Possibly Faulty GPU Causing Motherboard To Permanently Not Start Up Properly

Status
Not open for further replies.
Nagorak

Nagorak

Well-known member
I've recently encountered a very strange problem. My motherboard was working fine, then I pulled my GPU to reroute two of my case fan power cables. After reseating the GPU the motherboard would no longer start up properly when the power button was pressed. It would power up for a split second and then turn off, as if there was some sort of short. However, by hitting the power button immediately a second time, or holding it down the first time, the board would power up properly (it would turn on, turn off, then turn on again and stay on).

Trying to troubleshoot I unplugged everything, and finally removed the motherboard entirely from the case and tried a different CPU and PSU with nothing but power connected. It seems the motherboard has been damaged somehow and will no longer start properly even with nothing connected.

At first I thought maybe something got tweaked when I was reseating the GPU. It seemed weird, because I didn't use any sort of excessive force and didn't notice a problem, but it's sort of a cramped HTPC case where it's difficult to get the GPU in, so it seemed at least plausible.

Based on that, I took a second motherboard of the same type that I have for another build and installed it instead. I plugged things in partially and used a small GPU with no PCI-E connector to test and the board started up fine. Then I went ahead and plugged everything back in, including my EVGA RTX 3080. And tried to power up, and now that second board has the same problem as the first! :mad:

My best guess at this point is something has gone wrong with the EVGA RTX 3080 that is somehow causing a permanent fault with these boards. But, I also can't fully rule out it's somehow due to the PCI-E connector cables, the PSU powering them, or the 90 degree PCI-E power adapters I was using. The problem is, I can't really test it, because if I'm right about it being the GPU I'll just end up trashing another board (the two boards are under warranty so hopefully I can get them repaired).

The really weird thing is outside of this start up fault the system otherwise works fine with both boards.

So, I guess my question is, has anyone ever experienced anything like this? Does it seem plausible that the GPU is somehow faulty and causing some sort of permanent fault with the boards?

It seems like my only real course of action at this point is to RMA the boards, and I guess RMA the GPU?
 
Send it too me and I'll test it.

Pull the CMOS battery on the motherboard and let it sit for an hour. ( Outside of a system with no ram CPU or anything in it, just the board laying by itself. )

Then plug everyhting back in using a different GPU. leave the case fans unplugged for now and try booting. See if it helps.

Make sure you did not accidentaly hit the slow boot button on the boards while doing your swaps.

You can also try a USB Bios flashback to see if that helps.

Last but not least I can't find any info if there is a dual bios. If there is, switch to the second bios.
 
Last edited:
Nagorak I've heard of some evga 3080s pulling too much power from the pcie slot but I doubt it would fry the boards.
 
Can you test the GPU in someone else’s computer? Check any smell coming from the GPU. If GPU is toast then you would either see damage of some sort, get red lights over power cables or have a mild burning smell coming from front or back of the GPU. Unless it is core getting destroyed but then you will get artifacts.

I really can’t think of anything but I had an issue like this and the Power button panel on my Corsair 760T case was shorted and also causing a smell. So I threw the case out.
 
Sounds sort of similar to what happened to me (which led my last hardware upgrade).

I bought some cheap Chinese PCI adapter card. Plugged it in with the computer off but without switching off my PSU. As soon as I connected it, my computer booted even though I didn't press the power button.

I turned the computer off, removed the PCI card and tried to boot again. The fans would spin up for a second but then it automatically reset, and keep repeating that cycle of the fans spinning up and resetting. I think one time I managed to get to the BIOS screen, but never again.

I assumed my motherboard was fried and found a replacement on Ebay for $50. Swapped in the replacement motherboard and had the exact same problem. Fans spin up and then reset. So I figured the CPU was the problem.

So I gave up on trying to replace parts and bought a new Ryzen CPU and motherboard. Things seemed to work alright, but then in some games I would get a hard reset. I had a cheap backup video card to test and it worked fine.

So whatever shorted when I plugged in that PCI adapter card must have fried my CPU, the motherboard and the video card.

I eventually found a cheap CPU on ebay and tried that in the replacement motherboard I bought. Surprisingly even with the replacement CPU and the replacement motherboard, I still had the problem. So my original CPU must have managed to fry the replacement motherboard. I eventually found a second replacement motherboard and that one worked with the replacement CPU (so the replacement motherboard that was fried didn't also damage the replacement CPU).
 
Sounds like gpu might not have had appropriate extra plug in power after the rerouting, and maybe attempting to pull power over pcie instead?

Did you change which plugs you were using at all?
 
As an update, shortly after posting this I realized that I had a third board that was killed, either by this GPU or the power delivery to it. About a month ago my Gigabyte Aorus Z390 Pro WiFi suddenly had its primary PCI-E port stop working, forcing me to RMA it. That would actually be the first board where the problem occurred. (The sucky thing is that this board is bad for memory overclocking in general, but my previous one could still do 3800 MHz memory and this one won't even boot above 3600 MHz. :( It's a secondary machine though, so I guess not the end of the world.)

The reason I didn't immediately put the two problems together was because I thought that was just a fluke thing that the PCI-e slot died. And, initially, the behavior of the damage to the two Asus B550-F boards was so different. I thought something in the case was causing a fault, until I finally traced it to the GPU when the second board developed the same exact problem.

Hapatingjaky said:
Send it too me and I'll test it.

Pull the CMOS battery on the motherboard and let it sit for an hour. ( Outside of a system with no ram CPU or anything in it, just the board laying by itself. )

Then plug everyhting back in using a different GPU. leave the case fans unplugged for now and try booting. See if it helps.

Make sure you did not accidentaly hit the slow boot button on the boards while doing your swaps.

You can also try a USB Bios flashback to see if that helps.

Last but not least I can't find any info if there is a dual bios. If there is, switch to the second bios.
Click to expand...

I left the board with battery pulled and the clear CMOS jumper tripped for like 72 hours, reflashed the BIOS, tried all different components, and it made no difference. It's been permanently damaged by the fault. :(

koralis said:
Sounds like gpu might not have had appropriate extra plug in power after the rerouting, and maybe attempting to pull power over pcie instead?

Did you change which plugs you were using at all?
Click to expand...

I think it is some sort of power delivery problem. It may be a result of my Corsair CS750m only having two PCI-E power cables, both of which are terminated in two 8 pin connectors. The GPU takes three 8 pins and the instructions from EVGA said this was supposed to be OK since the PSU cable ships stock with enough connectors (the PSU technically has enough wattage), but maybe nonetheless not enough power is being delivered and it's pulling too much from the slot.

Maybe the 180 degree adapters I was using added ever so slightly more resistance which added to the problem? Or maybe the PSU is somehow not providing enough power at boot (could be defective). Or maybe the card itself was either defective, or it was damaged in the same event that killed the PCI-e slot on my Gigabyte Z390 (maybe the power connector was loose when that occurred setting off the whole problem--you can speculate endlessly). Some EVGA 3090s apparently have had issues pulling too much power from the slot, so maybe this problem is related to a similar design issue that is occurring under different circumstances (maybe it takes a combination of different component deficiencies to occur).

It's just weird because even after the B550-F boards were damaged, the GPU still ran perfectly fine under load. The only problem was the start-up fault.

I bought another power supply, an EVGA 750P2, that has three separate cables, and also ordered another B550-F motherboard from Amazon with the thought I can just resell one of the other boards once it's (hopefully) back from RMA.

Unfortunately, I don't know that I trust using the GPU again without RMAing it first. I don't really want to keep killing board after board now that there is strong reason to believe the problem is either the GPU, or some combination of the GPU and the PSU feeding it. Sure, it could just be the PSU, but if I guess wrong then another board is dead.

The new B550 is running fine with an old GPU in it, and I think I'll just have to RMA the GPU for safety at this point before putting it back in. I may go ahead and RMA the PSU as well just to be safe (can't risk it damaging another GPU under the current shortage circumstances if it is the source of the problem). It's just too hard to troubleshoot when there is a risk of damaging anything I plug the components into.
 
It's an EVGA GPU. I would've RMA-ed it first thing as EVGA seems to be tight about replacing components in US of A.
 
Yeah EVGA in NA is da bomb. Don't know about the res of the world.

Back on x99 I had a mobo die. Croshipped had it the next day.
 
I have nothing to add except just wanted to say hearing PC problem stories like this is the stuff of nightmares.

Honestly gives me anxiety to even think about this sort of thing any time I need to swap parts in a PC. :eek:
 
I assume you didn't turn the PSU off and brain-drain (push power button couple times with cable disconnected and PSU off) before taking components in and out?

Sounds like more than just the GPU doing all of this, but I guess it's possible. I'd make sure the stand-offs are properly mounted and you aren't grounding out your motherboards over and over.
 
Nunz said:
I assume you didn't turn the PSU off and brain-drain (push power button couple times with cable disconnected and PSU off) before taking components in and out?

Sounds like more than just the GPU doing all of this, but I guess it's possible. I'd make sure the stand-offs are properly mounted and you aren't grounding out your motherboards over and over.
Click to expand...
I only change components while PSU is powered on and computer is running. :bleh:
 
So, as a final follow up to this post. I got a new board and got the system up and running without issue.

In the end I have no clue what caused the problem. I speculated about different components above, but in the end everything checked out individually and together. As far as I can tell everything is working fine.

This is one of the weirdest things that I've ever experienced with computers because multiple boards were damaged across three different rebuilds, so whatever was causing the issue remained between taking the board and all other components out of the case besides the PSU.

It was also a weird failure on the Asus B550-F boards that it caused a start up fault that acted like a simple short, but then it didn't go away even when the boards were removed from the case and fully isolated, and it could also be "powered through" by holding the power button down.

Best guess is somehow the PCI-E cables were loose somehow, and since I was using the 90 degree adapters that wasn't fixed between rebuilds. Although how that would cause this problem when the GPU still functioned fully and could game without issue for hours is unclear? It could have also been that I had the power switch cables routed under the board and they had extensions on them, so maybe somehow the small exposed metal in the end clips somehow made contact with the board and shorted. It would just be bizarre, but I don't know what else could explain it? If it was just a static electric charge or something it seems unlikely that I'd have the same identical fault on two different boards.

Or it could be none of that.

Anyway, in the end just a big mystery all around. The damaged boards were all replaced by RMA, so I guess other than being out the wasted time and inconvenience, RMA shipping cost, and the and loss of value on reselling the extra boards, I didn't come out too far behind.
 
Status
Not open for further replies.
Back
Top