) tested by laying the case flat and seeing if it helps. Not sure if you can safely do this with a riser?
Motherboards can cause problems like this, especially if there's an issue with the PCIE slot, or if a big heavy card sags in the slot and is unsupported.
Well, the 16-pin connectors are known for going bad for no apparent reason, so the 6 months doesn't mean anything in that context. They can also be hard to seat and start to fall out of their own accord, so..
I've seen it a few times where black screens were resolved by swapping either to the adapter, or a native 16-pin PSU cable.
The pc boots into windows and the black screen happens sometimes 5 minutes in. So I think as you said cpu or bad ram should be a boot problem or bsod or complete restart. Gpu adapter: it is a brand new one from asus and still producing the same back screen. Gpu sag wont affect the motherboard, it would affect the riser. So according to this it is either a motherboard/riser or the psu most likely right?
Question is which one would be more likely? (Unfortunately I cannot borrow one so it has to be a new one.)
Question is which one would be more likely? (Unfortunately I cannot borrow one so it has to be a new one.)
Last edited:
So my guess would be the riser card, but I’ve seen similar instabilities and Nvidia errors mostly solved with new 12vHP cables. First I’d pull the graphics card and test the system. Take note of the DRAM temps. If the system is stable refit the card and retest again taking note of the DRAM temps as DDR5 needs to be well cooled. If the errors return swap the riser, or even better get shot of it and a good quality 12vHP cable.
Last edited:
The reason I originally suggested software, is because after 2-5 minutes stuff could be loading in the background, but you said it even happens in the BIOS.
So, my next best guess would be a bad connection, but that could be any one of those.
I'm assuming the PSU does not have a native 16-pin then? Does it support Corsair's cable? That would be cheaper than a new motherboard.
It is a pity you can't pull the riser with the build in the case. I'd suggest building it outside, but a 5090 is very heavy to be unsupported.